Chapter 3 Medical Image Processing

Transcription

1 Chapter 3 Medical Image Processing Medical image processing is application area of digital image processing in which the signal is medical image. The technique or process works as creating visual representations of the interior of a body for clinical analysis and medical intervention. Medical imaging helps in retrieval of internal structures of human body, skin and bones for diagnosis purpose and treatment of associated disease. Medical imaging is a method of acquiring medical images such as MRI, X-rays, CT images etc. and a database is created to identify the abnormalities or masses present in the images. Small tissues can be extracted and the stage of cancer or other disease could be classified. Various imaging technologies of X-ray radiography, magnetic resonance imaging, ultra-sonography, positron emission tomography; are used in medical imaging. Measurement and recording techniques used in medical imaging helps in acquiring the images such as electroencephalography (EEG), magnetoencephalography (MEG), electrocardiography (ECG), and others. For clinical applications medical imaging is generally used by radiologist or medical practitioner engaged in interpreting the images for diagnosis purpose. There are various medical modalities used in medical image processing. The radiologist is usually responsible for acquiring medical images of diagnostic quality, although some radiological interventions are performed by radiologists. Medical image processing also provides an evaluation of anatomy and functional assessment. Many of the techniques developed for medical imaging also have scientific and industrial applications such as automated diagnosis of disease, clinical analysis etc. Medical imaging is considered as the set of techniques that noninvasively produce images of the internal aspect of the body. Sometimes it can also be seen as the solution of mathematical inverse problems. Medical imaging helps in classifying different tissue types such as bone, muscle and fat. 3.1 Classification of Medical Image There are various types of medical images; few of them are explained here briefly. 31

2 3.1.1 X-rays X-rays are produced as radiation of electromagnetic waves. The images are created with the help of X-rays highlighting internal parts of the body radiation is different body parts because different tissues absorb different amounts of radiation. The amount of calcium present in bones absorbs X-rays more and therefore bones look white; fat and other soft tissues absorb less, and look gray. Air absorbs minimum radiation enhance lungs look black. The most important use of X-rays is in detection broken bones. For example, chest x- rays can spot pneumonia and Mammograms are used for breast cancer. Figure 3.1 shows an example of X-ray image. Fig.3.1: X-ray Image. An X-ray is a quick, painless test which produces images of the structures inside the body, particularly bones. The rays pass through the body, absorbed in different amounts depending on the density of the material they pass through. Sometimes iodine or barium is introduced into body provide greater detail on the X-ray images. X-ray are used to detect the following Fractures and infections in bones and teeth; Arthritis and dental decay; Osteoporosis as a density of bones; Lung Cancer and Bone cancer; 32

3 Breast cancer detection using Mammography which is special type of X-ray image; and Swallowed items can be detected using X-ray Tomography Tomography is the method of medical imaging which produces slice of an object. There are various types of tomography as under: Linear tomography is the basic form of tomography in which the X-ray tube is moved from one point to another. The fulcrum is set to the area of interest; and the points above and below the focal plane are blurred out. Poly tomography is a complex form of tomography. Zonography is a variant of linear tomography, where a limited arc of movement is used. Figure 3.2 shows a sample of Tomography image Fig.3.2: Tomography image Computed Tomography (CT) Computed Tomography (CT) is also referred as Computed Axial Tomography (CAT) which is helical tomography and 2D image of the structures in a thin section of the body is produced. CT scan uses X-rays and has a greater ionizing radiation dose burden than projection radiography. Generally, CT is based on the same principles as X-Ray projections but in case of CT the patient is enclosed in a surrounding ring of 33

4 detectors assigned with scintillation detectors. Figure 3.3 shows an example of CT scan image. Fig. 3.3: Computed Tomography image. The salient points about CT scan are summarized as: Computed tomography (CT) is a special type of X-ray imaging using X-ray equipment to produce cross-sectional pictures of the body; CT is also known as CAT (Ccomputerized Axial Tomography) which provides a different form of imaging known as cross-sectional imaging. The origin of the word "tomography" is from the Greek word "tomos" meaning "slice" or "section" and "graphe" meaning "drawing." CT scans are used to detect broken bones, cancers, blood clots, internal bleeding etc; and Positron emission tomography (PET) is used in conjunction with computed tomography and known as PET-CT Radiography Radiography is a very general purpose term used also as X-rays. Basically, two types of radiographic images are in use in medical imaging; projection radiography and fluoroscopy. This imaging modality uses a wide beam of x rays for image acquisition and is the first imaging technique available in modern medicine. 34

