In this study we propose the development of a near infrared (NIR) time- resolved optical imaging system for the detection of tumors in breast tissue. The early detection of cancer of the breast is a major public health issue. The standard method of X-ray mammography suffers from the need to expose patients to ionizing radiation. Because of the intense scattering of light by tissue, the ability to generate high quality images based on an optical measurement will require the need to evaluate signals which have been highly scattered. Development of this system will proceed along two problem areas; data collection and data analysis. The experimental program will involve a systematic investigation of model scattering media (microscopic latex beads in agar) in which phantoms will be placed. Data collection will involve performing one- and two-dimensional surface scans at multiple source locations using time-resolved (pico- and femtosecond NIR laser sources) and, in some cases, time-independent measurement schemes. The size and number density of the scattering particles will be varied to explore a range of scattering media (from highly forward directed to nearly isotropic scattering). Media having simple structures and regular geometries will be studied, initially, leading to the investigation of increasingly more complex media. Following this, tissue specimens will be studied, including those known to contain cancerous lesions. The analytical program will consist of the three parts. The first concerns the numerical modeling of photon transport in highly scattering media. Using Monte Carlo and other numerical methods, the spatial distribution of detected photon fluxes as a function of source-detector configurations, illumination scheme, and geometry of the medium will be modeled. In addition, the importance of reflection and refraction of photons at internal boundaries and at the exterior will be studied, as a function of complexity of the medium. Solution of the inverse problem will be evaluated by solving a linear perturbation equation we have recently developed. A variety of methods will be employed to solve the problem including gradient decent methods and methods which can readily incorporate apriori information needed to constrain the solution. In addition, hybrid methods which make use of multigrid techniques will be studied. Because of the inherent ill-posedness of the inverse problem, we also propose to examine a variety of strategies which can yield the necessary apriori information. Several have already been identified but further development and testing is needed. As a group, we have considerable expertise in each of the major problem areas that must be considered to rigorously pursue this program. Essentially all of the major experimental facilities required are in place and operational, and trained technical personnel are on-site. In addition, we have direct access to significant computing facilities that are essential to this program. Further, extensive preliminary work in all the major problem areas has been conducted, and results obtained have substantiated the feasibility of the proposed studies.
Showing the most recent 10 out of 11 publications
