Gastrointestinal tract cancer, the most common type of cancer, has a very low survival rate, especially for pancreatic cancer (five year survival rate of 5%) and bile duct cancer (five year survival rate of 12%). Here, we propose to use an integrated OCT-US catheter for cancer detection. OCT is targeted to acquire detailed information, such as dysplasia and neoplasia, for early detection of tumors. US is used for staging cancers according to the size of the primary tumor and whether or not it has invaded lymph nodes and other parts of the body. Considering the lumen size of the GI tract, an OCT system with a long image range (>10mm) and a US imaging system with a center frequency at 40MHz (penetration depth > 5mm) were used. The OCT probe was also designed for long-range imaging. The side-view OCT and US probes were sealed inside one probe cap piece and one torque coil and became an integrated probe. This probe was then inserted into a catheter sheath which fits in the channel of a duodenoscope and is able to be navigated smoothly into the bile duct by the elevator of the duodenoscope. We have imaged 5 healthy and 2 diseased bile ducts. In the OCT images, disorganized layer structures and heterogeneous regions demonstrated the existence of tumors. Micro-calcification can be observed in the corresponding US images.
Premalignant diseases of the gastrointestinal tract, such as Barrett's esophagus, long-standing ulcerative colitis, and
adenomatous polyps, have a significantly increased risk for development of adenocarcinoma, most often through an
intermediate stage of dysplasia. Adenocarcinoma of the colon is the second most common cancer in the United States.
Because patients with colorectal cancer often present with advanced disease, the outcomes are associated with
significant morbidity and mortality. Effective methods of early detection are essential. As non-polypoid dysplasia is not
visible using conventional endoscopy, surveillance of patients with Barrett's esophagus and ulcerative colitis is
performed via a system in which multiple random biopsies are obtained at prescribed intervals. Sampling error and
missed diagnoses occur frequently and render current screening methods inadequate. Also, the examination of a tissue
biopsy is time consuming and costly, and significant intra- and inter-observer variation may occur. The newer methods
discussed herein demonstrate the potential to solve these problems by early detection of disease with high sensitivity and
Conventional endoscopy is based on the observation of white light reflected off the tissue surface. Subtle changes in
color and shadow reveal structural changes. New developments in optical imaging go beyond white light, exploiting
other properties of light. Several promising methods will be discussed at this meeting and shall be briefly discussed
below. However, few such imaging modalities have arrived at our clinical practice. Some much more practical methods
to improve colorectal cancer screening are currently being evaluated for their clinical impact. These methods seek to
overcome limitations other than those of detecting dysplasia not visible under white light endoscopy. The current
standard practice of colorectal cancer screening utilizes colonoscopy, an uncomfortable, sometimes difficult medical
procedure. Efforts to improve the practice of colonoscopy will be described. Another limitation of the current practice is
the inability to detect polypoid neoplasia that is hidden from view under white light imaging by the natural folds that
occur within the colon. A device to overcome this limitation will also be described. Efforts to improve colorectal cancer
screening (and thereby decrease the death rate of this second leading cause of cancer death in the United States) are
progressing in many arenas. The researcher, basic or clinical, should maintain an up to date overview of the field and
how each new technological advance is likely to have a role in the screening and early detection of colorectal cancer.
In this investigation, an optical system is introduced for inspecting the interiors of confined spaces, such as the walls of containers, cavities, reservoirs, fuel tanks, pipelines, and the gastrointestinal tract. The optical system wirelessly transmits stereoscopic video to a computer that displays the video in realtime on the screen, where it is viewed with shutter glasses. To minimize space requirements, the videos from the two cameras (required to produce stereoscopic images) are multiplexed into a single stream for transmission. The video is demultiplexed inside the computer, corrected for fisheye distortion and lens misalignment, and cropped to the proper size. Algorithms are developed that enable the system to perform these tasks. A proof-of-concept device is constructed that demonstrates the operation and the practicality of the optical system. Using this device, tests are performed assessing validities of the concepts and the algorithms.
We presented two approaches for separating a diffusive component of the backscattered signal originated in deep tissue layers and a non-diffusive single backscattering component which backscattered from a thin epithelial layer. Both approaches can be effective and have their advantages and disadvantages. The modeling technique can provide important information about hemoglobin concentration, oxygenation, and average scattering properties of the mucosal tissue. On the other hand, when applied to new tissues, it has to be adjusted to take into account tissue morphology. Also, the polarization technique can be very robust and more effective in background removal. However, it lacks extracting capabilities of the modeling technique. Both techniques can be quite valuable and compliment each other in a future clinical device.
Previous in vitro studies showed that autofluorescence images of colonic mucosa collected endoscopically can be used to detect dysplasia with high sensitivity. This method is extended to collection of fluorescence images of adenomatous polyps in vivo. Fluorescence images were collected during colonoscopy in 30 patients. A total of 12 adenomatous and 6 hyperplastic polyps were identified. An optical fiber excitation probe, located in the instrument channel of the colonoscope, delivered 300 mW of near- ultraviolet light at (lambda) <SUB>ex</SUB> equals 351 and 364 nm. Mucosal fluorescence in the spectral bandwidth between 400 and 700 nm was imaged, processed, and displayed with various likelihoods of associated dysplasia. Adenomatous polyps exhibited decreased fluorescence intensity compared to adjacent mucosa with normal appearance. With the fluorescence threshold set to 80% of the average intensity of normal mucosa, a sensitivity of 83% for dysplasia detection was achieved. All hyperplastic polyps were correctly identified as being non-dysplastic. Optimal identification of dysplastic regions was obtained with the colonoscope oriented at near-normal incidence to the polyps. At higher angles of incidence, artifacts due to illumination shadows were introduced. The dysplasia associated with adenomatous polyps can be detected in vivo on fluorescence imaging with high sensitivity, thus demonstrating the potential to guide endoscopic biopsy.
