Vanderbilt University

Transcription

1 Vanderbilt University Department of Biomedical Engineering Using Fluorescence X-ray for Non-invasive Biopsy Team Members: Savannah Gill Kelvin Lin Mike McHugh Trey Reece Derric Williams Advisor: Frank E. Carroll, M.D. Date of Submission: April 24, 2007

2 Abstract Traditional detection of breast cancer includes a mammography and a needle biopsy. As an alternative to this invasive and painful needle biopsy, we explored a new technique of X-ray imaging relying on the angles at which the tissues diffract the X-rays. The goal of our project was to develop a way to differentiate between normal and cancerous breast tissue by use of small angle X-ray scattering (SAXS) generated from a collimated monochromatic X-ray developed by MXISystems. Using a polychromatic X-ray machine featuring a molybdenum tube, X-rays were fired at eight biological samples of pig and mouse tissue, both malignant and benign. The beam was collimated with a 0.2 cm aperture in an aluminum-lead collimator. Another lead collimator with an aperture of 0.07 cm was also used. The SAXS of the biological samples was recorded with a MAR345 image plate detector and the data was analyzed using ImageJ and Microsoft Excel. We found that the SAXS emission spectra of different samples, both cancerous and normal, exhibit qualitative differences that can be effectively plotted and analyzed. We were, however, unable to find specific key angles or significant differences between scattered angles of cancerous and normal tissue. Our design was originally intended for use with a monochromatic X-ray beam. Unfortunately, the monochromatic X-ray beam was not sufficiently repaired in time for us to conduct many experiments and gather conclusive data.

3 Introduction More than 180,000 women are diagnosed with breast cancer in the US every year 1. It is the second leading life-taking cancer for women. Aside from surgical removal of the tumor, there is currently no cure for breast cancer. Therefore, the greatest weapon to defeat this cancer is early detection. With early detection, the chances of the tumor metastasizing are less and the chances of a complete removal of the tumor are greater. Currently, the primary diagnostic phases for breast cancer detection are first physical examinations, then mammography, and finally biopsy 1. Physical examinations can only detect palpable tumors, and by that point the tumors are large and dangerous. Mammography has been the primary breast cancer diagnostic technology used for many years and has not been successfully exceeded by any other yet. It consists of using low-dose X-rays to create images of the breast on film 1. Different parts of the breast will absorb different amounts of X-rays and will show up on the film as black, gray or white depending on what the composition is 3. Especially with small tumors, distinguishing cancerous tissue from normal tissue by the different shades of black, gray, and white can be very difficult and consequently there are many cases of false positives and negatives. Biopsy is the removal of a lump or tissue sample for analysis. Due to the chances of getting diagnosed with breast cancer from a mammogram are as high as 10%, there are many women who must undergo unnecessary biopsies 1. About ¾ of breast lesions that are biopsied, as a result of questionable findings from a mammogram, turn out to be benign. The different forms of biopsy consist of fine needle aspiration, core needle, vacuum assisted, large core surgical, and open surgical. Each varies in the level of patient

4 discomfort, resultant scarring, and successful diagnosis. Although usually very definitive in the findings, biopsies can be painful to the patient and cause scarring or deformation of the breast, which can be difficult for further mammogram readings. The goal of our project is to develop a way to differentiate between normal and cancerous breast tissue by the use of a collimated X-ray beam and small angle X-ray scattering. With its success and further development, our project could eventually offer women another option of breast cancer detection with the accuracy of biopsies, but eliminating the invasiveness, needles, anesthesia, and pain associated with them. The theory behind small angle X-ray scattering involves coherent scattering of photons 2. When a sample is hit with an X-ray beam, photons are released at different angles. These elastic scatterings of photons are recorded at very low angles and can be detected by the same equipment used to detect the X-ray. Depending on the molecular makeup of the medium, a unique diffraction pattern will result leading to the characterization of the sample through diffraction patterns 4. In order to create clearer patterns, it is necessary to use a collimator to both create a smaller beam, and better isolate the scattering from the direct beam. Traditional X-rays used have been polychromatic, which consist of photons of multiple energies. Current technology allows the use of monochromatic X-rays, which consist of photons at a single energy. As low-angle scattering is highly dependent upon the energy levels used, additional energy levels present will result in ambiguous scattering angles. Therefore, when a monochromatic X-ray is used, a given molecule will scatter the photons at one angle giving a much more defined angle for analysis.

