Improving Pre-trained Segmentation Models using Post-Processing

TL;DR

This paper introduces adaptive post-processing methods to improve glioma segmentation accuracy from large pre-trained models, emphasizing efficiency, fairness, and clinical relevance.

Contribution

It presents novel post-processing techniques that significantly enhance segmentation quality, shifting focus from complex models to efficient refinement strategies.

Findings

Segmentation ranking improved by 14.9% in sub-Saharan Africa challenge.

Achieved a 0.9% improvement in the adult glioma challenge.

Highlights the importance of post-processing for clinical and sustainable AI applications.

Abstract

Gliomas are the most common malignant brain tumors in adults and are among the most lethal. Despite aggressive treatment, the median survival rate is less than 15 months. Accurate multiparametric MRI (mpMRI) tumor segmentation is critical for surgical planning, radiotherapy, and disease monitoring. While deep learning models have improved the accuracy of automated segmentation, large-scale pre-trained models generalize poorly and often underperform, producing systematic errors such as false positives, label swaps, and slice discontinuities in slices. These limitations are further compounded by unequal access to GPU resources and the growing environmental cost of large-scale model training. In this work, we propose adaptive post-processing techniques to refine the quality of glioma segmentations produced by large-scale pretrained models developed for various types of tumors. We demonstrated the techniques in multiple BraTS 2025 segmentation challenge tasks, with the ranking metric improving by 14.9 % for the sub-Saharan Africa challenge and 0.9% for the adult glioma challenge. This approach promotes a shift in brain tumor segmentation research from increasingly complex model architectures to efficient, clinically aligned post-processing strategies that are precise, computationally fair, and sustainable.