Analytical image reconstruction algorithm for diffuse optical tomography

• Main Author: Sim, Soon Kim.
• Other Authors: Lee Kijoon
• Format: Final Year Project
• Language: English
• Published: 2009
• Subjects: DRNTU::Engineering::Bioengineering
• Online Access: http://hdl.handle.net/10356/16640
• Institution: Nanyang Technological University

Diffuse Optical Tomography (DOT) is a relatively new technique in biomedical imaging that provides functional imaging on physiological parameters such as blood volume and oxygenation levels based on the optical properties of the target tissues. It is complement with other current imaging techniques such as MRI and CT to generate additional information. The main problem with DOT is that it requires the processing of large amount of data to generate images of acceptable resolution, thus straining the computer’s process and memory, reducing the reconstruction/ processing speed. An algorithm, recently developed by Konnecky et al, promises to reduce the increase the processing speed, with a minimal sacrifice in quality of the image. The algorithm makes use of symmetries of the Green function to reduce the reconstruction process, along with inversion of multiple small matrices instead of a large one to gain the large speed factor. Other than speed, the algorithm has been shown to be capable of imaging targets of complex geometries. Thus the objectives of this project are to construct a DOT program and investigate its capabilities in imaging. A simple setup was used for imaging in the slab geometry. three imaging targets were used: a small dark eraser, 2 small silicon phantoms, and a silicon phantoms shaped in the letters N, T and U. Reconstruction was performed using an MATLAB implementation of Konnecky’s algorithm. The reconstructed images for the dark eraser and 2 small phantoms were able to accurately show their relative positions and size, but not their shape. Reconstruction of the NTU-shaped phantom was partially successful: the letterings were barely visible on the reconstructed image due to a small set-up error. Further improvement of ii image quality can be obtained by modification of experimental procedures and data preprocessing steps.