WIRELESS ENDOSCOPY INTRODUCTION

Transcription

1 WIRELESS ENDOSCOPY INTRODUCTION The gastrointestinal diseases especially in the small bowel are hard to detect because of its location. Traditionally, long video endoscopic probes are used for the inspection of the gastrointestinal (GI) tract. Although this technique works well to obtain a detailed diagnosis, it is invasive and may cause significant pain for patients. Therefore, wireless capsule endoscopy has been developed to make standard endoscopy less invasive and, in addition, to reach less accessible areas of the GI tract (1). Figure 1 shows a wireless capsule endoscopy system for medical diagnostic applications. A miniature-sized wireless endoscopy device reaches areas such as the small intestine and delivers images and measurements of physiological signals wirelessly to an external receiver worn by the patient using one of the appropriate bands such as industrial, scientific and medical (ISM), medical implant communication service (MICS), and ultra wideband (UWB). The collected physiological signals and image data are transferred to a computer for diagnosis, review, and display of images. A high-resolution videobased capsule endoscope produces a large amount of data requiring a high-capacity wireless link. A wireless capsule endoscopy system is designed such that the patient will wear the external receiver (i.e., gateway) on the body closetothecapsuledevice(2).thegatewaycanbean array of receivers placed around the human abdomen to receive better signals from the capsule in the body. In current commercial capsule systems, the images are stored in the external receiver, and the patient returns the external receiver to a physician who downloads video images to a computer and uses special software to inspect and detect abnormalities. A remote receiver can be used to receive data from the external receiver close to the patient s body wirelessly. This additional wireless link will provide the patient freedom of movement within a hospital room. With this scenario, it is also possible for a health practitioner to view the data online using the Internet independent of the patient location (3). Commercially available wireless capsules used for diagnosis of diseases within the GI tract are passive (4). Actively controlled capsules are needed that can complete the journey in a shorter time and be stopped or navigated around a specific location to obtain a thorough diagnosis and treatment. Some studies have proposed concepts for magnetically propelled swimming robots (5). There are a number of similar projects around the world developing systems for control of capsule movements in the small and large intestines. For example, magnetic forces are used in Reference 6 to steer capsules in the GI tract to control and navigate capsules. These approaches are expensive and require many orientations of permanent magnets in the external device as well as inside the capsule. Location detection and position and movement control of capsules are existing issues that have awaited solutions for more than 50 years; in fact, these issues are described as dream features in Reference 7. Several commercial wireless capsules are already used in clinical environments for visualization of abnormalities in the GI tract. They offer almost similar design specifications. In addition to the camera-based wireless capsules, these companies also manufacture capsule devices for ph monitoring, temperature monitoring, and pressure monitoring. The first camera-based capsule product was the M2A capsule (8) manufactured by Given Imaging (Yoqneam, Israel) at the time when significant advances in integrated device technology enabled the design of miniature cameras. The M2A capsule received approval from the U.S. Food and Drug Administration (FDA) in Previously, wireless capsules measuring physiological parameters such as ph, temperature, and pressure existed since 1950 (2). The latest camera capsule PillCam SB by Given Imaging (9) is now widely used around the world in clinical settings for visualizing the small bowel to detect and monitor many abnormalities, including bleeding. With the help of camera-based capsules, the physicians can detect, monitor, and diagnose abnormalities in the entire small bowel without subjecting their patients to an uncomfortable procedure or sedation. The existing wireless capsules use narrowband wireless links. To achieve a higher frame rate transmission of images, they use some image compressing techniques. A high-data-rate transmission will increase the resolution and image quality of wireless capsules. Thus, a wideband communication system such as UWB communication will be an ideal wireless technology for future wireless capsule endoscopy by achieving a data rate equal to or higher than 100 Mbps. For such a high data-rate capacity, a wireless capsule can transmit raw video data without any compression, resulting in a low-power transmitter, less delay in real time, and increased picture resolution. With a highdefinition camera, such as 2 megapixels, UWB communication can send more than 30 frames per second (fps) (10, 11). At high frequencies, tissue absorbs significant signal energy, thus affecting overall performance of a telemetry link in a wireless endoscopy system. For high-frequency designs, signals through different layers of tissue should be optimized together with the wireless communication device as shown in Figure 1b. The existing wireless capsule devices use small batteries to power the electronics in a capsule. New developments consider a wireless power mechanism to provide the power supply or to recharge the battery of a wireless endoscopy device (12). In addition, future capsules are designed to be fully robotic with locomotion and motioncontrol features (13). Robotic capsules are targeted for drug-delivery applications in the human GI tract. The navigable wireless capsule pill will carry the right amount of drug and will go to the right location inside the body to deliver the exact amount of the drug required (14). Realtime wireless energy transfer via magnetic coupling is necessary for these types of capsule endoscopes to provide mechanical function, as they require a large amount of power for continuous movement (15). This article presents a detailed discussion of recent studies on motion control and localization that are targeted J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright#2014 John Wiley & Sons, Inc.

2 2 Wireless Endoscopy Figure 1. (a) A wireless endoscopy system for medical monitoring. (b) Signal transmission through tissue layers inside the human body. features for greater medical diagnosis, and wireless energy transfer to increase the operating life of current wireless capsules. These additional features will improve therapeutic capabilities and enable a fully robotic wireless capsule endoscopy system. WIRELESS ENDOSCOPY DEVICES A wireless capsule endoscope system consists of a tiny camera, light-emitting diodes (LEDs), a wireless transmitter, a battery, and an antenna. Recent research activities in the development of wireless capsule devices include a power management unit for receiving wireless energy and regulation, and a motion control unit. Figure 2 shows the components required for a wireless capsule device. It is important that integration of all these individual components will be low in complexity, small in form, and light in weight in order to reach a very small section of the GI tract. History of Wireless Endoscopy Devices The design of wireless capsules began in the 1950s. The complete design of swallowable radio transmitters for use in diagnosis of the organs within the digestive system first appeared in the literature in 1957 by two different groups, LEDS CAMERA/ SENSORS Motion Control Unit Battery/Power Management Microcontroller Transmitter Antenna Figure 2. A pill-sized wireless capsule and components. almost at the same time. These devices successfully operated in the GI tract and measurements such as ph, temperature, and pressure were recorded wirelessly. Since then, these devices have been called endoradiosondes, capsules, smart pills, electronic pills, radio pills, wireless capsules, wireless endoscopy, and so forth (16 18). Because the integrated circuit (IC) technology was not advanced, the early designs used simple electronic systems with fewer components. The design by Mackay and Jacobson (16) used a circuit given in Figure 3a to transmit pressure signals wirelessly using a core attached to a diaphragm moving inside the inductor. Temperature is measured through the transistor itself. The operation frequency selected for this capsule was 500 KHz considering the better signal penetration in the body and the effect of the skin depth at low frequencies. This wireless capsule system used a several array of coils around the body to compensate the changes resulted from the different orientations of the passive capsule device. Mackay and Jacobson (16) also described a wireless energy technique in their article by using a large coil around the patient to induce energy into a secondary winding at a frequency different than that of the outgoing signal. However, a battery was used in actual measurements because a better performance was obtained. The device was shaped like a pill and measured 9 mm in diameter and 28 mm in length (Figure 4a). The capsule designed by Farrar and Zworykin (17, 18) also contains an oscillator-based transmitter powered by a battery having lifetime of 15 h. It has a cylindrical pill shape and measured 10 mm in diameter and 30 mm in length. It uses a transistor oscillator circuit similar to the one in Figure 3a. The components were packaged as given in Figure 4b. The capsule had a flexible rubber at one end that covers a pressure transducer. Similar to the previous design, the pressure applied to the transducers modulates the oscillation frequency of the transmitter [known as frequency modulation (FM)], which is received and demodulated by a frequency modulation receiver. The

3 Wireless Endoscopy 3 Figure 3. Circuits used in earlier capsules: (a) a pressure- and temperature-measuring capsule (16); (b) a ph, temperature- and pressuremeasuring capsule (7); and (c) a pressure-measuring radio pill (19). pressure signal was recorded using an oscilloscope. The precise measurement of pressure activity in the intestine was complicated for this design as the capsule would move downstream. This indicates that position control was also an important problem that time. Figure 3b is a echo capsule that was designed in 1962 by a Japanese group (7) and was energized wirelessly. A wireless energy link was investigated because miniature batteries were not available in Japan at that time. The circuit is a Hartley oscillator in which a capacitor C c is used as a storage element instead of a battery. The resonant part of this circuit (L-C) receives a power signal from an external coil, and the signal is rectified through the p-n-p transistor. C b is a blocking capacitor, and the electric charge on this capacitor is discharged from base to collector. This circuit acts as a receiver to receive power and as a transmitter to transmit information on various quantities such as temperature, pressure, and ph. Temperature is measured through the transistor, pressure is measured through the movement of core inside the coil, and ph is measured using the AgCl electrode. RD is a Zener diode used for voltage regulation. The cylindrical pill-shaped Figure 4. Structures and package styles of earlier capsules in 1950s and 1960s: (a) pressure-measuring capsule (16); (b) pressure-sensitive radio transmitter (17); (c) a ph-, temperature- and pressure-measuring capsule (7); and (d) mechanical layout of pressure measuring radio in Reference 19.