5 Fluoroscopy produces real-time images of internal structures of the body in a similar manner to radiography. But this modality employs a constant input of x-rays, at a lower dose rate. Contrast media, such as barium, iodine, and air are used to visualize internal organs as they work. An image receptor converts the radiation into an image after it passes through the area of interest. Projection radiographs, also known as x- rays are used to determine the type and extent of a fracture as well as for detecting pathological changes in the lungs. Figure 3.4 shows a sample of radiographic image. Fig. 3.4: Radiographic image Magnetic Resonance Imaging (MRI) A magnetic resonance imaging modality is mostly used in detection of brain tumour detection. An MRI instrument also known as MRI scanner or nuclear magnetic resonance (NMR) imaging scanner is employed and powerful magnets polarise and excite hydrogen nuclei in water molecules in human tissue which produces a detectable signal which is spatially encoded, resulting in images of the body. A RF (radio frequency) pulse is emitted which binds to hydrogen and the instrument sends the pulse to the area of the body to be examined. The pulse makes the protons in that area absorb the energy needed to make them spin in a different direction. Salient features of MRI modality are reported as under: 35

6 Similar to CT, MRI also creates a two dimensional image of a thin "slice" of the body; MRI is considered as a topographic imaging method; MRI instruments can produce images in the form of 3D blocks, which may be considered a generalisation of the single-slice; CT and MRI are sensitive to different tissue properties and the appearance of the images obtained with the two techniques differ; Any nucleus with a net nuclear spin can be used, the proton of the hydrogen atom remains the most widely used, especially in the clinical setting; Magnetic resonance imaging (MRI) uses a large magnet and radio waves to look at organs and structures inside human body; Physicians use MRI scans to diagnose a variety of disease conditions, from torn ligaments to tumors; MRIs are very useful for examining the brain and spinal cord; problems; MRI scan is painless and the machine makes a lot of noise; and Physician inquires while scanning; if the patient is pregnant; have pieces of metal in the body; and have metal or electronic devices in your body, such as a cardiac pacemaker or a metal artificial joint. Figure 3.5 shows a sample of MRI image. Fig. 3.5: MRI image. 36

7 3.1.6 Ultrasound Images Medical imaging uses high frequency broadband sound waves in the Megahertz (MHz) range which are reflected by tissue to varying degrees to produce medical images. This modality is very commonly used in imaging the foetus in pregnant women. Other important applications of ultrasound images are in imaging of abdominal organs, heart, breast, muscles, arteries and veins. Salient features of this modality are given under: This provides less anatomical details as compared to that of CT or MRI but has several advantages such as it provides monitoring of moving structures in body; it does not ionizing radiation etc. Ultrasound is used as an important tool for capturing raw data that could be used in tissue characterization; This modality is very user friendly; The ultrasound images are digitally acquired and analyzed by the radiologists; The foetus status could be determined and the age of the foetus can also be determined with the help of ultrasound images; The noise present in the images could create problems in determining the status of the foetus; Ultrasound imaging is also used in detection of abnormalities in pancreas; Ultrasound scanners can be taken to critically ill patients in intensive care units without any risks while moving the patient; Doppler capabilities of the scanners allow the blood flow in arteries and veins to be assessed; Diagnostic ultrasound imaging is also called as sonography which uses highfrequency sound waves to produce images of structures within human body; and Ultrasound may be used for several purposes such as, assessment of foetus, diagnosis of gallbladder disease, evaluation of breast lump, cancer detection etc. Figure 3.6 shows a sample of ultrasound image. 37

8 Fig. 3.6: Ultrasound image Thermographic Images This is used in breast cancer detection and imaging of breast images. This modality is of three types: tele-thermography, contact thermography and dynamic angio-thermography. Few salient features of the thermographic imaging are: The modality is basically infrared imaging technique; This works on the concept of metabolic activity and vascular circulation in both pre-cancerous tissue and the surrounding area; Cancerous tumors need more nutrients and this is met by increasing circulation to their cells by holding open existing blood vessels and opening dormant vessels. This can be seen in thermograms; Tele-thermography and contact thermography result in increase in regional surface temperatures of the breast; Thermography is considered as an accurate means of identifying breast tumours; Warnings are issued against thermography in few countries; Dynamic angiothermography exploits thermal imaging; This imaging can be used in combination with other techniques for diagnosis of breast cancer; and The method is a low cost as compared with other techniques. 38

