Adaptive feature recombination and recalibration for semantic segmentation: application to brain tumor segmentation in MRI

TL;DR

This paper introduces adaptive feature recombination and a novel segmentation SE block to improve brain tumor segmentation in MRI by better leveraging spatial and contextual information in CNNs.

Contribution

It proposes a new feature recombination method and a segmentation-specific SE block for enhanced spatially-aware feature recalibration in FCNs.

Findings

Improved segmentation accuracy on brain tumor MRI data.

Enhanced feature representation through recombination and recalibration.

Better utilization of spatial and contextual information in CNNs.

Abstract

Convolutional neural networks (CNNs) have been successfully used for brain tumor segmentation, specifically, fully convolutional networks (FCNs). FCNs can segment a set of voxels at once, having a direct spatial correspondence between units in feature maps (FMs) at a given location and the corresponding classified voxels. In convolutional layers, FMs are merged to create new FMs, so, channel combination is crucial. However, not all FMs have the same relevance for a given class. Recently, in classification problems, Squeeze-and-Excitation (SE) blocks have been proposed to re-calibrate FMs as a whole, and suppress the less informative ones. However, this is not optimal in FCN due to the spatial correspondence between units and voxels. In this article, we propose feature recombination through linear expansion and compression to create more complex features for semantic segmentation. Additionally, we propose a segmentation SE (SegSE) block for feature recalibration that collects contextual information, while maintaining the spatial meaning. Finally, we evaluate the proposed methods in brain tumor segmentation, using publicly available data.