Image Segmentation of Thyroid SPECT Using Edge-Based Active Contour Model

Nuclear medicine is a reliable method to analyze or diagnose some type of diseases using functional imaging, for example SPECT imaging to obtain organs uptake and biodistribution. In order to get information needed, SPECT images have to be segmented first. Region of interest (ROI) method by the expertise is generally performed for image segmentation in SPECT imaging, but the task is daunting and time-consuming due to large number of images to be segmented. Among many methods that can be used for image segmentation, edge-based active contour is one of the most popular and widely used method in medical image segmentation. In addition to Level Set Method that fails to stop gradient magnitude by only using gradient information, here the edge-stop function (ESF) is proposed for edge-based active contour models to segment images with unclear boundaries using Matlab R2019a. In this study the Level Set Method with k-Nearest Neighbours classifier and fuzzy ESF are used to segment the thyroid images from the ^99mTc-sestamibi SPECT imaging. The number of iterations and initialization are assessed to obtain better segmented images which closely related to segmented images using ROI technique from the experts. The results showed that the proposed method could be used for segmentation of thyroid SPECT images.