9 Figure 3.7 shows a sample of thermographic image. Fig. 3.7: Thermographic image PET Scan Images Positron emission tomography (PET) scan is an imaging method which utilizes a radioactive substance known as a tracer to search for disease in the body. A PET scan highlights the working of organs and tissues. The salient features of PET imaging are: This is different from MRI and CT scan imaging, which actually shows the structure of and blood flow; It can be seen in Fig. 3.8 that various functions can be explicitly seen as different structures; PET is very useful modality in detection of anatomy of various structures of the body; PET is also used in combination with CT and MRI; and referred as PET-CT and PET-MRI respectively; PET scan is performed for capturing brain, breast, heart and lung in the body; PET is a nuclear medicine used as functional imaging technique; and PET produces a 3D image of functional processes in the body. Figure 3.8 shows an example of PET scan image. 39

10 Fig. 3.8: PET scan image. 3.2 Problems with Medical Images As discussed in previous Chapter, there are several factors affecting the accuracy of CAD based system of medical image analysis and interpretation. One of the most important challenges is noise present in medical images. There are numerous techniques for medical imaging such as such as CT scan, ultrasound, digital radiography, magnetic resonance imaging (MRI), spectroscopy and so on. The imaging techniques have been being used as revolutionary methods in diagnostic radiology. The images which are used by physicians in their analysis are prone to suffer with noise. Under this condition, the detection accuracy might suffer. CAD suggests using appropriate algorithms for noise removal. Generally, image brightness is desired to be uniform except where it changes to form an image. The variation in brightness or contrast value is usually random and has no particular pattern. This can reduce image quality and is especially significant when the objects being imaged are small and having low contrast. This random variation in image brightness or low contrast is known as noise. 40

11 3.2.1 Noise Noise is an undesired signal which corrupts the digital images. The noise may be contributed by various sources such as errors in the image acquisition process that might result in pixel values not reflecting the true nature of the scene. During acquisition, transmission, storage and retrieval processes, the noise may be mixed with original image. An image which is being sent electronically from one place to another place is contaminated by noise sources. Noise can be caused in images by random fluctuations in the image signal. The noise signal present in medical images poses a great challenge in automatic medical image analysis. Some of the important points related to noise in medical images are: All medical images contain some type of noise; The noise may be of grainy, textured, or snowy appearance; Noise in image may contributed by several sources: No imaging method is free of noise; Image de-noising becomes essential step in all CAD systems; Nuclear images are more noisy; Noise creates more difficulty in MRI, CT, and ultrasound imaging as compared to other imaging modalities; Radiography produces images with the minimum noise; Noise also reduces the visibility of certain features within the image; The loss of visibility is especially significant for low-contrast images; The main aim of the image processing is to extract clear information from the images corrupted by noise; and Noise removal is called filtering or denoising. Huan et al. (2010) discussed three important types of noise signals: impulse noise, multiplicative noise and additive noise. Noise in digital images is found to be additive in nature with uniform power in the entire bandwidth with Gaussian probability distribution. This type of noise is called as additive white Gaussian noise (AWGN). This noise is multiplicative noise and an unwanted random signal gets multiplied into some relevant signal during acquisition, transmission, or other image processing. An important example of noise in medical images is the speckle noise that could be commonly observed in radar imaging modality. Few examples are shadows due to undulations on the surface of the imaged objects, shadows cast by complex objects 41

12 like foliage, dark spots caused by dust in the lens or image sensor, and variations in the gain of individual elements of the image sensor. Multiplicative noise appears in various image processing applications, such as synthetic aperture radar, ultrasound imaging, particle emission-computed tomography and positron emission tomography. Therefore, removal of multiplicative noise is very essential in imaging systems and image processing applications. Since the speckle noise is mostly found in medical images, this is briefly discussed here Speckle Noise Speckle is a basically a granular 'noise' which exists in the images inherently. This noise degrades the quality of the active radar and synthetic aperture radar (SAR) images. Salient features of speckle are: Speckle noise in conventional radar results from random fluctuations in the return signal from an object that is no bigger than a single image-processing element; The speckle increases the mean grey level of a local area; This noise in SAR is very serious and causes difficulties for image interpretation; Speckle is caused by coherent processing of backscattered signals from multiple distributed targets; This is also caused by signals from elementary scatterers and the gravitycapillary ripples; Several different methods are available in literature for removing speckle noise; The methods are based on various mathematical models; Adaptive non-adaptive filters are used to eliminate the noise; Adaptive speckle filtering is better at preserving edges and detail in hightexture areas; Non-adaptive filtering is simpler to implement and requires less computational time; There are two types of non-adaptive speckle filtering; There are many forms of adaptive speckle filtering, including Frost filter Speckle noise in SAR is a multiplicative noise where the noise is present in direct proportion to the local grey level; and 42

