An Uncertainty-Driven GCN Refinement Strategy for Organ Segmentation

Abstract

Organ segmentation in CT volumes is an important pre-processing step in many computer assisted intervention and diagnosis methods. In recent years, convolutional neural networks have dominated the state of the art in this task. However, since this problem presents a challenging environment due to high variability in the organ's shape and similarity between tissues, the generation of false negative and false positive regions in the output segmentation is a common issue. Recent works have shown that the uncertainty analysis of the model can provide us with useful information about potential errors in the segmentation. In this context, we proposed a segmentation refinement method based on uncertainty analysis and graph convolutional networks. We employ the uncertainty levels of the convolutional network in a particular input volume to formulate a semi-supervised graph learning problem that is solved by training a graph convolutional network. To test our method we refine the initial output of a 2D U-Net. We validate our framework with the NIH pancreas dataset and the spleen dataset of the medical segmentation decathlon. We show that our method outperforms the state-of-the-art CRF refinement method by improving the dice score by 1% for the pancreas and 2% for spleen, with respect to the original U-Net's prediction. Finally, we perform a sensitivity analysis on the parameters of our proposal and discuss the applicability to other CNN architectures, the results, and current limitations of the model for future work in this research direction. For reproducibility purposes, we make our code publicly available at https://github.com/rodsom22/gcn_refinement.

Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future

“…However, compared to other modalities such as video, image, text, and audio, annotating graphs is more complex. Although the issue of missing labels is a general problem not specific to the graph domain, only a few works have adopted the training paradigms previously discussed, with semi-supervised examples being the segmentation of cerebral cortex [28] and organs [217].…”

“…This could be partially dealt with by a method that can learn how to construct graphs rather than hand-designing them. Soberanis-Mukul et al [217] also determined the uncertainty along with model expectation to perform a semi-supervised pancreas and spleen segmentation refinement task.…”

“…Related to this idea, CNN uncertainty estimation has been used as an attention mechanism for finding potentially misclassified regions for segmentation refinement [229]. Soberanis-Mukul et al [217], [218] proposed a semi-supervised graph learning problem operating over CT data for pancreas and spleen segmentation refinement task. First, a Monte Carlo dropout process is applied to a CNN (2D U-Net) to extract the model's expectation and uncertainty which are used to divide the CNN output into high confidence points (i.e.…”