FFT BASED IMAGES WITH DIFFUSING LIGHT WAVES

点击翻译按钮获取中文摘要

DESCRIPTION (Adapted from Applicant's Abstract): The applicants proposed to improve the tumor diagnostic capabilities of optical spectroscopy and imaging based on diffusing near infrared light. The proposed work aims to develop and assess the utility of a new tomographic image reconstruction scheme that may be applicable, for example, to tumor detection and characterization within the human breast. The theoretical approaches will be tested experimentally using existing photon migration instrumentation and computer codes primarily developed for compressed breast and box geometries. The detrimental effects of boundaries on image fidelity will be mitigated using a novel theoretical approach that incorporates experimental measurements of the photon density waves on the boundary. Finally, a realistic prototype parallel plate instrument will be built and tested in phantoms. If successful, the applicants anticipate combining the optimal device with existing instrumentation for diffuse-optical / MRI breast diagnostics at their institution. In this scenario, the applicants plan to apply for additional funding to support these clinical studies. The primary benefit derivable from the proposed research will be improved image fidelity and reconstruction speed, which in turn, will enable clinicians to fully exploit the new spectroscopic and scattering contrast mechanisms available with the optical method for increased tumor sensitivity and specificity. The scheme is a near-field Fast Fourier Transform (FFT) approach that affords a rapid imaging of absorption and scattering variations that are either intrinsic properties of the tissue or are selectively induced by administration of optical contrast agents. The method offers the possibility of Projection (2-D) images and full three- dimensional reconstructions. The projection images may be combined with complementary localizing techniques to deduce optical properties of tissue without full volume reconstruction. The method is new to the field; it differs from least squares techniques such as ART and SIRT in that it is fast and predominately non-iterative; its computational complexity scales with voxel number as N2lnN rather than N3. It also offers a well defined prescription for handling boundary effects. If successful, it is anticipated that the anatomical information generated by the devices based on these theories will be useful for improving our ability to specify the nature of tumors inside breast in a non-invasive, non-ionizing and cost effective way.