5 Methods In order to work with the X-ray equipment in the W.M. Keck Free Electron Laser Center, the group had to take a course on radiation safety training. Once the course was completed, each group member received the required radiation safety badge provided by the university. The first task was to design and construct a collimator in order to properly reduce the spot size of the X-ray Io I beam. Using the Beer-Lambert Law, which is seen in Equation 1 and represented in Figure 1, three metals, aluminum, lead and l Figure 1. According the Beer-Lambert Law, an X-ray with energy Io incident on a material will be absorbed based on the material's attenuation coefficient. iron, were tested based on their attenuation coefficients, µ, and thickness, l, to determine the amount of X-rays transmitted through the metal. I I o = µ l Eq. 1 Among the three metals tested, lead blocked the most X-rays with a material thickness of just 1 cm. However, since lead is slightly hazardous to handle and extremely Figure 2. Aluminum-lead sandwich design for the original collimator. malleable at such a small thickness, an aluminum-lead sandwich design, seen in Figure 2, was used for the collimator. This provided a sturdy backing to prevent the lead from experiencing deformation and also made handling the collimator easier by having the non-hazardous aluminum on the outside. Each lead sheet was 1.5 mm

6 in thickness, providing a total lead thickness of 3 mm. This was more than enough to successfully block out almost the entire X-ray beam at 30 kev. Each aluminum sheet was also 1.5 mm in thickness. The four sheets had a length of 24 cm and a height of 10 cm. Six holes were drilled into each of the four metal sheets and they were then riveted together. Five apertures of varying sizes were drilled into the collimator to determine which X-ray spot size elicited on a sample would provide the best scattering results. The apertures needed to be spaced appropriately apart to prevent the X-ray beam incident on the collimator from passing through multiple holes. The design Figure 3. Dimensions of the aluminum-lead collimator. dimensions of the collimator are displayed in Figure 3. The apertures were drilled into the collimator using a press drill. The next step was to design and construct a beam stop which would block out the central beam of the X-rays transmitted θ Distance to Detector Scattering Radius through the sample. Based on the group's research, it was estimated that the SAXS of human tissue would exist between the angles of Figure 4. Basic geometry used to determine the radius of X-ray scattering. Knowing the distance from the sample to the detector, we were able

7 to calculate the final radius of X-ray scattering using basic geometry, as seen in Figure 4. A tangent equation, Equation 2, was applied given both the maximum and minimum scattering angles, θ, and maximum and minimum distances to the detector, x, in order to find the scattering radius, R s. x*tan(θ) = R s Eq. 2 Having found the maximum and minimum values for radius spot size, we were able to construct a beam stop that would be large enough to sufficiently block out the central transmitted beam and yet small enough to allow the scattered X-rays to pass. A circular beam stop 5 cm in diameter was then constructed out of lead. Using a Kevex X-ray machine featuring a molybdenum tube, X-rays were fired through the collimator onto various samples. The spectrum for the Kevex molybdenum tube is visible in Appendix A, Figure A1. The energy varies over a wide range of values, and holds two large peaks at 17.5 kev and at 19.6 kev, with a local maximum around 25 kev. While the two spikes dominate the results, the extra energy adds noise to the diffraction patterns. As the diffraction angle is directly related to the energy of the incoming X-rays, the range of energies causes the angle to blur over a range of values. The purpose of the monochromatic X-ray developed by MXISystems is to have a tunable X-ray that generates just the peaks without any of the extra noise. The SAXS of the samples was collected and recorded with a MAR345 image plate detector. The experimental set up is displayed in Appendix A, Figure A2. The first samples fired at included pure samples of metal, crystal, numerous breast phantoms and wood. It was determined that the smallest aperture of 0.2 cm provided the best scattering images. However, in order to obtain better results, the collimator was modified to have a second