13 The speckle can also represent some useful information in some cases, for example; laser speckle where the changes of the speckle pattern is a measurement of the surface's activity. Figure 3.9 shows speckle noise as granular noise in a digital image. Fig. 3.9: Speckle noise as granular noise in a digital image. 3.3 Image De-noising Methods An extensive literature survey suggests that there are numerous research contributions as discussed in Chapter 2. Few of prominent methods contributing towards enhanced robustness are discussed here. Gonzalez et al. (2009) suggested Median Filtering Technique that could successfully remove Impulse noise from the distorted image. However, the filtered image suffers the blurring effect. Here, each pixel is considered to calculate the median and every pixel is replaced by that calculated median. So affected pixels are considered to calculate the median and unaffected pixels are also replaced by this calculated median. Achieving higher PSNR or better quality image could be further improved. Bo et al. (2007) suggested an adaptive threshold median filter (ATMF) as a combination of the adaptive median filter (AMF) and two dynamic thresholds. This work gives a balance between the removal of multiple impulse noise and the quality of image. 43

14 Pandey et al. (2008) presented an Improved Switching Median filter as effective method to remove salt-and-pepper noise in images, but for uniformly distributed impulse noise, the method does not perform well. Chen et al. (2001) proposed an Adaptive Center-Weighted Median Filter (ACWM) using a novel adaptive operator for estimating the differences between the current pixel and the outputs of centerweighted median (CWM) filters with varied center weights. Ko et al. (1991) implemented a Multi-State Median Filter (MSM) as generalized framework of median based switching schemes. This method is also called multi-state median (MSM) filter. The output of the MSM filter is adaptively switched among a group of center weighted median (CWM) filters having different center weights. The method is equivalent to an adaptive CWM filter with a space varying center weight which is dependent on local signal statistics. Chen et al. (2001) proposed a Tri-State Median Filter (TSM) for preserving image details while effectively suppressing impulse noise by using the standard median(sm) filter and the center weighted median (CWM) filter into a noise detection framework. This was also determined whether a pixel is corrupted, before applying filtering unconditionally. Noise detection is realized by an impulse detector, which takes the outputs from the SM and CWM filters and compares them with the origin or center pixel value in order to make a tri-state decision. Ma et al.(1999) studied an advanced Impulse Detection Based on Pixel-Wise MAD (PWMAD) which has a robust estimator of variance, MAD (median of the absolute deviations from the median), is modified and used to efficiently separate noisy pixels from the image details. Senk et al. (2004) presented Signal-Dependent Rank Order Mean (SDROM) Filter using a new framework for removing impulse noise from images, in which the nature of the filtering operation is conditioned on a state variable defined as the output of a classifier that operates on the differences between the input pixel and the remaining rank-ordered pixels in a sliding window. The method can effectively restore images corrupted with Gaussian noise and mixed Gaussian and impulse noise. Mitra et al. (1996) proposed a directional Weighted Median Filter (DWM) used for removal of random-valued impulse noise. This filter uses a new impulse detector, which is based on the differences between the current pixel and its neighbours aligned with four main directions. 44

15 Soft computing based methods are recently used for image de-noising using various tools such as neural network and fuzzy logic concept. Now, hybrid approaches are also developed for removing the noise from medical images Fuzzy Based Methods Fuzzy based methods mainly employ the concept of fuzzy logic and membership functions to reduce the amount of uncertainty. Fuzzy set theory and fuzzy logic was discussed by Kerre et al. (1998) and the concept of fuzzy logic proved to be powerful tool to represent and process human knowledge represented as fuzzy if-then rules. Fuzzy image processing suggested by Tizhoosh et al. (1997) has three main stages namely, image fuzzification, modification of membership values and image defuzzification. The main ability of fuzzy image processing lies in modification of membership values. After the image data is transformed from input plane to the membership plane (fuzzification), appropriate fuzzy techniques modify the membership values. This can be a fuzzy clustering, a fuzzy rule-based approach, a fuzzy integration approach, etc. Following are main advantages of fuzzy based approaches: The complexity of the method is reduced; Execution time is reduced; Noise reduction performance is improved; and Amount of uncertainty is reduced Neural Network Based Methods Artificial Neural Networks (ANNs) are used generally for classification and feature extractions tasks. ANNs are also used as a new model to the image denoising problems. Yu et al. (2002) proposed an ANN as a mathematical or computational model that attempts to simulate the functional features of biological neural networks. The networks consist of an interconnected group of artificial neurons to process input information. ANN can be an adaptive system that changes its construction based on external or internal information that flows through the network during the learning phase Hybrid Methods Improved performance could be achieved by combining neural network as well as fuzzy logic algorithms. The method becomes hybrid method. Pusphavalli et al. (2013) 45

