Wu Zitong, Wang Xiaodong, Yu Jingjing, Yi Huangjian, He Xiaowei
J Opt Soc Am A Opt Image Sci Vis. 2018 Feb 1;35(2):328-335. doi: 10.1364/JOSAA.35.000328.
Fluorescence molecular tomography (FMT) is an important in vivo molecular imaging technique and has been widely studied in preclinical research. Many methods perform well in the reconstruction of a single fluorescent target but may fail in reconstructing multiple targets because of the severe ill-posedness of the FMT inverse problem. In this paper the original synchronization-inspired clustering algorithm (OSC) is introduced into FMT for resolving multiple targets from the reconstruction result. Based on OSC, a synchronization-based clustering algorithm for FMT (SC-FMT) is developed to further improve location accuracy. Both algorithms utilize the minimum spanning tree to automatically identify the number of the reconstructed targets without prior information and human intervention. A serial of numerical simulation results demonstrates that SC-FMT and OSC can resolve multiple targets robustly and automatically, which also shows the potential of the proposed postprocessing algorithms in FMT reconstruction.
荧光分子断层扫描（FMT）是一种重要的体内分子成像技术，已在临床前研究中得到广泛研究。许多方法在单个荧光靶点的重建中表现良好，但由于FMT反问题的严重不适定性，在重建多个靶点时可能会失败。本文将原始的同步启发聚类算法（OSC）引入FMT，以从重建结果中解析多个靶点。基于OSC，开发了一种用于FMT的基于同步的聚类算法（SC-FMT），以进一步提高定位精度。这两种算法都利用最小生成树在无需先验信息和人工干预的情况下自动识别重建靶点的数量。一系列数值模拟结果表明，SC-FMT和OSC能够稳健且自动地解析多个靶点，这也显示了所提出的后处理算法在FMT重建中的潜力。
