SPECT Image Segmentation for Target Volume Estimation of Iodine-131 Radionuclide Therapy

An essential stage in the dosimetry process utilizing this imaging method involves the volume segmentation on single photon emission computed tomography (SPECT) images. The segmentation process needs to be both resilient and precise to ensure the reliability of the estimated values. However, the low sensitivity of the sodium iodide (NaI) SPECT detector to capture the high energy gamma rays from radionuclide Iodine-131 (131I) is causing poor resolution of the reconstructed images. Hence, this study investigated three distinct methods for automatic tumor delineation in 131I SPECT imaging. The three segmentation methods taken into consideration were a 42% fixed threshold (FT), adaptive threshold (AT), and gradient-weighted threshold (GWT). For the image reconstruction, the NEMA IQ phantom, having six spheres inside, was filled with 131I as a volume of interest (VOI). The results of each accuracy segmentation method were compared with actual volume via percentage deviation (%) and root mean square deviation (RMSD) for all the VOI calculated. GWT method showed a high-accuracy segmentation for the sphere with the ranging percentage deviation − 5.54 to 2.79%, and 2.18–54.62% for non-sphere VOI in the range of both actual volume 27.79–2.43 mL, while 2.71 and 12.1 for RMSD, respectively. The GWT method is found suitable for volume estimation of 131I SPECT imaging studies for sphere volume 2.43 ml and above.