Investigations into high resolution imaging and contouring for diagnostic bio-applications

• Main Author: Valiyambath, Krishnan Mohankumar
• Other Authors: Associate Professor V. M. Murukeshan
• Format: Theses and Dissertations
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering::Mechanical engineering
• Online Access: http://hdl.handle.net/10356/66284
• Institution: Nanyang Technological University

Cancer is a leading cause of deaths in the world, with lung, breast and colorectal cancers being the top killers. As early detection is important to prevent deaths due to cancer, current emphasis is on periodic screening for persons over the age of 50. Depending on the organ being assessed, a variety of diagnostics tools are presently available to detect abnormal state of the tissue, with each technique having its own set of advantages and disadvantages. Of these, currently, only endoscopy provides a possibility of abnormality detection based on direct tissue imaging and visual differentiation in body cavities such as colon. However, the effectiveness of endoscopy is limited in that polyps are missed as they fail to catch the attention of the endoscopist during the visual examination. They are missed out in blind biopsies or are not well delineated as abnormalities, especially as in the case of flat dysplasia. If abnormal regions are well identified, targeted biopsies can be carried out minimising the pain and risk of tissue perforations to the patient. Research is in progress around the world for finding methods which are minimally invasive and which can provide accurate results with minimum patient discomfort. The hallmarks of cancer are changes in the tissue’s biochemical makeup, biological activity level, vasculature, blood perfusion and cellular structure. In this context, this research thesis focuses on coming up with such methodologies using optical concepts and imaging modalities for diagnostic medical imaging in a probe based scheme. Complementary optical methodologies, which can enable easy identification and diagnosis of potential abnormal growth sites that represent cancerous state of tissue, were investigated in this research, as it is envisioned that information from one modality can be used to strengthen judgment from another independent modality. For the modalities in whole field imaging mode, a target size of 5 mm and below was set as detection capability for the abnormalities for early diagnosis, as polyp sizes up to 5 mm were deemed as low risk and are only marked for observation in subsequent screening sessions. An enlargement of the size of cell nucleus from about 5-10 pm to about 20 pm being an indication of pre-cancerous state, a resolution capability of less than 5 pm was set as a target for high resolution imaging mode. A novel RGB grayscale value based spectral shift approach was successfully developed for easy identification of abnormality based on biochemical signature differences that can be identified optically, namely fluorescence. A CCD camera based image analysis scheme was conceptualised and configured using spectral shift calculations. The Euclidian distances for the change in Red, Green and Blue gray values between the illuminating wavelengths and the points in the image were used to reconstruct a pseudo-image which shows the differences between normal and abnormal areas in the tissues. The proposed algorithm was validated with phantom tissues as well as real mouse colon tissues and mouse with tumour to distinguish abnormal regions, with abnormality sizes ranging from 2 mm to about 12 mm. The proposed algorithm and developed methodology apply pixel level processing, and hence, processing of both whole field and high resolution images can be carried out.Contouring of abnormality based on bioactivity differences that can be identified optically, namely dynamic speckle based methodology, was the second target area of the research. A novel thermal perturbation based approach was proposed and researched to enable identification and contouring of abnormalities by way of differences in speckle pattern changes arising out of tissue mass and perfusion differences. Thermal perturbation of the target tissue with infrared beam was used to effect the required change in the bio-speckle pattern. Subsequent contouring of the abnormal region in the tissue could be delineated due to the speckle pattem differences that evolved over. It was found, using phantom tissue as test samples, that a time lapse of about 1 second between imaging frames captured with a CCD camera could enable identification and contouring of sub-surface abnormalities. An image reconstruction scheme using variation of intensity at corresponding pixel locations was also investigated. The effects of perfusion and simulated mucus layer were investigated. Experimental verification was carried out on silicone phantoms and to a limited extent, with animal model, with abnormality sizes ranging from about 5 mm to 8 mm. A fiber based high resolution imaging methodology was investigated that is capable of producing microscopic images of the targeted tissues. A quasi- confocal imaging approach with an imaging fiber bundle was developed to image and contour microscale objects. Non-scanning (fixed frame) as well as scanning (lD as well as 2D) based approaches were investigated. For the nonscan based approach, with axial and lateral resolutions of 15.98 ± 0.94 pm and 3.47 pm, respectively, the system capability was demonstrated with glassbubbles, polystyrene beads, cells and micro-fluidic flow channels. The scan based methodology developed, with