Solutions for clinical problems of near infrared (NIR) light imaging reconstruction with ultrasound localization

Diffuse optical tomography (DOT) using near infrared (NIR) light has a great potential for non-invasive imaging and monitoring tumor angiogenesis development. In recent years, the Optical and Ultrasound Imaging Lab at University of Connecticut has developed a dual-modality technique, which utilizes co-registered ultrasound images to guide the optical image reconstruction. We already collected data from over hundreds of biopsied patients, and obtained promising initial results. However, we encountered many problems during our clinical experiments. This dissertation focuses on the problems encountered in the clinical experiments and possible solutions. ^ First, we discuss a problem related to shallow lesions. Imaging shallow lesions of less than 1.0-1.5 cm in depth with DOT in reflection geometry is a challenge. This is because photons traveling from sources to detectors follow banana paths because of an absorbing boundary condition, therefore probe shallow regions with lower sensitivity. To solve this problem, we have designed a new probe which incorporates reflection geometry with either reflection boundary condition or absorption boundary condition with an angled source to improve the illumination of the shallow region underneath the ultrasound probe. The simulations and phantom experiments confirm this prediction. The clinical experiments show better overall reconstruction of the lesion. ^ Secondly, in NIR experiments, the patients with small breasts usually have a chest-wall problem. The chest-wall layer underneath the breast tissue consists of muscles and bones, which induce distortion to near infrared diffused wave measured at distant source-detector pairs when the reflection geometry is used. Two methods are proposed to solve this problem. The first method is using a priori information obtained from co-registered real-time ultrasound on the chest-wall to assist the removal of distant measurements, which are disturbed most by the chest-wall. The second method is using an analytical layered reconstruction method. The first method is easier to apply; however the second method provides better accuracy. ^ Thirdly, we discuss the shadow effect of highly-absorbing larger targets. For deeply located, highly absorbing large tumors, we have found that the reconstructed absorption values and the resulting total hemoglobin concentration are highly dependent on depth. In other words, the reconstructed absorption coefficients of the top layer are higher than those of deeper layers. We call this light shadow effect . A Monte Carlo method is used to simulate the photon trace inside the medium. The results indicate that the photon group which passes through the bottom part of a big target is much smaller than the photon group which passes through the upper part of a big target. The shadow effect, which is similar to the sound shadow effect frequently seen in pulse-echo ultrasound images, is an intrinsic property of optical diffuse tomography.