Weakly-supervised Convolutional Neural Networks of Renal Tumor Segmentation in Abdominal CTA Images

Background: Renal cancer is one of ten most common cancers in human beings. The laparoscopic partial nephrectomy (LPN) is an effective way to treat renal cancer. Localization and delineation of the renal tumor from pre-operative CT Angiography (CTA) is an important step for LPN surgery planning. Recently, with the development of deep neural networks in medical images, fully supervised deep neural networks can provide accurate pixel-wise organ and lesion segmentation. However, constructing the training dataset with manual labels consumes a lot of time. Methods: Therefore, in this work, we proposed a novel weakly-supervised convolutional neural network for renal tumor segmentation. A new three-stage training strategy was introduced to train a convolutional neural network, which includes group and weighted training phases. Results: We evaluated the proposed method on abdominal CT angiographic images of 200 patients. Extensive experimental results show that the proposed method achieves higher dice coefficient of 0.826 than the other two existing weakly-supervised deep neural networks. Furthermore, the segmentation performance is close to the fully supervised deep convolutional neural networks. Conclusions: The proposed strategy improves not only the efficiency of network training but also the precision of the segmentation.