16 stated that there has been a growing interest in the applications of soft computing techniques, such as neural networks and fuzzy systems, to the problems in digital signal processing. Salient points related to soft computing tool are: Neural networks are low level computational structures that perform well when dealing with raw data; Fuzzy logic deals with reasoning on a higher level, using linguistic information acquired from domain experts; Fuzzy systems lack the ability to learn and cannot adjust themselves to a new environment; Hybrid neuro-fuzzy systems can combine the parallel computation and learning abilities of neural networks with the human like knowledge representation and explanation abilities of fuzzy systems; Neural networks become more transparent, while fuzzy systems become capable of learning and Neuro-Fuzzy (NF) systems offer the ability of neural networks to learn from examples and the capability of fuzzy systems to model the uncertainty; Hybrid systems may be utilized to design efficient signal and image processing operators with much less distortion than the conventional operators; A Neuro-Fuzzy System is a flexible system trained by heuristic learning techniques derived from neural networks can be viewed as a 3-layer neural network with fuzzy weights and special activation functions is always interpretable as a fuzzy system uses constraint learning procedures is a function approximation (classifier, controller) (Sun et al. (1994); Wang et al. (1999)); This filter can efficiently eliminate impulse noise; Filter performance deteriorates the image quality for higher level impulse noise; and Preserving edges and fine details of images at higher level impulse noise could also be achieved using appropriate hybrid methods. 46

17 3.4 Summary This chapter discusses about medical image processing, various modalities, different types of medical images, various types of noise signals and noise removal methods. The chapter can be summarized as: Medical image processing is application area of digital image processing in which the signal is medical image. The technique or process works as creating visual representations of the interior of a body for clinical analysis and medical intervention. Medical imaging helps in retrieval of internal structures of human body, skin and bones for diagnosis purpose and treatment of associated disease. X-rays are produced as radiation of electromagnetic waves. The images are created with the help of X-rays highlighting internal parts of the body radiation is different body parts because different tissues absorb different amounts of radiation. Tomography is the method of medical imaging which produces slice of an object. Computed Tomography (CT) is also referred as Computed Axial Tomography (CAT) which is helical tomography and 2D image of the structures in a thin section of the body is produced. CT scan uses X-rays and has a greater ionizing radiation dose burden than projection radiography. Radiography is a very general purpose term used also as X-rays. Basically, two types of radiographic images are in use in medical imaging; projection radiography and fluoroscopy. A magnetic resonance imaging modality is mostly used in detection of brain tumour detection. An MRI instrument also known as MRI scanner or nuclear magnetic resonance (NMR) imaging scanner is employed and powerful magnets polarise and excite hydrogen nuclei in water molecules in human tissue which produces a detectable signal which is spatially encoded, resulting in images of the body. Medical imaging uses high frequency broadband sound waves in the Megahertz (MHz) range which are reflected by tissue to varying degrees to produce medical images. 47

18 Positron emission tomography (PET) scan is an imaging method which utilizes a radioactive substance known as a tracer to search for disease in the body. A PET scan highlights the working of organs and tissues. The images which are used by physicians in their analysis are prone to suffer with noise. Under this condition, the detection accuracy might suffer. CAD suggests using appropriate algorithms for noise removal. Noise is an undesired signal which corrupts the digital images. The noise may be contributed by various sources such as errors in the image acquisition process that might result in pixel values not reflecting the true nature of the scene. During acquisition, transmission, storage and retrieval processes, the noise may be mixed with original image. Speckle is a basically a granular 'noise' which exists in the images inherently. This noise degrades the quality of the active radar and synthetic aperture radar (SAR) images. Fuzzy based methods mainly employ the concept of fuzzy logic and membership functions to reduce the amount of uncertainty. Artificial Neural Networks (ANNs) are used generally for classification and feature extractions tasks. ANNs are also used as a new model to the image denoising problems. Hybrid neuro-fuzzy systems can combine the parallel computation and learning abilities of neural networks with the human like knowledge representation and explanation abilities of fuzzy systems; 48