%0 Journal Article %A Mohammed Goryawala %A Seza Gulec %A Rekha Suthar %A Ruchir Bhatt %A Anthony McGoron %A Malek Adjouadi %T A low-interaction automatic 3D liver segmentation method for selective internal radiation therapy %D 2012 %J Journal of Nuclear Medicine %P 2251-2251 %V 53 %N supplement 1 %X 2251 Objectives Selective Internal Radiation Therapy (SIRT) with Yttrium-90 (Y-90) microspheres is an emerging technique for effective liver-directed therapy. SIRT requires accurate determination of the functional tumor volumes and the anatomical liver volumes to gauge the Y-90 microsphere dose to be delivered to the tumor. The study presents a liver segmentation approach directed towards SIRT for accurate calculation of the anatomic liver volumes as an alternative to the tedious manual segmentation process currently employed. Methods The novelty of this algorithm is in using a single slice in the initialization process to segment the entire liver in 3D. The algorithm relies on a localized contouring algorithm in combination with a modified k-means approach. Newly developed initialization masks in support of the contouring algorithm allowed for much reduced human intervention. Intensity based region growing along with a new volume of interest (VOI) based correction technique are brought together to achieve the single slice initialization process. The performance of the algorithm is assessed experimentally using 20 liver Computed Tomography (CT) scans. Statistical analysis was conducted to determine the consistency of the segmentation process and volume estimation errors in comparison to the manual gold standard. Results Experiments conducted on the CT scans show an average accuracy of 96.8% for volumetric calculation and an average Dice coefficient of 0.92. Statistical tests show that the algorithm is highly consistent (p = 0.67) and independent of the user initialization (p = 0.20). Conclusions The study demonstrated the development of a robust and accurate liver segmentation technique requiring minimal human interaction. An accurate calculation of the liver volume would help in gauging the most effective radioactive dosage to the patient %U