Mitigating False Predictions In Unreasonable Body Regions

TL;DR

This paper introduces a novel loss function and a Body Part Regression model to improve the generalization of 3D medical image segmentation, reducing false predictions in body regions outside the training FOV and enhancing robustness across diverse image distributions.

Contribution

It proposes a new loss function that penalizes implausible predictions and utilizes a Body Part Regression model to improve segmentation generalization across varying FOVs.

Findings

Reduces false positive tumor predictions by up to 85%

Significantly improves segmentation performance across diverse FOVs

Enhances robustness of models in clinical settings

Abstract

Despite considerable strides in developing deep learning models for 3D medical image segmentation, the challenge of effectively generalizing across diverse image distributions persists. While domain generalization is acknowledged as vital for robust application in clinical settings, the challenges stemming from training with a limited Field of View (FOV) remain unaddressed. This limitation leads to false predictions when applied to body regions beyond the FOV of the training data. In response to this problem, we propose a novel loss function that penalizes predictions in implausible body regions, applicable in both single-dataset and multi-dataset training schemes. It is realized with a Body Part Regression model that generates axial slice positional scores. Through comprehensive evaluation using a test set featuring varying FOVs, our approach demonstrates remarkable improvements in generalization capabilities. It effectively mitigates false positive tumor predictions up to 85% and significantly enhances overall segmentation performance.