4 4 Wireless Endoscopy Figure 5. Sample design of recent wireless capsules. capsule had 8 mm in diameter and 25 mm in length (Figure 4c). Another FM oscillator-based design shown in Reference 19 measured pressure using the oscillator circuit shown in Figure 3c. The design of inductor (L1) was done by winding around a small ferrite core and placed parallel to a ferrite disk. The inductance changes as the air gap between this disk and the core changes, which in turn leads to a change in the oscillation frequency. The carrier frequency is selected between 300 KHz and 500 KHz. The mechanical layout is shown in Figure 4d. The pill had a battery lifetime of 80 h for recording. The volume of the pill is approximately 2 cm 3 (less than 1 cm 3 cm). Current wireless endoscopy devices integrate more complex systems on the same platform when compared with the systems in 1950s and 1960s (Figure 5). Recent significant technology improvements have enabled the design of small cameras and batteries. In the last 10 years, research projects examining developing wireless capsules have concentrated mostly on the visual sensor system. In 2000, First clinical trial of a camera-based wireless endoscopy (8) has successfully been achieved. An example of recent wireless capsule design is depicted in Figure 5. In addition to image data, detection and subsequent measurement of physiological signals such as temperature, ph, and pressure, are usually necessary to improve patient diagnosis, as listed in Table 1. The image data are obtained from a complementary metal oxide semiconductor (CMOS) camera and is in a digital format. Physiological signals obtained from inside the human body are initially analog and go through an amplification/filtering (A/F) process to increase the signal strength and to remove the unwanted signals and noise. A multiplexer is used to switch between each data. An analog-to-digital conversion (ADC) stage in the microcontroller is used to convert the analog physiological signals Table 1. List of Sensors Used Inside Wireless Endoscopy Devices Devices Signal Types Camera Camera images (data rates: 10 Mbps) Temperature Physiological signal (30 C to 40 C) Oxygen Physiological signal Pressure Physiological signal ph value Physiological signal (1 13) into a digital format for digital processing. The microcontroller will code the data before the data are sent to the wireless transceiver. The multiaccess protocol to enable a multiuser function is implemented in the microcontroller. A multiuser function would be an important feature if more than one patient is in the same care room in a hospital. It contains a battery and power-management circuitry. The power-management circuit is usually a voltage regulator chip used to distribute the power source to the individual blocks. Commercial Endoscopy Devices Table 2 presents the details of commercially available wireless capsule endoscopy technologies that are being used in clinical environments. These devices are used to provide diagnostic images for the esophagus, small bowel, and the colon, and they can identify the source of GI bleeding. For example, PillCam SB (Given Imaging, Yoqneam, Israel), EndoCapsule (Olympus Corporation, 3500 Corporate Parkway, P.O. Box 610, Center Valley, PA , U.S.A.), and MiroCam (IntroMedic, Seoul, Korea) capsules are for small bowel. The current wireless endoscope devices are efficiently being used to diagnose disorders such as Crohn s disease, Celiac disease, cancerous tumors, iron-deficiency anemia (IDA), obscure GI bleeding, and ulcerative colitis. One state-of-the-art technology for wireless endoscope devices is PillCam SB 3. PillCam SB 3 recently received the FDA approval; it uses adaptive frame rate technology to deliver more detailed images and coverage in the small bowel (9). It has a physical dimensions of 11 mm 26 mm and weighs less than 4 g, transmitting images at a rate between two and six images per second. Commonly used wireless transmission frequencies have been ultrahigh frequency (UHF) around 400 MHz. Given Imaging devices use the Zarlink s radio frequency (RF) chip (20) for wireless transmission in the MICS band ( MHz). The allowable channel bandwidth (BW) for this band is 300 khz. Because of the limited transmission bandwidth used for the commercial capsule devices, the image transfer rate has been limited to 2 35 frames per second. As high-definition cameras continue to be developed, they will become more attractive for use in wireless capsule endoscopy devices. A higher pixel camera will require higher image transfer rate. Thus, future wireless capsule devices will target higher-bandwidth data transmission that could facilitate a better diagnosis.

5 Wireless Endoscopy 5 Table 2. Comparison Of Commercial Wireless Capsule Devices a Model Company Camera (Sensor) Frequency (MHz) Data Rate Power Source Physical Dimension Image Rate and Resolution Operation Time PillCam (SB 1) (SB 2) (SB 3) PillCam ESO 3 PillCam Colon Given Imaging Micron, CMOS & 433 ISM (Zarlink) Given Imaging Given Imaging pixels Two cameras, CMOS, pixels Two cameras, CMOS, EndoCapsule Olympus Optical CCD camera, Sayaka RF System Lab CCD Image sensor MiroCam IntroMedic pixels OMOM capsule ChongQing JinShan Science & Technology SmartPill Smartpill Corp. Acidity (ph), pressure, and temperature 800 kbps (FSK) Battery Battery Battery 11 mm 26 mm, <4g 11 mm 26 mm, 3.7 g mm, 2.9 g Battery 11 mm 26 mm, 3.8 g Wireless power 12 MHz body as trans. Channel (HBC) Battery 11 mm 24 mm 3.3 gr Rechargeable battery 2 6 images/s 8 h 50,000 images 35 images/s, 2,600 images 30 min 4 35 images/s 10 h 144,000 images 2 images/s 8 h 9mm22 mm 30 images/s 8 h (870,000 images) 13 mm 27.9 mm, <6g 3 images/s 11þ h (118,800 images) 2 15 frame/s 8 h Battery 13 mm 26 mm Only sensor discrete data a All data within this table are obtained from the follownig websties: and Currently, all video-based commercial systems use LED illumination. Sayaka, by RF System Lab (Traverse City, MI), has both wireless power transfer and localization capabilities. This battery-free capsule contains three rotor coils for posture control and four LEDs for focus adjustment. This capsule with posture and orientation control has the ability to stay in a specific area of the intestine to obtain higher quality images. Another endoscope, EndoCapsule, developed by Olympus Corporation (Tokyo, Japan), was mainly used in Europe but in 2007 received marketing clearance from the FDA. The device contains six LEDs with adjustable illumination to maintain optimal imaging. The capsule by SmartPill is designed to measure pressure, ph, and temperature as it passes through the GI tract. A receiving device worn by the patient collects data that are later examined by a physician. Another commercially available capsule for endoscopy is MiroCam. This system has a different wireless transmission technique than other capsule technologies. Instead of using RF signals to transmit images, MiroCam uses natural electrical impulses in the human body as the transport medium (21). Figure 6 presents the physical shapes of commercial capsules. Current video-based wireless capsule developments will continuously be improved to have better features. For example, at the moment they have an average battery life of 8 h and provide 2 35 frames/s transmission. A high-capacity radio system is currently necessary for this technology to examine visually the digestive tract wirelessly with a better resolution. In most of capsules, small batteries are used to supply the energy to the electronic boards. A wireless power mechanism either charging a battery externally or directly powering from an external wireless source will be a significant design improvement for wireless capsule endoscopy technology. Wireless Power Link A critical challenge for wireless capsule devices is the limited energy source. Currently, the battery in wireless capsules provides the energy to the active electronic components in the device. Although miniature rechargeable battery technologies are available, the lifetime they provide may not satisfy the desired operation time for detecting and transmitting useful data. One way to enhance this operation lifetime is to charge the battery wirelessly. Inductive links have been used to power medical implants to eliminate the use of batteries or to charge batteries to extend the lifetime of the implants (22). Unlike

6 6 Wireless Endoscopy Figure 6. Wireless video capsule endoscopes: (a) PillCam SB2 (Given Imaging), (b) MiroCam (IntroMedic, Seoul, Korea), (c) Endo Capsule (Olympus Corporation), and (d) OMOM (Jinshan Science and Technology, Chongqing, China). the short-range inductive implant systems where the wireless link has a distance of 1 2 cm (centimeters), wireless capsule devices are placed deep inside the body with a distance of up to 10 cm or 20 cm for some patients and thus require a long range wireless transfer (2). The length and width of commercially available wireless capsule devices is usually 26 mm 11 mm, which is barely smaller than the average diameter of the adult upper esophagus. Wireless power can substantially reduce the overall size and weight because the need for batteries is eliminated. As described previously, the study of wirelessly energizing wireless capsule devices started with the development of the early capsule devices. The designs discussed in References 7, 23, and 24 used an inductive link for wireless power transfer. One or two large circle-shaped coils as shown in Figure 7a are placed around the human body Figure 7. Wireless power link using a class E driver for energy transfer in wireless capsule applications: (a) using a Helmotz coil configuration for wireless energy transmission; or (b) using an array of power transmitting coil for wireless energy. (c) Class E power transmitter and receiver circuits.

