Structure-aware scale-adaptive networks for cancer segmentation in whole-slide images

TL;DR

This paper introduces a structure-aware, scale-adaptive network for cancer segmentation in whole-slide images, effectively handling vague boundaries and small regions, and demonstrating superior performance on liver and colorectal cancer datasets.

Contribution

It proposes a novel scale-adaptive feature selection module and a structural similarity metric, enhancing segmentation accuracy and robustness in challenging cancer image regions.

Findings

Achieves top performance in PAIP 2019 liver cancer segmentation challenge.

Outperforms existing attention mechanisms in efficiency and accuracy.

Improves baseline network with the proposed scale-adaptive module.

Abstract

Cancer segmentation in whole-slide images is a fundamental step for viable tumour burden estimation, which is of great value for cancer assessment. However, factors like vague boundaries or small regions dissociated from viable tumour areas make it a challenging task. Considering the usefulness of multi-scale features in various vision-related tasks, we present a structure-aware scale-adaptive feature selection method for efficient and accurate cancer segmentation. Based on a segmentation network with a popular encoder-decoder architecture, a scale-adaptive module is proposed for selecting more robust features to represent the vague, non-rigid boundaries. Furthermore, a structural similarity metric is proposed for better tissue structure awareness to deal with small region segmentation. In addition, advanced designs including several attention mechanisms and the selective-kernel convolutions are applied to the baseline network for comparative study purposes. Extensive experimental results show that the proposed structure-aware scale-adaptive networks achieve outstanding performance on liver cancer segmentation when compared to top ten submitted results in the challenge of PAIP 2019. Further evaluation on colorectal cancer segmentation shows that the scale-adaptive module improves the baseline network or outperforms the other excellent designs of attention mechanisms when considering the tradeoff between efficiency and accuracy.