8 aluminum-lead panel. Another two 1.5 mm thick sheets of lead were riveted to a sheet of aluminum also 1.5 mm in thickness. This second panel was glued to the first with the two panels separated by an inch of wood. Adding a second panel was done to simply help further collimate the X-ray the beam. Eight biological samples of pig and mouse tissue were prepared by the group's advisor, Dr. Frank Carroll. These samples included skin, malignant and benign breast tissue and malignant and benign muscle. X-rays were fired at these biological samples using the smallest aperture in the two-panel collimator. However, difficulties involved in set up and alignment of the collimator led to a new design for a second collimator, which was constructed out of a lead brick 5 cm in thickness with an aperture 0.25 cm in diameter. This collimator proved ineffective as the aperture was too large for the desired scattering and a new collimator was needed. It was determined that the thickness of the 5 cm lead brick was unnecessary, as only a few millimeters were needed. Also, at this thickness the collimator was exceedingly heavy and difficult to move. Thus, the final collimator was designed to be 0.25 inches thick with an aperture 0.18 cm in diameter. Several more trials were run on both the phantoms and the biological samples using this third collimator.. The obtained images of SAXS were transferred to the computer program ImageJ to be analyzed. To plot the emission spectrum, ImageJ was used to subtract the control spectrum of the cuvette from the pattern generated by the sample within the cuvette. ImageJ was then used to plot a radial profile from the center of the pattern to the edge of the detector. The data generated was then transferred to Microsoft Excel as a function of scattered angle, which was easily calculated from the pixel size and the distance of the detector from the sample. A five-pixel convolution was applied to smooth each spectrum

9 and reduce the effects of white noise. The plots from the biological samples were compared to determine any differences between malignant and benign tissue, or key angles that could provide insight about the characteristics of each sample. Results & Discussion Calibration A variety of samples were imaged before any tests were done on biological samples. Aluminum was one of the first samples, since scattering patterns for this and other crystalline materials traditionally generate a unique spray of dots upon irradiation. Rather than the expected dot pattern, though, our setup generated a network of smears. The width of the smear is due to the divergence from the holes in the collimator, and the length is due to the change in diffraction angle due to the range of energies that the X-ray tube generates. The blurs are also darker near the center where the two maxima for the molybdenum X-ray energy peak. Other substances that were imaged were wood, calcite, water, polystyrene, beryllium, and fluorite. Some of these images are shown in Appendix B. Our system was first calibrated for biological tissues using adipose/glandular human breast tissue phantoms designed specifically for breast cancer imaging modalities. These were imaged to ensure that our design was capable of producing SAXS spectra, and are plotted along with their unique radial profile in Appendix C. These phantoms were each designed to represent different ratios of adipose and glandular tissue, and these ratios are observed in their SAXS spectra as the samples transform from 100% adipose to 100% glandular tissue. The transition from adipose to glandular was found to coincide with the disappearance of a characteristic ring at approximately 15.5, which is also clearly shown in each sample s radial profile. Another breast tissue phantom was also

10 imaged that contained a wax insert of unknown composition. The SAXS spectrum for this insert is also featured in Appendix C, and exhibits a far more spectacular multilayered ring pattern. Biological Samples Eight biological samples were imaged using our design, and emission spectra were analyzed for each specimen. A plot of relative intensity versus scattering angle is displayed in Appendix D, Figure D1 for all eight samples. Each sample generated a unique spectrum, the most radical being the non-cancerous adipose tissue taken from an adult pig breast. The fact that each sample generates a different spectrum indicates that each sample contains a different ratio of scattering molecular species, with each species scattering its unique angle at different relative intensities. There are obvious similarities as well, with each sample exhibiting a peak effect at 9.5, 11.5, 18.7, and 20.6, as well as an absorptive trough effect at 3.3, 5.6, 7.5, 13.6 and This is indicative of the fact that all biological tissues share some base characteristic molecular species at similar levels. This experiment demonstrates that the SAXS emission spectra of different samples, both cancerous and normal, exhibit qualitative differences that can be effectively plotted and analyzed. We were, however, unable to find specific key angles or significant differences between scattered angles of cancerous and normal tissue. This shortcoming, we believe, does not stem from flaws in our experimental design, but from insufficient samples to study. We could only image, for example, one sample of cancerous mouse muscle rather than many. To test the reproducibility of our results, the same sample was imaged several times, with each trial plotted as relative intensity versus scattered angle, shown in