7 Wireless Endoscopy 7 to energize the capsule placed inside the body by an external source. These earlier battery-less devices operate based on passive telemetry. They use a resonant circuit whose characteristic frequency is sensed from outside. It operates exactly in the similar fashion with the radiofrequency identification (RFID) technology. The capsules with wireless power source are generally smaller in size than a battery-powered capsule in addition to the unlimited lifetime they provide. Figure 7b shows another efficient and less complex structure to energize wireless capsules inside the human body. With the use of multiple arrays of coils, in which each array transmits a power signal, the device inside the human body will receive energy even if it moves in the digestive system (25). The autonomous robotics capsule device in (13) uses wireless energy for power supply. It successfully provides a wireless power source more than 400 mw. The device s driving source is based on a direct current (dc) motor used for the linear actuators of the microrobot. A very low (10 KHz) transmitting frequency is used for wireless power to reduce human body absorption. Another study implements a multicoil technique for inductive powering of an endoscopic capsule (12, 15). The system is shown in Figure 8, which consists of external two Helmholtz coils transmitting energy to a 9-mm three-dimensional (3D) coil power receiver (similar to Figure 7a). Three orthogonal coils have been designed for receiving energy to supply the power to the device. The wireless power link is able to transmit a power around 300 mw at 1 MHz transmission frequency, which is sufficient enough for use in locomotion-based systems. A block diagram of an inductive link used for wireless power transmission system is depicted in Figure 7c. The power signal is sent from the primary site (i.e., external unit) to the secondary site using two coils. A class-e transmitter is preferred to achieve a highly efficient power transfer (15, 22). The transistor device (i.e., T 1 transistor) acting as a switch is driven by a square wave at an oscillation frequency f osc. The network s resonant frequency should be the power transmission frequency and Figure 8. Nine-millimeter power-receiving coils and capsule prototype implementation. The capsule system uses a Helmholtz coil method and class E power transmitter for wireless energy transmission (12). is given by equation 1. The capacitor C 2 connected in series with the transmitting coil (L 2 ) is acting as a dc-blocking capacitor, and its value should be small enough to be nearly resonant with L 2. p f osc 1= 2p ffiffiffiffiffiffiffiffiffiffiffi p L 2 C 1 ¼ 1= 2p ffiffiffiffiffiffiffiffiffiffiffi L 3 C 3 It is clear from the battery-less systems that it is necessary to bring the power transmitter very close to the skin. This way, a wireless power system can transfer energy through the 10- to 20-cm thick skin to reach the device inside the body (26). As described previously, a wireless capsule device can also be used as a method of drug-delivery systems. Such capsules should be controlled mechanically to navigate the device to the correct location. This will require a wireless energy capability to maintain sufficient energy in a capsule device to achieve the required navigation and motion control. Wireless Communication Links It is commonly known that propagation through the body has higher losses at higher frequencies, and the characteristics vary depending on characteristics of the antenna. It is therefore important to optimize the through-body signal propagation together with the antenna to obtain optimum antenna radiation at the design frequency (27). Unlike the wireless transmission in the air, the signal propagates through a few layers of skin when a wireless capsule is making its journey. Because of the different tissue layers, the signal propagation will have different electrical properties in the intestine system with the dielectric constant ranging from 40 to 70 and the conductivity in the range of 0.7 to 1.9 S/m (28). The wireless capsule endoscopy communication system should be optimized according to these characteristics and should have following features: Compact and forming a pill shape Low power Wideband for high-quality image data transfer A communication between the capsule and the external unit should be formed using an appropriate frequency band. Commonly used frequency bands are UHF-433 MHZ ISM, MICS, and UWB. International communications authorities allocate the MHz band with 300-kHz channel bandwidths for communication between an implanted medical device and an external controller. The MICS band has been the most popular frequency band choice and is used in Given Imaging s wireless capsule endoscopy devices. The narrowband-based transmitter designs including 433 MHZ and MICS bands can support only a few hundred kilobytes per second (2). It is thus necessary to use higher frequencies with wider bandwidth to achieve a similar video quality obtained with that of wired endoscopy systems. As high-definition cameras are continuously being developed, they will be attractive for use in wireless (1)

8 8 Wireless Endoscopy capsules. However, a higher pixel camera will require a higher image transfer rate. As an example, if pixel (2 megapixels) charge-coupled device (CCD) sensors to be integrated in capsule device, it will require a data rate of 33.2 Mbit/frame, considering 2 bytes are used per pixel. Currently, such a high data rate transmission is not possible with any of the available frequency bands. If UHF frequencies are used because of the restricted bandwidth, then transmission of such a rate will result in a transmission time of 10 s per frame or more, which will result in very small motion for a video streaming. Although compression techniques could be used to some extent (29), it reduces the image equality. Thus, a wireless capsule endoscopy device should use a frequency band with larger bandwidth for high-definition image transmission. Wideband (Including UWB) Communication Link. A wireless endoscope is a biomedical system requiring a large amount of data that will be delivered to outside the body to achieve high-resolution pictures and images. For highdata-rate and short-range applications, wideband communication (e.g., UWB) is an ideal physical layer solution by achieving a data rate equal or higher than 100 Mbps (30). The UWB communication uses an operation frequency higher than 3 GHz ( GHz). A wideband wireless capsule device can transmit raw video data without any compressing, resulting in low-power, less delay in real time, and increased picture resolution. With a high-definition camera such as 2 megapixels, UWB telemetry can easily transmit up to 10 fps or more. In addition to its high-capacity wireless link, a wideband telemetry link can have some additional benefits because it operates based on pulses rather than sinusoidal carrier signal, enabling the low-power transmitter to increase the battery life and decrease the interference effect on the other wireless systems in medical centers. A high-frequency communication requires small electronic components such as capacitors and inductors, which helps to design a compact communication system. The radio signal propagation through tissue layers at the GHz frequency presents significant signal loss, and thus designing UWB link for wireless capsule is a challenging task. Recent developments have shown that through careful arrangement of signal power and optimizing channel model between the capsule and the receiver, a UWB communication link can be established that can enable a high-capacity wireless link for high-quality video data transmission (10, 30 33). Ultra-wideband communication transmits data using short pulses with 1 ns width or shorter. It is easier to obtain a modulated transmit signal with pulses compared with narrowband-based continuous signal transmission. Generating an ultra-wideband signal for data communication is more energy efficient. There have been different methods of generating transmission pulses to obtain wideband (34, 35). Figure 9 shows a very low-power circuit technique to generate narrow pulses and hence a wideband spectrum. As illustrated in Figure 9, a reference oscillator signal is digitized and its delayed replica is passed through the Exclusive-OR (XOR) gate to obtain a narrow square pulse (i.e., UWB pulse). After the signal is mixed with the medical data, the modulated signal is then passed through a band pass filter. The signal at the output of the filter is a wideband signal that can carry a large amount of data (Figure 9b). In addition to its high-data-rate capability, UWB technology provides some desirable advantages for wireless capsule technology including low-power transmitted design, low radio frequency and electromagnetic interference (EMI) effects in medical environment, and a size antenna. Unlike the narrowband technologies, UWB can support scalable data rates. For example, it can easily be designed to transmit from 10 Mbps to 100 Mbps without introducing much complexity in the hardware. In addition, UWB frequency bands are less crowded; thus, it will provide robust and reliable data communication from the capsule to the external monitoring device. Wireless capsule devices are located in the body with a distance of up to 10 cm. Thus, accurate knowledge of the propagation though inhomogeneous structure body is required to optimize a UWB-based wireless communication link. The performance of the communication link is affected by the dielectric properties of the human tissues, (a) fosc. Inverter/ bufer Digitised signal Tosc. XOR, AND..etc Image data x(t) Band Pass Filter Antenna a(t) Delay τ τ A Less than 1 ns 0 (b) Narrow pulse dbm Figure 9. Wideband signal-generation technique.