Several groups have shown that laser-induced fluorescence spectroscopy can detect dysplastic changes in human colon tissues. We present an approach based on analysis of the underlying tissue microstructure for extracting histological information from such spectral signals. The method employs fluorescence microscopy and tissue optics to model the `bulk' fluorescence collected with an optical fiber probe in a clinical setting. For both colonic normal and adenoma, we measured the intrinsic fluorescence lineshapes, the spatial distributions of the fluorophores, and optical parameters of tissue. Numerical and analytical solutions to the radiative transfer equation were then used to compute fluorescence spectra. The results of the model were in excellent agreement with clinical spectra collected during colonoscopy, using 370 nm excitation. Four factors were found to be responsible for the spectral differences between normal tissue and adenoma: fluorescence of mucosal collagen, dysplastic cell, and submucosa, and hemoglobin attenuation. Preliminary results indicate that these parameters can be extracted from individual clinical spectra by reversing the modeling procedure.
We are studying the use of the laser-induced fluorescence (LIF) endoscopic images of colonic mucosa for detection of pre-malignant lesions. LIF images were collected through a fiber optic colonoscope, and adenomatous polyps were used as a model of dysplasia. A total of 12 tissue samples containing 29 adenomas, obtained from colectomy specimens from 3 familial adenomatous polyposis patients, were studied. Regions of colonic mucosa were illuminated by a quartz optical fiber with near-UV light from an argon-ion laser. Autofluorescence between 400 and 700 nm was detected by means of an intensified CID camera. In the LIF images, adenomatous polyps appeared lower in intensity than normal mucosa by about a factor of 2. The LIF images were processed by dividing the raw image by a spatially averaged one to correct for differences in the distance to the tissue and in the light collection efficiency of the optics. Relative intensity thresholds were set at values varying between 55% and 90% compared to the spatial average to determine likely areas of disease. The results were compared to histology taken at 2 mm intervals along several transverse cross-sections of the specimens. At a threshold of 75%, 26 true positives, 256 true negatives, 22 false positives, and 3 false negatives were identified, resulting in a sensitivity, specificity, positive predictive value and percentage of correct determinations of 90%, 92%, 54%, and 92%, respectively. These values are comparable to results of independent diagnoses by two pathologists, demonstrating the potential of LIF to guide biopsy.
Laser-induced fluorescence spectroscopy is a promising technique for detecting colonic dysplasia in vivo but, at present, the biological basis for the success of the method is poorly understood, and little information is provided as to the morphological/molecular origin of tissue fluorescence. We present a methodology for establishing this, taking as a starting point a recently completed prospective LIF clinical study of colonic polyps. The method is based on a morphological model of tissue fluorescence with three components: the intrinsic lineshapes of the fluorophores, the spatial distributions of their intensities, and the optical parameters of tissue. We measure these using fluorescence microspectroscopy, microscopic imaging and tissue optics, respectively. The model predicts the features of the clinical spectra, and quantifies the respective intrinsic and architectural contributions. The results can be inverted to extract biological parameters from the spectra, and used to select optimal excitation wavelength(s) and guide probe design. Implications for the detection of non-polypoid dysplasia are discussed.
Laser-induced fluorescence spectroscopy is a promising technique for detecting colonic dysplasia in vivo but, at present, the biological basis for the success of the method is poorly understood, and little information is provided as to the morphological/molecular origin of tissue fluorescence. We present a methodology for establishing this, taking as a starting point a recently completed prospective laser-induced fluorescence (LIF) clinical study of colonic polyps. The method is based on a morphological model of tissue fluorescence with three components: the intrinsic lineshapes of the fluorophores, the spatial distributions of their intensities, and the optical parameters of tissue. We measure these using fluorescence microspectroscopy, microscopic imaging and tissue optics, respectively. The model predicts the features of the clinical spectra, and quantifies the respective intrinsic and architectural contributions. The results can be inverted to extract biological parameters from the spectra, and used to select optimal excitation wavelength(s) and guide probe design. Implications to detection of nonpolypoid dysplasia are discussed.
Raman spectroscopy can provide quantitative molecular information about the biochemical composition of human tissues exhibiting various stages of disease. Fluorescence interference is ubiquitous in Raman spectra of biological samples excited with visible light. However, it can be avoided by using near-infrared (NIR) or ultraviolet (UV) excitation. We are exploring the potential of these methods for detecting precancerous/cancerous changes in human tissues. The NIR studies use 830 nm excitation from a Ti:sapphire laser. Raman signals are collected by an imaging spectrograph/deep-depletion CCD detection system. High quality tissue spectra can be obtained in a few seconds or less. The UV resonance Raman studies employ wavelengths below 300 nm for selective excitation of nucleic acids, proteins and lipids. Excitation is provided by a frequency tripled/quadrupled mode-locked Ti:sapphire laser, and Raman light is collected by a one meter spectrograph/UV-enhanced CCD detector. The two systems can be coupled to appropriate microscopes for extracting morphological and biochemical information at the cellular level, which is important for understanding the origin of the Raman spectra of bulk tissue. The results of the initial studies for cancer detection in various human tissues are reported here.