11 Appendix D, Figure D2. The scattering signature of the cuvette that held the sample was also subtracted from every pattern. Similarities in each trial are apparent at many angles. Notable angles are 4.5, 5.4, 7, 10.5, 12, 13.8, and 19.4 degrees, though closer inspection reveals additional similarities throughout. Since many trials of the same sample generated the same unique pattern, we can conclude that the process we designed is a sound way to generate SAXS spectra of biological samples for analysis. Safety The largest safety concern is exposure to X-rays generated by the machine used for testing. Before being allowed into the laboratory, we were required to undergo radiation safety training with the staff on campus. In addition to protecting ourselves, there are several safeguards built into the equipment. This includes an interlocking door to the X-ray room, as well as a door that interlocks with the X-ray itself. Additionally, there is a lead cover on the X-ray tube to block any low-energy X-rays generated when the tube is active but not firing. Finally, in order to keep track of the dosage we receive from our time in the laboratory, we are required to keep X-ray sensitive badges on our bodies at all times, and turn these badges in for measurement once a month. If these badges show a dangerously high amount of X-rays we would be forced to retire from X- ray related activities for a period of time to wait for the biological effects to subside. An additional danger existing in the laboratory came from the lasers used to generate the monochromatic X-rays. As the voltages used for these lasers are extremely high, looking into the lasers for any length of time would be enough to cause blindness. In order to protect against this, all of the lasers are covered by heavy black cloth, and separated from the rest of the equipment by an interlocking door.

12 The only other risk in the lab comes from the material used to block most of the X-rays: Lead. As lead is a toxic material, whenever prolonged exposure to the lead was necessary, we wore gloves for our safety, and our hands were washed after contact. Economic Considerations As our project was conducted with the purpose of testing a technique for possible use, there is no "product" per say. Both the collimator and the beam stop are constructed of extremely cheap materials, costing less than a few dollars. However, in the future, a medical implementation of this technique would require a much more extravagant setup, including many costs for registration and licensing, as well as testing for safety. Patient risk would be minimized, most likely at or below the risk of normal mammogram exposure. Traditional mammography exposes the breast to approximately 13.0 mr, over 1.14 seconds at 25 kvp 5. Our setup confers X-rays through an area 0.2 inches in diameter, at 30kvp for seconds. Despite the increased exposure time, the difference in area of effect is overwhelming. As this research was specifically begun for the purpose of testing a new use for the monochromatic X-ray created by MXISystems, any clinics wishing to use this technique would first need to invest $3 million for one of the new monochromatic X-rays that MXISystems is producing. This purchase could be a replacement of the clinic's original X-ray tube, as the monochromatic can do mammograms and all of the other tasks of the original X-ray. Our technique would be a side-application for this device, as a substitute for needle biopsy. Until this product, the only way to generate monochromatic X-rays was in a synchrotron. As a result, monochromatic X-ray beams have only been

13 used in a few locations, and purely for research. Our experiment attempts the first clinical use for this product, though potentially many could follow. Conclusions Our experimental design and method has successfully generated SAXS emission spectra from biological samples. We have shown that scattering spectra from different biological samples exhibit strong similarities and differences, and that our design generates reproducible results, confirming the possibility of using this technique for cancer detection. Our design was originally intended for use with a monochromatic X-ray beam, which would have substantially reduced noise and sharpened the scattering patterns, which would allow us to make more conclusive results. Unfortunately, the monochromatic X-ray beam was not sufficiently repaired in time for us to conduct many experiments and gather conclusive data.

14 References 1. Nass, S., Henderson, C., Lashof, J. Mammography and Beyond: Developing Technologies for the Early Detection of Breast Cancer. National Academy Press. Washington, DC Changizi, V., Oghabian, M., Speller, R., Sarkar, S., Kheradmand, A. "Application of Small Angle X-ray Scattering for Differentiation between Normal and Cancerous Breast Tissue." International Journal of Medicinal Sciences pp "Mammography." Radiology Info. 15 Aug Apr < 4. L., Dunn. "Small Angle X-Ray Scattering." 7 Apr Apr < 5. Lavoy, Thomas R. "Radiographic Techniques in Screen-Film Mammography." Journal of Applied Clinical Medical Physics 3 (2002): E.F. Donnelly, R.R. Price, and D.R. Pickens. Dual focal-spot imaging for phase extraction in phase-contrast radiography. Med. Phys. 30, (2003). 7. F.E. Carroll, J.W. Waters, et al. Attenuation of Monochromatic X-Rays by Normal and Abnormal Breast Tissues. Investigative Radiology. 29, (1994). 8. Elshemey, Elsayed, Lakkani. "Characteristics of low-angle x-ray scattering from some biological samples" Phys. Med. Biol. 44, (1999). 9. Fernandez, Keyrilanen, Serimaa, et al. "Small-angle x-ray scattering studies of human breast tissue samples." Phys. Med. Biol. 47, (2002). 10. Kidane, Speller, Royle, and Hanby. "X-ray scatter signatures for normal and neoplastic breast tissues." Phys. Med. Biol. 44, (1999). 11. E. F. Donnelly and R. R. Price. Quantification of the effect of kvp on edgeenhancement index in phase-contrast radiography. Med. Phys. 29, (2002). 12. "NIST Scientific and Technical Databases." National Institute of Standards & Technology. 19 June Jan < 13. Cullity, R D., and S R. Stock. Elements of X-Ray Diffraction. 3rd ed. Prentice Hall, 2001.