9 Wireless Endoscopy 9 Figure 10. Path loss in different tissue regions with the distance from the antenna toward the front of the human body (32). in addition to the distance between the transmitter inside the capsule and the external receiver. Several studies in the literature have reported UWB channel characteristics that can be used to configure transmitter and the receiver parameters such as transmit and receive power levels and the sensitivity (32, 36). Figure 10 demonstrates a simulation result for UWB signal propagation through the different tissue layers of the digestive system (32). Assuming the distance from the GI tract to the skin surface is about 8 9 cm, a receiver can easily be placed around the human belt or up to cm from the wireless capsule accommodating a signal loss from 80 db to 90 db. Assuming a UWB transmit signal has a power level between 10 dbm and 0 dbm, a receiver with a sensitivity of 100 dbm or 90 dbm is required to detect UWB signals. Several path loss studies for in body applications are demonstrated in Figure 11. The loss is plotted against its actual depth inside the body to a body surface device (11, 37, 38). Characteristics for commonly used frequencies are presented for a reference to the in-body UWB communication link. Table 3 gives average loss values for various transmission frequencies within the human body. It is clear that the loss varies from 40 to 90 db when a wireless link is used for wireless capsule applications, depending up the frequency selected. Therefore, the designers should consider accommodating this type of loss when designing a wireless link for a wireless capsule device. A simulation shown in Figure 12 was conducted in Reference 32 to obtain the corresponding temperature increase for different UWB signal power levels absorbed by the human body. The temperature increase is obtained after the steady state is achieved. The simulation considers the variation of basal metabolic rate and blood perfusion. The initial body temperature is considered to be 37 C. The results in Figure 12a illustrate that the temperature of the whole body has increased to C from the initial body temperature of 37 C. In short, the selected signal power Figure 11. Implant to body surface path loss studies: (a) a path loss study for in-body UWB communication link conducted in (11) and (b) a path loss study for implant to body surface at MICS frequencies (403.5 MHz) (37).

10 10 Wireless Endoscopy Table 3. Average Path Loss Measurements of Different Tranmission Frequencies (The External Receiver is on the Surface of the Body) Antennas Loss Frequency Penetration UWB (32) 86 db 4 GHz 10 cm UWB (11) 80 db 4 GHz 7 cm MICS (37) 60 db MHz 10 cm 418 MHz (38) 21 db 418 MHz 3 cm 916 MHz (38) 27.5 db MHz 3 cm levels of UWB pulses are not high enough to cause a significant temperature increase. It is necessary to analyze the electromagnetic effects of UWB signals. Specific absorption rate (SAR) and specific absorption (SA) are two indicators measured to define the safety limits of absorbed electromagnetic power by the human body. The internationally recommended SAR level regulated by the International Council on Non-Ionizing Radiation Protection (ICNIRP) standard and the Institute of Electrical and Electronics Engineers (IEEE), standard is 2 W/kg, which is 10 g averaged in the frequency range of 10 khz to 10 GHz (39, 40). Usually the ICNIRP regulated SAR level of 2 W/kg determines the maximum allowable delivered signal power to the antenna for UWB communication (32). The exposure of electromagnetic fields producing higher than 2 W/kg can cause an increase in body temperature that could affect the biological mechanism of the human body. The SAR variations for the power levels that are shown in Figure 12 meet the limit of 2 W/kg defined by the safety standards. Antenna Design The antenna is a crucial element in a wireless capsule endoscopy device. The design method used for antennas in wireless capsules is the optimization of an antenna with respect to a range of tissue properties surrounding the device (27) (as shown in Figure 1b). A device that travels in the GI system comes into contact with various tissue types and fluids, which have variable permittivity and conductivity. In addition, it is important to consider the physical shape of a wireless capsule package when designing an Figure 12. Side view of the temperature variation in the human for infrared-uwb signals with peak spectral input power limit of (a) 41.3 dbm/mhz and (b) 1.33 dbm/mhz, and a total signal power of (a) mw (SAR ¼ 0.2 mw/kg) and (b) 24 mw (SAR ¼ 2 W/kg) (32). antenna. The antenna can be inserted on top of the electronics boards. Alternatively, the capsule shape can also be divided into two regions where the antenna can be placed into the upper half whereas the remaining electronic units are packed into the lower half. Placing the antenna at one side of the electronic units in the capsule is another possibility. Recent antenna designs in the literature that use the shape of a capsule efficiently are shown in Figure 13. One of the earlier antennas designed for capsule endoscopy was reported in Reference 41, with dimensions of 8 mm diameter and 5.6 mm height. The antenna was matched in the band MHz inside a homogeneous tissue simulating phantom with a dielectric constant 56 and conductivity 0.8 S/m. The antenna has seven loops with 0.5 conductor width (Figure 13b). The antenna demonstrates an omnidirectional radiation pattern to obtain a signal independent of the transmitter positions. The bandwidth enhancement is achieved by connecting the end of the helical antenna to the ground. A similar antenna (42) reported by the same authors has a conical helix shape and can be placed at the top or bottom site inside a cylindrical capsule. The antenna has bandwidth of 101 MHz from (418 MHz 519 MHz) with an omnidirectional radiation pattern. It has a diameter of 10 mm and a height of 5 mm. A modified stacked spiral-shaped wideband antenna for capsule endoscopy that operates from 411 MHz to 600 MHz was presented in Reference 43. In this antenna, two spirals are connected to a feeding line and it has a diameter of 10 mm and a height of 7 mm. Two spirals have about five turns and are separated with a gap of 4 mm (Figure 13c). In Reference 44, another spiral antenna for a wireless capsule endoscope system was presented. The antenna has successfully been used in wireless telemetry based on ON- OFF keying (OOK) modulation with pixel image transmission. This design is similar to those in References 42 and 43 operating at 500 MHz with bandwidths of 104 MHz. The spiral antenna is structured on a 1-mm-thick ground plane based on a spiral arm with width of 4 mm. The antenna has a diameter of 10 mm and a height of 5 mm. Some researchers have investigated antenna designs printed directly on the surface of the capsule casing. For example, the antenna shown in Figure 13f (45) was proposed for wireless endoscope systems. The antenna operates in the 1.4 GHz telemetry band and considers the presence of the small intestine and other tissue materials. The antenna has good polarization characteristics that support various orientations of the capsule and has a dipole-like radiation pattern with an omnidirectional pattern in two planes. The antenna is designed taking into consideration the average body permittivity of 58.8 and a conductivity of 0.84 S/m. Figure 13a shows a UWB antenna designed for wireless capsule applications (46). This antenna operates from 3 to 5 GHz to enable a large bandwidth for a high data rate transmission. Another UWB antenna (Figure 12e) for capsule applications is described in Reference 47. It is based on a planar loop antenna. It operates in the UWB low band of 3.4 GHz to 4.8 GHz. A compact monopole spiral antenna based on medical implant band (MICS) is presented in Reference 48. It has dimensions of 7 mm 14 mm. The length of the wire used to form the spiral

11 Wireless Endoscopy 11 Figure 13. Antenna designs for wireless capsules: (a) capsule-shaped UWB antenna, (b) helical antenna (41), (c) stacked spiral antenna (43), (d) wideband spiral antenna (44), (e) planar UWB loop antenna (47), and (f) conformal dipole antenna (45). antenna is 26 mm, which is a quarter wavelengths for the frequency of 402 MHz for a permittivity e r of 55 and a conductivity of 0.7 S/m. The antenna is designed on a substrate with a relative permittivity of 2.2 and a thickness of 0.5 mm. The main differences between in-body antennas and external antennas are the strict size constraints and demand for high efficiency. Substrates with high dielectric constants are helpful in miniaturization, but the increased conductivity of the body tissues with frequency becomes an important problem for antenna operation (27, 46). Table 4 summarizes the existing antennas described in the literature in terms of their operation frequency, size, and S11 performance. The operating frequency bandwidth (BW) of antennas is obtained by the 10-dB band. It is important to use a wide BW antenna because a communication system for wireless capsule endoscopy technology will work efficiently despite the frequency shift in the antenna operation. The signal transmission will still fall within the 10-dB frequency BW. A wide BW will be useful for video-based wireless capsules because of the high data rate requirement. LOCALIZATION AND CONTROL TECHNIQUES WITHIN THE GI TRACT It is essential to have an accurate knowledge of the position and orientation of the capsule when it moves along the GI tract where the endoscopic images are captured (Figure 14). The success of drug administration, follow-up interventions and other therapeutic operations heavily depend on the accuracy of this spatial information. Table 4. Comparison of Antennas for Wireless Capsule Devices Antennas Therefore, having a precise and reliable localization system plays an important role in enhancing the capabilities of a capsule. In general, two localization methods exist: the magnetic localization methods and the electromagnetic localization methods (49). Magnetic Methods 10 db frequency band Dimensions (mm mm) central frequency (db) UWB (46) GHz Planar loop GHz 11 mm 20 UWB (47) diameter Spiral dipole 1.4 GHz (45) Spiral (44) MHz MHz BW Stacked MHZ spiral (43) 189 MHz BW Conical MHz spiral (42) 101 MHz BW Helical (41) MHz MHz BW Spiral antenna (48) MHz (402 MHz) Magnetic localization methods are used because 1) lowfrequency and dc magnetic signals can pass through human tissue without any attenuation (50) and 2) magnetic

