Automated Detection of the Occurrence and Changes of Hot-Spots in Intro-subject FDG-PET Images from Combined PET-CT Scanners

Dual-modality PET-CT imaging has been prevalently used as an essential diagnostic tool for monitoring treatment response in malignant disease patients. However, evaluation of treatment outcomes in serial scans by visual inspecting multiple PET-CT volumes is time consuming and laborious. In this paper, we propose an automated algorithm to detect the occurrence and changes of hot-spots in intro-subject FDG-PET images from combined PET-CT scanners. In this algorithm, multiple CT images of the same subject are aligned by using an affine transformation, and the estimated transformation is then used to align the corresponding PET images into the same coordinate system. Hot-spots are identified using thresholding and region growing with parameters determined specifically for different body parts. The changes of the detected hot-spots over time are analysed and presented. Our results in 19 clinical PET-CT studies demonstrate that the proposed algorithm has a good performance.