融合三维统计形变结构的光学分子断层成像稀疏重建方法

Optical molecular tomography can provide effective means of accurate diagnosis at early stage of diseases, gene therapy and drug development, and consequently has a wide range of biomedical applications prospects.In order to break through the present limitation of this sort of techniques on imaging accuracy and speed, this project will establish tomography methods fused with computed tomography information to address the preclinical need for dynamic and continious evaluation on small animal model in vivo. Tissue segmentation methods based on a three-dimension statistical deformable model will be developed to achieve fast organ segmentation in complex organisms, and registration techniques based on mark points or feature points will be applied to accomplish precise registration between different imaging modalities. Fast numerical scheme for solving the radiative transfer equation will be combineded with the general graphics processing unit technique to decrease the time cost dramatically. By incorporating a priori information regarding anatomical structure and sparse distrution of targets, steady and robust reconstruction algorithms with high efficiency will be designed to obtain accurate recovery accuracy and high quality images with less measurement. Furthermore, this project will explore the application of optical molecular tomography combing CT structural informaiton in new drug development to quantitatively evaluate the drug-target interactions. The original contribution of this project will provide new methods and novel techniques for the research of optical molecuare imaging in quantitative evaluation of drug efficacy and early detection of cancers.Consequently, this project will have important scientific significance and theory value.

光学分子断层成像可为疾病早期诊断、基因治疗和药物研发等提供有效的在体研究手段，具有广泛的生物医学应用前景。为突破光学分子断层成像在成像精度和成像速度上的瓶颈，本项目拟建立融合计算机断层成像(CT)结构信息的光学断层成像方法，满足预临床研究对小动物模型在体、动态、连续量化评估的需求。创建基于三维统计形变模型的器官分割方法，解决复杂动物模型器官的快速分割难题；发展基于标记/特征点的光学与CT图像配准技术实现影像精确配准；利用通用图形处理单元加速辐射传输方程的数值求解，在此基础上融合结构、稀疏等先验信息，设计高效、鲁棒的重建算法在减少测量信息的同时提高重建精度和成像质量。此外，本项目将探索融合结构信息的光学分子断层成像技术在新药研发中的应用，为靶药相互作用评估提供分子影像学评价标准。本项目的研究为肿瘤早期检测及药物疗效定量评估等提供新方法与新技术，具有重要的理论价值和科学意义。

光学分子断层成像可为疾病早期诊断、基因治疗和药物研发等提供有效的在体研究手段，具有广泛的生物医学应用前景。为突破光学分子断层成像在成像精度和成像速度上的瓶颈，本项目建立了融合计算机断层成像(CT)结构信息的光学断层成像方法，满足预临床研究对小动物模型在体、动态、连续量化评估的需求。创建了基于三维统计形变模型的器官分割方法，解决复杂动物模型器官的快速分割难题；发展了基于标记/特征点的光学与 CT 图像配准技术实现影像精确配准；并利用通用图形处理单元加速辐射传输方程的数值求解，在此基础上融合结构、稀疏等先验信息，设计了高效、鲁棒的重建算法和策略在减少测量信息的同时提高重建精度和成像质量。此外，本项目融合结构信息的光学分子断层成像技术在靶向药物评估方面，提供分子影像学评价标准。本项目的研究为肿瘤早期检测及药物疗效定量评估等提供新方法与新技术，具有重要的理论价值和科学意义。