A method to detect iodine-125 seeds in lung CT

Detection of implanted iodine-125 seeds in postoperative CT is a necessary step for evaluating the output of seed implantation brachytherapy of lung tumor. In this paper, we propose a semi-automated method to detect implanted seeds in postoperative lung CT. Three main steps are included in our approach. Firstly, the ROI (Region Of Interest) containing all seeds is extracted from the original image. It is segmented using binary thresholding and needles in the binary image are removed based on the truth that they have longer length than seeds. Then, the touching seeds in each slice are split using a method based on seed gray feature. Finally, a label matrix is generated using a new geometry-based method and used to calculate the number, size, position and orientation of seeds. The efficiency of the method is tested and validated using eight clinical cases and the experimental result tells that our algorithm can achieve a 98.4 % detection rate.