BackSplit: The Importance of Sub-dividing the Background in Biomedical Lesion Segmentation

TL;DR

BackSplit improves small lesion segmentation in medical images by subdividing the background into detailed classes, leveraging richer anatomical context, and enhancing model training without increasing inference complexity.

Contribution

The paper introduces BackSplit, a novel approach that subdivides background classes in lesion segmentation, backed by information theory and extensive empirical validation.

Findings

BackSplit consistently improves small-lesion segmentation accuracy.

Using auxiliary labels from pretrained models enhances robustness.

The approach is effective across multiple datasets and architectures.

Abstract

Segmenting small lesions in medical images remains notoriously difficult. Most prior work tackles this challenge by either designing better architectures, loss functions, or data augmentation schemes; and collecting more labeled data. We take a different view, arguing that part of the problem lies in how the background is modeled. Common lesion segmentation collapses all non-lesion pixels into a single "background" class, ignoring the rich anatomical context in which lesions appear. In reality, the background is highly heterogeneous-composed of tissues, organs, and other structures that can now be labeled manually or inferred automatically using existing segmentation models. In this paper, we argue that training with fine-grained labels that sub-divide the background class, which we call BackSplit, is a simple yet powerful paradigm that can offer a significant performance boost without increasing inference costs. From an information theoretic standpoint, we prove that BackSplit increases the expected Fisher Information relative to conventional binary training, leading to tighter asymptotic bounds and more stable optimization. With extensive experiments across multiple datasets and architectures, we empirically show that BackSplit consistently boosts small-lesion segmentation performance, even when auxiliary labels are generated automatically using pretrained segmentation models. Additionally, we demonstrate that auxiliary labels derived from interactive segmentation frameworks exhibit the same beneficial effect, demonstrating its robustness, simplicity, and broad applicability.