the probe in proximity with the target tissue, could overcome loss of visual information due to the fiberlet interspaces and achieved a significant improvement in lateral resolution to 2.19 pm with the same optical elements as with the non-scan based scheme. With the fiber bundle enabling multiple scan points, thus saving scan time, and scanning steps designed to match the imaging pixel size for best resolution, a scheme to reconstruct high resolution images from the whole-field scan, at any region of interest was worked out. It was found that the central pixel in each of the fiberlet image are the optimal scan points as it gave the best contrast and the least offset in the scan image compared with other pixels within the fiberlet. The capability of the system was demonstrated using micron-sized fluorescent polystyrene beads and cells mounted on a microscope slide. Besides providing imaging capability, the imaging fibre can double up as a light conduit to illuminate the target tissue, which would be in synergy with the first two modalities, namely, the RGB and the dynamic speckle analyses. The research carried out in this thesis involved exploring and laying the foundations for a synergistic combination of optical methods which could potentially be integrated in future into a single probe for endoscopic applications. It is capable of identifying early stage cancerous regions of less than 5 mm in size from a distance and also micro-level imaging of structures less than 5 pm at proximity of the targeted area. As a future work direction, an actual probe system could be engineered integrating a micro-CCD camera and an imaging fiber bundle by applying the proposed multiple optical bubbles, polystyrene beads, cells and micro-fluidic flow channels. The scan based methodology developed, with the probe in proximity with the target tissue, could overcome loss of visual information due to the fiberlet interspaces and achieved a significant improvement in lateral resolution to 2.19 pm with the same optical elements as with the non-scan based scheme. With the fiber bundle enabling multiple scan points, thus saving scan time, and scanning steps designed to match the imaging pixel size for best resolution, a scheme to reconstruct high resolution images from the whole-field scan, at any region of interest was worked out. It was found that the central pixel in each of the fiberlet image are the optimal scan points as it gave the best contrast and the least offset in the scan image compared with other pixels within the fiberlet. The capability of the system was demonstrated using micron-sized fluorescent polystyrene beads and cells mounted on a microscope slide. Besides providing imaging capability, the imaging fibre can double up as a light conduit to illuminate the target tissue, which would be in synergy with the first two modalities, namely, the RGB and the dynamic speckle analyses. The research carried out in this thesis involved exploring and laying the foundations for a synergistic combination of optical methods which could potentially be integrated in future into a single probe for endoscopic applications. It is capable of identifying early stage cancerous regions of less than 5 mm in size from a distance and also micro-level imaging of structures less than 5 pm at proximity of the targeted area. As a future work direction, an actual probe system could be engineered integrating a micro-CCD camera and an imaging fiber bundle by applying the proposed multiple optical bubbles, polystyrene beads, cells and micro-fluidic flow channels. The scan based methodology developed, with the probe in proximity with the target tissue, could overcome loss of visual information due to the fiberlet interspaces and achieved a significant improvement in lateral resolution to 2.19 pm with the same optical elements as with the non-scan based scheme. With the fiber bundle enabling multiple scan points, thus saving scan time, and scanning steps designed to match the imaging pixel size for best resolution, a scheme to reconstruct high resolution images from the whole-field scan, at any region of interest was worked out. It was found that the central pixel in each of the fiberlet image are the optimal scan points as it gave the best contrast and the least offset in the scan image compared with other pixels within the fiberlet. The capability of the system was demonstrated using micron-sized fluorescent polystyrene beads and cells mounted on a microscope slide. Besides providing imaging capability, the imaging fibre can double up as a light conduit to illuminate the target tissue, which would be in synergy with the first two modalities, namely, the RGB and the dynamic speckle analyses. The research carried out in this thesis involved exploring and laying the foundations for a synergistic combination of optical methods which could potentially be integrated in future into a single probe for endoscopic applications. It is capable of identifying early stage cancerous regions of less than 5 mm in size from a distance and also micro-level imaging of structures less than 5 pm at proximity of the targeted area. As a future work direction, an actual probe system could be engineered integrating a micro-CCD camera and an imaging fiber bundle by applying the proposed multiple optical methodologies for effective identification, contouring and confirmation of abnormalities that can lead to early diagnosis of disease, such as cancer, in body cavities such as colon. These findings are expected to contribute to future research towards realising an in vivo optical biopsy probe.