12 12 Wireless Endoscopy Figure 14. Position and orientation of the capsule when it is inside the GI tract. localization is a non line-of-sight method, in which the capsule does not need to be in the line of sight with magnetic sensors in order to be detected (51). In addition to being used for the localization, the negligible interaction between the magnetic field and the human body is also employed to establish locomotion systems that can guide endoscopic capsules magnetically such as robotic magnetic steering (52 54), helical propulsion by a rotational magnetic field (55 58), magnetic levitation (59, 60), and remote magnetic manipulation (61). Furthermore, a localization system can be used to provide feedback control information for an actuation system (62). In this regard, the actuation and localization systems should work together during a diagnostic procedure. However, the interference between the two magnetic fields is a challenging research problem (49, 52, 63). Depending on how a capsule is propelled in the GI tract, the magnetic localization methods can be divided into two distinct classes. Magnetic Localization of Capsules Under Natural Peristalsis. A magnetic source and a sensor are the most important elements of a magnetic localization system. A stable and reliable source of the magnetic field is essential for any real-time magnetic localization system. Depending on how the magnetic source is created and whether the capsule acts as a field generator or contains a sensor, the localization systems in this group are divided into three subgroups. In the first subgroup, a permanent magnet, which is placed in the capsule, creates a uniform magnetic field. The magnitude and direction of the magnetic field depends on the magnet s position and orientation. Magnetic sensors placed on the exterior of a patient s body measure the magnetic field. An analytical mathematical model describing the relationship between the magnetic field strength and location of the permanent magnet can be employed to determine the position and orientation of the capsule, i.e., to localize it. For example, if the magnet is modeled as a magnetic dipole, then the magnetic flux density around the dipole source is given by (64) ~B ¼ B x ~ i þ By ~ j þ Bz ~ k ¼ m 0 4p 3ð~m ~rþ~r j~r j 5 ~m j~r j 3 where B x, B y,andb z are the components of the magnetic flux density; ~m is the magnetic dipole moment of the magnet; ~r is the position vector of the magnet; and m 0 is the air magnetic permeability (4p 10 7 JA 2 m 1 ). The localization parameters of the magnet (i.e., the capsule) can be determined from equation 2. Weitschiles et al. (65, 66) have employed this technique with a 37-channel, superconducting quantum interference device sensor system to monitor the position of the capsule in the GI tract, without considering the capsule orientation. However, one issue associated with this position monitoring system was that it had to take place in a magnetically shielded room to reduce the effect of the environmental magnetic noise on the measured magnetic flux density. Schlageter et al. (67, 68) employed a two-dimensional (2D) array of sixteen Hall sensors to determine both position and orientation of a pill-size magnet coated with silicone. When the magnet with the volume of 0.2 cm 3 movedupto20cmfromthesensor plane, its position and orientation were determined at the rate of at least 20 Hz. One limitation of this approach is that the system s accuracy drops significantly with the distance (>20 cm) between the magnet and the sensor array. Chao et al. (50) employed 50 cm 50 cm 50 cm cubic sensor arrays to deal with this limitation, as shown in Figure 15. Each of the four sensor planes contains 16 magnetic sensors. An average position error of 1.8 mm and orientation error of 1.6 were obtained. Many attempts including using 3D sensor planes (69 76) have been proposed to enhance the accuracy and widen the sensing volume of this localization system,butwithanincreased cost and complexity.! (2)

13 Wireless Endoscopy 13 Figure 15. Scheme of the cubic magnetic sensor array and its setup (50). To reduce the cost, Aziz et al. (77) proposed a tracking system based on only three 3-axis magnetic sensors placed orthogonally in 3D space and an extra sensor for cancelling environmental magnetic noise. A position error of up to 3 cm was obtained when a F5mm L6.0 mm cylindrical magnet was tested in a volume of 10 cm 10 cm 10 cm. Wu et al. (78) built a wearable magnetic localization system for the volume of 40 cm 25 cm 40 cm with six sensing modules at the front frame and the other four at the back frame, as shown in Figure 16. Each module is composed of six linear magnetic sensors that form three pairs of back-to-back sensors arranged perpendicularly to each other in three dimensions. Each pair is responsible for measuring one dimension of the magnetic field. In this arrangement, the top two modules at the front were employed to eliminate the interference of the earth magnetic field. Because this localization system is based on the mathematical model of a magnetic dipole, when the capsule is close to the sensing module, this model is no longer valid, resulting in a decrease in the localization accuracy (78). Moussakhani et al. (79, 80) have derived a path loss model for the wireless channel inside the human body to simulate the electromagnetic wave propagation in human tissues. This model is then employed to determine numerically the theoretical performance limit, which is reported to be 1 cm. Assuming that the capsule endoscope contains a cylindrical magnet magnetized in the axial direction, they have theoretically demonstrated that the localization accuracy can be in the order of 2 3 mm at best. No experimental results are presented to substantiate these claims. In the second subgroup, a coil is placed in the capsule. Plotkin and Paperno (81, 82) proposed the idea of tracking a receiving coil by a large 2D array of transmitting coils. The magnetic field seen by the receiving coil is given by ~B r ¼ ~ B t ~ M (3) where ~ M is the magnetic moment of the receiving coil and ~B t is the magnetic flux intensity generated by the transmitting coil. Because ~ B t can be expressed approximately by equation 2, computing the localization data is similar to that of a permanent magnet. In other words, an optimization algorithm can be employed to solve the inverse problem based on equation 3 once the electromotive force induced in the receiving coil has been measured. In an attempt to apply this localization concept to a capsule, Nagaoka and Uchiyama (83) designed a single-axis coil with 160 turns of copper wire with the size of F6.5 mm L2.3 mm to be inserted into a capsule. A magnetic field generator placed outside the body produced five alternating magnetic fields, each with different frequencies. The electromotive force created by mutual induction was sent to the outer detector through transmitting 75-kHz signals from an integrated FM circuit. Because the magnetic field strength decreases proportionally with the inverse third power of the distance between primary and secondary coils, the current flow in the primary coils was controlled automatically to keep the induced electromotive force within a constant range. To implement this approach, a power amplifier was connected to the generator. The received data at the detector not only were used for Figure 16. Wearable sensing modules by Wu et al. (78).

