Splenomegaly segmentation on computed tomography (CT) abdomen anatomical scans is essential for identifying spleenbiomarkers and has applications for quantitative assessment in patients with liver and spleen disease. Deep convolutionalneural network automated segmentation has shown promising performance for splenomegaly segmentation. However,manual labeling of abdominal structures is resource intensive, so the labeled abdominal imaging data are rare resourcesdespite their essential role in algorithm training. Hence, the number of annotated labels (e.g., spleen only) are typicallylimited with a single study. However, with the development of data sharing techniques, more and more publicly availablelabeled cohorts are available from different resources. A key new challenging is to co-learn from the multi-source data,even with different numbers of labeled abdominal organs in each study. Thus, it is appealing to design a co-learningstrategy to train a deep network from heterogeneously labeled scans. In this paper, we propose a new deep convolutionalneural network (DCNN) based method that integrates heterogeneous multi-resource labeled cohorts for splenomegalysegmentation. To enable the proposed approach, a novel loss function is introduced based on the Dice similarity coefficientto adaptively learn multi-organ information from different resources. Three cohorts were employed in our experiments,the first cohort (98 CT scans) has only splenomegaly labels, while the second training cohort (100 CT scans) has 15 distinctanatomical labels with normal spleens. A separate, independent cohort consisting of 19 splenomegaly CT scans withlabeled spleen was used as testing cohort. The proposed method achieved the highest median Dice similarity coefficientvalue (0.94), which is superior (p-value<0.01 against each other method) to the baselines of multi-atlas segmentation(0.86), SS-Net segmentation with only spleen labels (0.90) and U-Net segmentation with multi-organ training (0.91). Ourapproach for adapting the loss function and training structure is not specific to the abdominal context and may be beneficialin other situations where datasets with varied label sets are available.
展开▼
