论文信息 - A new image-reconstruction approach to optical computed tomography based on BP neural network

A new image-reconstruction approach to optical computed tomography based on BP neural network

A new image reconstruction approach to optical computed tomography is proposed in which a BP neural network is used to express the nonlinear relation between the change of optical properties inside the biology tissue and photons distributing change around the surface. A modeling and simulation tool named Femlab and finite element method has been tested wherein the forward model on a basis of the diffusion equation. Based on the forward problem, a BP neural network was established to solve the inverse problem. Thus, the position and its corresponding optical properties of tissue change could be recognized by the network. New approach was suitable for clinical application for its fast reconstruct characteristic.

Qiong Wu | Zhiyu Qian

[1] Eiji Okada,et al. Relationship between position of brain activity and change in optical density for NIR imaging , 2000, European Conference on Biomedical Optics.

[2] M. Tamura,et al. Dynamic multichannel near-infrared optical imaging of human brain activity. , 1993, Journal of applied physiology.

[3] Charles A. Bouman,et al. Nonlinear multigrid algorithms for Bayesian optical diffusion tomography , 2001, IEEE Trans. Image Process..

[4] K D Paulsen,et al. Three-dimensional simulation of near-infrared diffusion in tissue: boundary condition and geometry analysis for finite-element image reconstruction. , 2001, Applied optics.