14 14 Wireless Endoscopy estimating the capsule s position but also were used for providing feedback signals continuously with the power amplifier. Because the capsule cannot move with a high speed in the GI tract, such a feedback signal does not affect the tracking rate of the system (83). Although it was reported that the system demonstrated an accuracy of 5 mm when the capsule was 50 cm away from the generator, the experiments failed at several locations. As an example to the third subgroup, Guo et al. (84) developed another solution for the localization problem by placing a three-axis magnetoresistive sensor inside a capsule to measure the intensity of the external magnetic field generated by three energized coils placed on the patient s abdomen. The three coils are excited in turn by square waves with the same period of 0.03 s. At the end of every cycle, there is a break period of 0.1 s when the coils are not activated to estimate the Earth s magnetic field magnitude. To obtain the real data of the magnetic field generated by the coils, the Earth s magnetic field is subtracted from the total magnetic field measured by the sensor at the capsule location. A neural-network algorithm is employed to estimate the three positional and three angular coordinates of the capsule. However, it must be noted that this is not a real-time localization system because the estimation procedure is done after the completion of the experiments. Another important point to keep in mind is that when the sensor is close to the coils (less than 50 mm), the magnetic dipole assumption fails, which causes significant localization errors. This situation becomes even worse when the coils diameters are increased for enlarging the localization range (84). Therefore, an improved localization model based on Biot-Savart law was proposed to replace the magnetic dipole model (85). It was reported that the position and orientation errors range from 6.25 mm to mm and from 1.2 to 8.1, respectively. Eight energized coils excited by sinusoidal signals instead of square waves are employed to improve the accuracy of this method (86). After implementing an adaptive particle swarm optimization technique, the mean position and orientation errors of 14 mm and 6.9 were obtained, respectively. SUMMARY One disadvantage of the magnetic localization systems is that all devices or equipment around them must be nonferromagnetic. Another disadvantage is that their coverage volume is limited (87). Additionally, how to eliminate or minimize the interference between the magnetic localization and magnetic actuation (if applicable) is a significant issue. Alternately switching ON or OFF between the actuation and sensing (localization) has been suggested to solve this issue (63). But, because of the hysteresis characteristics of a magnetic field, the external magnetic field in the actuation mechanism is still ON for a certain period of time after it has been turned OFF. As a result, it would be virtually impossible to do a real-time localization until the external magnetic field has completely diminished. Needless to say, during the off period of the magnetic actuation, the capsule may move to a new position and orientation. In that case, the tracking system will not provide accurate feedback data for the actuation system. Another solution for this problem, is to use a high-frequency alternating magnetic field (88, 89), which is presented in the next section. Magnetic Localization of Capsules Under Active Actuation. Magnetic Localization via Measuring High- Frequency Alternating Magnetic Field. The capsule propulsion is based on a spiral or spirals assembled on the surface of a capsule that contains a permanent magnet (88). An external rotating magnetic field was generated around the patient s body using three pairs of coils (i.e., Helmholtz s coils) placed in three perpendicular directions to rotate the capsule and hence propel it forward or backward depending on the direction of the current applied. The frequency of the rotating magnetic field should be less than 10 Hz because the capsule is not allowed to move too fast in the GI tract. Because the lowfrequency (several Hz) rotating magnetic field does not interfere with a high-frequency (from 1 khz to 1 MHz) alternating magnetic field (88, 89), exciting coils were placed around the patient s body to produce the high-frequency magnetic field for localization purposes. The operating frequency within the range of 1 khz to 1 MHz is sufficient to avoid the attenuation of a magnetic field while passing through the human body. Detecting coil arrays are also placed around the patient to measure the magnetic field induced by a resonating coil integrated inside the capsule and thus to determine the position and orientation of the capsule (90, 91). Magnetic Localization via Inertial Sensing. The capsule can be propelled via magnetic steering, for example, using a 6 degrees of freedom robotic manipulator carrying a Figure 17. Permanent magnet mounted at the end effector of a 6 degrees of freedom robot and four cylindrical permanent magnets placed in the capsule (52).

15 Wireless Endoscopy 15 permanent magnet at its end effector, as shown in Figure 17 (52). A permanent magnet assembled in the capsule will create a magnetic link between the capsule and the external permanent magnet and, hence, drag and steer the capsule in the body. A three-axis accelerometer is assembled in the capsule to provide the approximate location and orientation of the capsule in the digestive tract and, if needed, to provide valuable feedback data to the actuation system in order to verify the magnetic link between the external permanent magnet and the capsule. It was reported that this system could provide position and orientation accuracies of 3 cm and 6, respectively. The position of the capsule can be estimated from the position of the end effector as long as the magnetic link is still maintained during the steering process. This system does not have a conflict between its actuation and localization modules. However, the space available in a capsule is limited to accommodate both the permanent magnet and the inertial sensor. Perhaps the most significant disadvantage of this method is that the capsule has to be dragged on the tissue, which can cause significant damage to the contacting tissue. Magnetic Localization Via Measuring a Rotational Magnetic Field from a Permanent Magnet. Kim et al. (92) assembled a helical structure on the surface of the capsule and used an externally generated rotational magnetic field to rotate the capsule containing two permanent magnets. Instead of using six coils around the patient s body, a big parallelepiped permanent magnet consisting of seven smaller rectangular magnets was rotated to generate the rotational magnetic field. This magnetic field generator was driven by an electric motor mounted on a manipulator so that it could rotate and be moved while propelling the capsule, as shown in Figure 18a. One notable feature of this method is that the external magnetic field generated for actuation can be used for localization. When the external permanent magnet is spinning, the magnetic field strength at the capsule location changes periodically. Three Hall effect magnetic sensors set up orthogonally inside the capsule, as shown in Figure 18b, are used to provide the capsule position data. Once the capsule position is determined, a rotation matrix that represents the orientation of the capsule in three dimensions is obtained by comparing the three calculated orthogonal components and the three measured orthogonal components of the magnetic flux density at this position. This localization system generated x, y, andz position errors within the ranges of (þ2 mm, þ15 mm), ( 9mm, þ12 mm), and ( 10 mm, þ3 mm), respectively. The orientation errors were within the ranges of ( 2, þ13 ) in pitch direction and ( 4, þ11 )inyaw direction. The methods presented previously can be used for localization and actuation simultaneously. However, they are costly and bulky and far from being used in clinical applications yet. Therefore, there is still an increasing need to establish an effective localization technique. Electromagnetic Localization Methods The electromagnetic waves-based localization methods can be used together with the magnetic actuation, which appears to be a feasible method to remotely propel, localize and control the motion of the robotic capsules. While radio waves, visible waves, X-ray, and gamma ray of the electromagnetic waves schematically shown in Figure 19 can be used for capsule localization or tracking, microwaves, infrared waves, and ultraviolet waves, which have very low penetrability through the human tissue, are not suitable for localization or tracking. Radio Waves. Although radio waves have been widely used for locating an object in outdoor and indoor environments with the accuracy of hundreds of millimeters (93), applying the radio waves in tracking a capsule within the GI tract is not straightforward. This is because high-frequency signals attenuate significantly at different levels when they pass through different human tissues, whereas low-frequency signals because of their long wavelengths cannot deliver the desired precision of several millimeters (76). Based on the radio frequency, received signal strength indicator (RSSI), angle of arrival (AOA), time of arrival (TOA), time difference of arrival (TDOA) and RFID techniques can be employed for localization. Of these, RSSI and RFID have a practical significance for the capsule localization. In contrast, TOA is not feasible because the radio waves travel with a very high speed ( m/s); thus, an extremely strict time synchronization less than 1 ns is required to obtain an acceptable position resolution. AOA is not reliable even in well-structured environments (94, 95). Figure 18. (a) Rotating a permanent magnet for generating a rotational magnetic field. (b) Sensor module scheme inside the capsule (92). DoF ¼ degrees of freedom.

16 16 Wireless Endoscopy Localization of WCE Radio waves Hz Microwaves Hz Infrared Hz Visible waves Hz Ultraviolet Hz X-ray Hz Gamma-ray Hz Figure 19. Electromagnetic waves for wireless capsule endoscopy localization. Received Signal Strength Indicator. The wireless capsule endoscope is equipped with a telemetry capability. Fisher et al. (96, 97) used this advantage to measure the strength of the received RF signals at eight sensors placed uniformly on the patient s abdomen. The closer the receiver is to the transmitter, the stronger the signal. These signals are used to determine the position of the capsule. When two adjacent antennas receive equal strength signals, the capsule is assumed to be in between them. This localization technique isbeingusedinthegivenimagingm2acapsuletoestimate the 2D position of the capsule with the accuracy of 3.77 cm. Although this technique does not require any additional hardware in the capsule, its low accuracy makes it unsuitable for providing feedback data with a possible actuation system. Arshak and Adepoju (95) have proposed an empirical signal propagation model that describes a relationship between the RSSI value and distance from the transmitter to the receiver (98); RSSIðdÞ ¼P T PLðd 0 Þ 10n log 10 d d 0 þ X s (4) where d is the distance between transmitter and receiver, P T is the transmit power, PL(d 0 ) is the path loss for a reference distance d 0, n is the path loss exponent, and X s is a Gaussian random variable. Equation 4 can be used to determine the distances between the capsule and each of the sensors. A trilateration method is used to calculate the capsule location for these distances from the transmitter to the receivers. In place of equation 4, Shah et al. (99) proposed an algorithm based on a lookup table for the position estimation. In order to reduce the estimation error of this method, it is necessary to develop a more appropriate attenuation model when RF signal travels within the human body. Radio Frequency Identification. A cubic antenna array is built surrounding a patient s body to receive signals from an RFID tag placed in a capsule ( ) (Figure 20). The center of gravity principle is applied to the collected data to estimate the position of the tag, which is the position of the capsule. However, this tracking algorithm produced large errors because of the limited accuracy of the center of gravity principle. To deal with this problem (101, 102), a bidirectional antenna has been used to transmit RF signals in two opposite directions. The position errors of 0.5 cm in x and y directions and 2 cm in z direction were estimated using simulation data. The frequency is limited below the UHF band for the RF signals in order to pass through the human body. In this band, it is impossible to generate directional radiation by a compact antenna less than 1 cm in length. Another significant drawback of this system is that when the longitudinal axis of the tag s antenna is in the same direction with the main axis of the patient, the radiation pattern does not intersect with the cubic array, and thus the matching algorithm will not be able to determine the position of the capsule. To solve this problem, at least one more tag in a perpendicular direction with the first tag needs to be assembled in the capsule. Another technique for the localization of an RFID tag placed in a capsule is based on phase difference. In a system with one transmitter and several receivers, the Figure 20. Antenna array and radiation pattern of RF signals transmitted from an RFID tag (101).

