基于迭代支撑检测的生物发光断层成像

发布时间：2018-08-12 08:18

【摘要】：生物发光断层成像(BLT)是一种高灵敏度、高特异性的光学分子影像技术,可根据探测到的生物体表光强来重建发光光源在生物体内的三维分布。由于生物体表面测得的光强信息有限,光源重建面临巨大的挑战。为了在有限的测量条件下获得更精确的重建光源,结合BLT中光源稀疏分布的特征,将重建问题转化为L1范数优化问题,并采用迭代支撑检测(ISD)算法实现快速重建,该算法交替执行支撑集检测和信号重构两个模块,直至重建精度达到要求。为了评估ISD算法的光源定位能力,设计数字鼠仿真实验,并与三种典型的稀疏重建算法比较。仿真结果表明ISD算法对于单光源和双光源目标均可以实现准确的重建。
[Abstract]:Bioluminescent tomography (BLT) is a highly sensitive and specific optical molecular imaging technology, which can reconstruct the three-dimensional distribution of luminescent light source in vivo according to the detected light intensity of biological body surface. Because of the limited information of light intensity measured on the surface of organism, the reconstruction of light source is faced with great challenge. In order to obtain more accurate reconstruction light source under limited measurement conditions, combining the characteristics of sparse distribution of light source in BLT, the reconstruction problem is transformed into L1 norm optimization problem, and the iterative support detection (ISD) algorithm is used to realize fast reconstruction. The algorithm alternately performs two modules: support set detection and signal reconstruction until the reconstruction accuracy reaches the requirement. In order to evaluate the ability of ISD algorithm to locate the light source, a digital rat simulation experiment was designed and compared with three typical sparse reconstruction algorithms. Simulation results show that ISD algorithm can achieve accurate reconstruction for both single and double light source targets.
【作者单位】：陕西师范大学物理学与信息技术学院;
【基金】：国家自然科学基金(61401264) 陕西省自然科学基金(2015JM6322) 中央高校基本科研业务费专项资金(GK201603025)
【分类号】：TP391.41

【参考文献】