Domain Generalization for Endoscopic Image Segmentation by Disentangling Style-Content Information and SuperPixel Consistency

TL;DR

This paper introduces a style-content disentanglement method using instance normalization and whitening, combined with superpixel-based grouping, to improve domain generalization in endoscopic image segmentation across different imaging modalities.

Contribution

It proposes a novel style-content disentanglement approach with ISW and SUPRA for better domain-invariant segmentation in endoscopy, outperforming existing methods.

Findings

Achieved up to 18% improvement over SOTA methods on polyp dataset.

Surpassed the second-best method on Barrett's Esophagus dataset by nearly 2%.

Demonstrated significant performance gains in cross-modality endoscopic image segmentation.

Abstract

Frequent monitoring is necessary to stratify individuals based on their likelihood of developing gastrointestinal (GI) cancer precursors. In clinical practice, white-light imaging (WLI) and complementary modalities such as narrow-band imaging (NBI) and fluorescence imaging are used to assess risk areas. However, conventional deep learning (DL) models show degraded performance due to the domain gap when a model is trained on one modality and tested on a different one. In our earlier approach, we used a superpixel-based method referred to as "SUPRA" to effectively learn domain-invariant information using color and space distances to generate groups of pixels. One of the main limitations of this earlier work is that the aggregation does not exploit structural information, making it suboptimal for segmentation tasks, especially for polyps and heterogeneous color distributions. Therefore, in this work, we propose an approach for style-content disentanglement using instance normalization and instance selective whitening (ISW) for improved domain generalization when combined with SUPRA. We evaluate our approach on two datasets: EndoUDA Barrett's Esophagus and EndoUDA polyps, and compare its performance with three state-of-the-art (SOTA) methods. Our findings demonstrate a notable enhancement in performance compared to both baseline and SOTA methods across the target domain data. Specifically, our approach exhibited improvements of 14%, 10%, 8%, and 18% over the baseline and three SOTA methods on the polyp dataset. Additionally, it surpassed the second-best method (EndoUDA) on the Barrett's Esophagus dataset by nearly 2%.