17 Wireless Endoscopy 17 RF waveforms at the ith receiver can be described by the following equations (103) I i ðtþ ¼A i cosð2pðf r f c Þt þ f i Þþs i n i1 Q i ðtþ ¼A i sinð2pðf r f c Þt þ f i Þþs i n i2 (5) where I(t) and Q(t) are the in-phase and quadrature components of the signal; A i, f r, f c, f, s, and n denote the received signal magnitude, the frequency at the receivers, the carrier frequency, the phase difference between the carrier at the tag and the carrier at the receiver, the noise level, and Gaussian noise, respectively. Hekimian- Williams et al. (103) showed that although the exact phase value (1 i ) of a single signal received at an antenna cannot be used to calculate the distance that the signal has traveled, the phase difference (1 i 1 j ) between signals within the same burst arrived at different antennas can be employed for the location estimation. Wille et al. (104) developed an RFID navigation system using the phase difference to track medical instruments such as needles or catheters. Support vector regression (SVR), a machine learning algorithm, was applied to estimate the position of the RFID tag by employing the phase difference data collected at different RFID receivers. Although the experimental results indicate the feasibility of the phase difference method for accurate localization of RFID tags, a lot of improvements are needed before applying this method to capsule tracking. The reason is that two important factors that could affect the accuracy of the localization method were ignored during the experiments. One of these factors is the orientation of the tag, which was kept constant in all of the tests. The second factor is that the attenuation of the RF signals through the human body was not considered. Visible Waves. Even though visible waves cannot penetrate the human body, it has still been considered for the capsule localization through computer vision ( ). These localization methods provide only basic information about the location of a capsule, which is not sufficient. Therefore, the localization information can be considered only as reference or complementary information for the endoscopists. X-Ray. X-rays can be employed to track an endoscopic capsule. Fluoroscopy, which is a type of imaging technique based on the X-ray radiation, is used to display continuous X-ray images in real-time, which shows the location of the capsule (53). However, this method can only supply visual information of the capsule location via radiation images. It is impossible to obtain actual parameters of its position and orientation to serve as feedback data for an actuation system. Aiming to solve this issue, Kuth et al. (108) proposed a method that takes advantage of both X-ray imaging and image processing for automatically determining position and orientation of the wireless capsule endoscopy. Gamma Ray. Gamma rays are used in gammascintigraphy technique to visualize the position of an Enterion capsule, a drug-delivery type capsule, in real time (109). The capsule that is loaded with gamma-emitting radioisotopes can be detected by scintillation cameras. Because gamma rays are partly absorbed by the human tissues when they travel from the radioactive source to the camera, both dorsal and ventral images are taken to enhance the tracking accuracy (110). However, similar to an X-ray based localization system, this method can be harmful to patients. Recently, Than et al. (111) used gamma rays to estimate the position and orientation of the capsule in a phantom replicating the conditions in the human body. No battery consumption and zero space occupation inside the capsule are the primary advantages of this localization method. This method can provide less than 0.5 mm position error, and 2.1 orientation error in a localization time interval of 50 ms with an average computational time of 7 ms per time interval. These are the best localization performance data reported in the literature. Other Localization Methods Magnetic resonance imaging (MRI) and ultrasound, which are the widely used diagnostic imaging techniques, can be used for localization ( ). The need for custom-programmed pulse sequences, which are different from the standard pulse sequences of commercial MRI scanners, would be a disadvantage for an MRI-based method (117, 118). In contrast, although the bones and gas shield ultrasonic signals (119), the localization method based on ultrasound offer the features of high speed, safety, and low cost (120). Comparison and Discussion As presented in Table 5, the approaches that have a promising potential for an accurate localization are either influenced by the magnetic field to be used for actuation or too complex and still at their proof-of-concept stage. The next generation wireless capsule endoscopy is expected to have full robotic capabilities (121, 122) such that it will be able to accomplish both diagnosis and disease treatment. To deliver these functions, building a complete localization system, which is reasonably accurate in real time, minimally invasive, able to work with an active actuation system, and easily implementable, is greatly desirable. This is a significant challenge for researchers to address within the next 5 years. Possible solutions include designing novel approaches, improving the proposed methods, or even developing hybrid strategies to exploit a combined advantage of different techniques. ROBOTIC WIRELESS CAPSULES AND DESIGN CONSIDERATIONS Minimally invasive surgery uses cutting-edge technology to diagnosis and implement therapy with the expected advantages of fewer traumas to the body, a shorter recovery time, and shorter hospital stay than traditional surgical methods. Wireless capsule endoscopy is one of the successful medical technologies for the minimally invasive surgery (123). Recently, substantial national and international research efforts have been directed to actively propel the capsule endoscopy and accurately localize it in order to

18 18 Wireless Endoscopy Table 5. A Comparison of the Key Methods Presented in this Study. Take Extra Space of the Capsule Consume Extra Power of the Capsule Accuracy Interference with Magnetic Actuation Real Time Adverse Health Effects Magnetic Permanent magnet Yes No High Yes Yes No localization (passive WCE) (50, 78) Secondary coil (74) Yes Yes Moderate Yes No Magnetoresistive Yes Yes Moderate Yes No No sensor (84 86) Magnetic HF alternating Yes No High No Yes localization (active WCE) magnetic field (90, 91) Inertial sensing (52) Yes Yes Low No No Rotating external Yes Yes Moderate No No permanent magnet (92) Electromagnetic Radio frequency (96, No No Low No Yes No waves 97) Visible waves (105, No No Low No No 107) X-ray (108) No No No Yes Yes Gamma ray (109, 110) Yes No No Yes Yes Others MRI (112, 114) Yes Yes High Yes Little Ultrasound (115, 116) Yes Yes No Yes Little a The accuracy may be high (position error <2 mm), moderate (position error is from 2 mm to 20 mm), or low (position error >20 mm). The symbol within a cell indicates that the information is unknown. enhance its diagnosis and therapy features, and hence to make it a truly robotic minimally invasive medical device. To this end, on-board and off-board activation propulsion concepts have been proposed and demonstrated under overly simplified operation conditions. The literature contains several propulsion systems such as earthworm-like (124, 125), paddling-type (126, 127), legged (128), electrically driven (129), and propeller-driven (130) capsules. All of these systems require onboard batteries or an energy source to provide power to the locomotion systems. Because the size of wireless capsules is of primary importance, it is better to take the power supply out of the capsule so that more space can be saved for other functional modules and the operation time can no longer be limited. To achieve this, the magnetic actuation is considered as the best choice to propel a robotic capsule externally with an on-board magnetic source ( ). In addition to these efforts, there is a need to incorporate biopsy, suturing/clipping, and on-site drug-delivery and other similar medical interventional capabilities into a robotic capsule to widen its versatility and effectiveness in diagnosing, and treating abnormalities related to the GI tract ( ). One effective and straightforward method to propel a robotic capsule is to generate magnetic gradient fields (via a Maxwell coil system), which induces a direct pulling force on a robotic capsule containing a permanent magnet ( ). Another method is to assemble a spiral structure on the outer surface of a magnetic capsule and then use magnetic torques to rotate the capsule, as shown in Figure 21. When the helices interact with the working Figure 21. (a): Propulsion of a millisized robotic device with spirals. The cylindrical shape will be the initial shape for performance optimization studies. (b) The propulsion concept based on a rotational magnetic field.