15 Appendix A Experimental Set Up Figure A1. Intensity spectrum of the Kevex X-ray cm X-Ray Collimator Sample mar345 Detector Translational Stage 24.1 cm 38.1 cm 100 cm Figure A2. Top-down view of experimental set up with dimensions.

16 A Kevex Molybdenum X-ray Tube B Polychromatic X-ray beam C Lead Collimator D Biological Sample E Scattered X-rays F Beamstop G MAR345 X-ray detector G F D E C B A Figure A3. 3-Dimensional view of the experimental set up. G C B D θ F B Polychromatic X-ray beam C Lead Collimator D Biological Sample E Scattered X-rays F Beamstop G MAR345 X-ray detector E Figure A4. Side view of the experimental set up.

17 Appendix B X-ray Emission Spectra Figure B1. Cuvette 30 second exposure 0.4 cm aperture Figure B cm aluminum block 10 second exposure 0.4 cm aperture Figure B3. Aluminum foil 30 second exposure 0.4 cm aperture Figure B4. Calcite 0.4 cm aperture

18 Figure B5. Pyrite 0.4 cm aperture Figure B cm Aluminum block 0.4 cm aperture Figure B7. Water in cuvette 0.4 cm aperture Figure B8. Phantom bolus 0.4 cm aperture

19 Figure B9. Phantom fat 0.4 cm aperture Figure B10. Vertical wooden 0.2 cm aperture Figure B11. Horizontal wooden block 0.2 cm aperture Figure B12. Horizontal phone book 0.2 cm aperture

20 Figure B inch sheet of rubber 0.2 cm aperture Figure B14. Wax phantom 0.18 cm aperture Figure B15. Mouse - small cancerous breast tissue (sample 1) 0.18cm aperture Figure B16. Mouse - large cancerous breast tissue (sample 2) 0.18cm aperture

21 Figure B17. Mouse normal muscle tissue (sample 3) 0.18cm aperture Figure B18. Mouse - cancerous muscle tissue (sample 4) 0.18cm aperture Figure B19. Pig skin (sample 5) 0.18cm aperture Figure B20. Pig fat tissue (sample 6) 0.18cm aperture

22 Figure B21. Pig normal muscle tissue (sample 7) 0.18cm aperture Figure B22. Pig normal muscle breast (sample 8) 0.18cm aperture

23 Appendix C Phantom Appendix Adipose/Glandular 0/100

24 Adipose/Glandular 30/70

25 Adipose/Glandular 50/50

26 Adipose/Glandular 70/30

27 Adipose/Glandular 100/0

28 Wax Adipose Phantom

29 Appendix D Biological Sample Analysis I/Io Radial Profile of Several Specimens (Relative Intensity vs scattered angle) Scattered Angle (θ) Breast 1 (Cancerous) Breast 2 (Cancerous) Muscle (Normal) Muscle (Cancerous) Skin (Normal) Adipose (Normal) Muscle (Normal) Breast (Normal) Figure D1. Relative intensity vs. scattered angle of eight biological samples. 1 Cancerous Muscle (Sample 4) Relative Intensity vs Scattered Angle I/Io Trial 1 Trial 2 Trial 3 Trial 4 Average Scattered Angle (θ) Figure D2. Relative intensity vs. scattered angle of cancerous mouse muscle.