19 Wireless Endoscopy 19 environment (fluid or solid) inside the organ, the microrobot can be propelled by converting the rotation to a translational movement like a screw ( ). To induce a magnetic torque constantly on this kind of spiral-type microrobot for medical use, a rotating magnetic field with uniform magnitudes is required. As far as generating external magnetic fields is concerned, electromagnets show better performance than permanent magnets in terms of controlling the field strength and direction (130). Therefore, an electromagnetic system should be employed as the external magnetic source when accuracy is at a premium. The passive capsules are moving under the natural peristalsis within the GI tract (that is, the contraction of smooth muscles to propel contents through the digestive tract). They cannot be stopped or turned around while within the body, and they cannot be actively navigated in the GI tract. These distinct disadvantages greatly diminish the effectiveness of these capsules for accurate diagnosis and limit therapeutic capabilities. Therefore, significant research efforts and resources have recently been directed toward making the wireless capsule as an actively controllable capsule robot to realize diagnostic, therapeutic, and surgical functions, such as noninvasive GI surgery and targeted drug delivery (4, 119, 138, ). Traditional Robotic Approaches. Despite the significant research progress in the traditional macrosized robotic systems, the progress in miniaturized robotics has been limited because of the difficulties associated with powering and actuating them using onsite means, such as an electric motor and a battery with enough energy density (119). The existing actuation and power supply cannot be scaled further to realize functional in-body robotic systems. This is not a feasible and practical method for functional in-body robotic systems because a battery with required energy density will increase the physical dimension of the system. Magnetic actuation is of particular interest for miniaturized robotic systems operating in confined spaces as a magnetic force acts without any physical contact over large distances. Swimming Microrobots. Some studies have proposed propulsion concepts for magnetically propelled swimming robots for operation within veins and arteries ( ). Most of these studies are empirical studies characterizing the behavior of microrobotic devices experimentally and using simplified mathematical models to understand the theory behind their operation. Abbott et al. (149) have compared three possible biologically inspired propulsion methods based on external magnetic actuation for swimming microrobots. Martel et al. (153) has pioneered the research and development of 1) magneto-taxis propelled in the blood vessels by magnetic gradients generated by a magnetic resonance imaging (MRI) system and 2) development of new microelectromechanical and nanoelectromechanical systems based on the integration of bacteria as biological components. Sitti (150) reported that 1) there are many fundamental and applied research issues associated with the propulsion of microrobots before they find applications in medicine and other areas, and 2) making, powering, and steering miniaturized robots are equally crucial research problems, which must be solved urgently. Dreyfus et al. (151) built a flexible artificial flagellum, which is a microswimmer, consisting of a chain of magnetic particles linked by the DNA and attached to a red blood cell. They demonstrated experimentally that the microswimmer was propelled with an external uniform magnetic field to control its velocity and direction of motion. External Magnetic Actuation. As stated, space has been the primary constraint preventing the use of onsite actuators and power sources for propulsion. Some attempts have been made to transfer energy wirelessly to the actuators to navigate the robotic capsule, but these efforts failed because of the space issue and the complex energy transduction principle based on coils and storing harvested energy before use. Therefore, almost all realistic propulsion concepts proposed in the literature for robotic capsules are based on magnetic propulsion. Even so, there is still no functional robotic capsule externally controlled to operate safely within the GI tract to perform diagnostic and therapeutic functions. Lien et al (154) reported on a hand-held magnetic controller for a robotic capsule for 3D movements within the stomach. A permanent magnet is placed in a cylindrical capsule to allow remote maneuverability through the handheld device containing a permanent magnet and a stepper motor. Yim and Sitti (155) have proposed a magnetically actuated robotic capsule based on rolling locomotion for operation in stomach. In many previous studies (4, 138, 148, ) and in our recent study (2, 10, 131, 132, 137) to evaluate the feasibility of a robotic capsule propulsion concept based on a rotary magnetic field, an effective distance between the external magnetic source and internal magnetic source has to be maintained to create enough propulsion force. Furthermore, in an unpublished study, we equipped a robot manipulator (ABB IRB120) with a permanent magnet to navigate a cylindrical robotic capsule containing a permanent magnet in a transparent model of the GI tract, as shown in Figure 22. We found that the motion of the capsule could be discontinuous and the friction between the capsule and the GI tract played a critical role for smooth propulsion. If this propulsion concept were tested in a real GI tract or porcine intestine sample, the tissue squeezed between the capsule and the external magnetic field would easily be Figure 22. Robot manipulator propelling a robotic capsule inside the transparent model of the GI tract.

20 20 Wireless Endoscopy damaged. The majority of propulsion concepts proposed in the literature for robotic capsules for operation in the digestive tract are made of rigid surfaces, which potentially apply significant stress on the tissue. It follows that it is difficult to control the magnetic forces propelling such capsule robots with no damage to the contact tissue. More importantly, these devices cannot stably resist the peristalsis forces in the GI tract to anchor themselves for detailed imaging of target lesions, and they carry out other tasks such as delivering therapeutic agents (4, 119). Therefore, there is an increasing demand for novel robotic capsule propulsion concepts. Design Considerations for a Robotic Capsule Assuming that the capsule preserves its current features of telemetry, imaging, swallowable size, minimum invasion, and safe operation, the new features the next-generation wireless capsules should have include active locomotion, accurate localization, soft anchoring, on-board ph, temperature, pressure, blood sensors, interventional capabilities such as local drug delivery, biopsy, and suturing. In addition, the capsule should allow parameter variations in quick, easy, nondestructive, and straightforward manners before using it for a patient. As argued in the previous section, the propulsion of a spiral-type robot based on a rotating magnetic field is advantageous because the maximum torque available to the robot is proportional to the magnetic field intensity, which declines slower than the magnetic field gradient over a long distance (131, 132). The most important element of such a robot is its traction element, which is the spiral converting the rotational movement into a rectilinear motion. We have optimized the geometry of the helical structure because it plays a significant role in determining the propulsion efficiency. As a medical robot traveling in a deflated, winding, and slippery lumen, the complexity of its working environment makes this optimization problem even more critical. Therefore, the resistant characteristics of the GI tract should be evaluated to provide more accurate data with the optimization process. The results in Figure 23 indicate that the frictional torque increases as the rotation frequency increases, indicating the rotational resistance has a relationship with the rate of strain of the small intestine. This dependence reveals the viscoelasticity of the GI tract to some extent. After the introduction of the spiral, the cross-section of the capsule in the lateral direction is raised, which increases the deformation of the intestine. This increase becomes larger when the helical angle gets smaller. Therefore, when the number of spirals is the same, a capsule with a smaller helical angle causes more strain in the intestine and, consequently, confronts a higher frictional resistance. In this case, from 0.5 Hz to 3 Hz, the capsule with the helical angle of 10 causes a torque in the range of 0.6 to 1.8 mnm, whereas the capsule helical angle of 5 results in the torque magnitude between 0.8 and 2.3 mnm. At low frequencies, the torque is almost proportional to the frequency and the proportionality constant is larger when the helical angle is smaller. We have proposed an analytical model to estimate this fictional torque and experimentally verified it (131). Also, we have established a conceptual framework for designing and optimizing the topology of a spiral-type robotic capsule propelled in a fluidic and tubular environment using electromagnetic actuation (132). For each capsule, a segment of copper wire (w1 mm) was wound around the outer surface and acted as the spiral structure. The winding area was within the cylindrical part of the capsule so that every spiral structure had the same dimensions (15 mm) in the longitudinal axis, as shown in Figure 24. Dummy Pillcam SB2 capsules (Given Imaging) are used the base of the robotic capsule. We have found that adopting a relatively small lead and using at least two spirals can make a spiral-type capsule more balanced. Regarding the optimized number of spirals, the simulations and two sets of experiments are in agreement, indicating that a capsule wound with two spirals performs satisfactorily from the propulsion efficiency point of view. The capsule assembled with two spirals with a square cross section of 1mm 1 mm and a lead of 12 mm exhibits the best performance when magnetically propelled in the small intestine of a pig under a rotating magnetic field with the frequency of 4 Hz (132). Clearly, the current research efforts show (157) that there is a need to provide solutions for many fundamental and applied research issues associated with the nextgeneration capsule robots with active navigation including anchoring and localization. To this aim, breakthrough methodologies and solutions that will enable the next generation pill-sized capsule robots and establish novel methodologies to predict and optimize the performance of these remotely actuated robotic capsules are needed. Figure 23. Variation of the frictional torque with the helical angle of the spirals and the rotational frequency of the capsule. Figure 24. Capsules assembled with different spirals.