SiNGR: Brain Tumor Segmentation via Signed Normalized Geodesic Transform Regression

TL;DR

This paper introduces a novel brain tumor segmentation method that models uncertainty near tumor boundaries as a regression problem using a signed geodesic transform, improving accuracy over traditional classification approaches.

Contribution

It proposes a new ground truth transformation based on signed geodesic transform and a focal-like regression loss, enabling effective voxel-level uncertainty modeling in tumor segmentation.

Findings

Outperforms state-of-the-art segmentation models in accuracy.

Architecture-agnostic approach adaptable to various models.

Provides publicly available code for reproducibility.

Abstract

One of the primary challenges in brain tumor segmentation arises from the uncertainty of voxels close to tumor boundaries. However, the conventional process of generating ground truth segmentation masks fails to treat such uncertainties properly. Those "hard labels" with 0s and 1s conceptually influenced the majority of prior studies on brain image segmentation. As a result, tumor segmentation is often solved through voxel classification. In this work, we instead view this problem as a voxel-level regression, where the ground truth represents a certainty mapping from any pixel to the border of the tumor. We propose a novel ground truth label transformation, which is based on a signed geodesic transform, to capture the uncertainty in brain tumors' vicinity. We combine this idea with a Focal-like regression L1-loss that enables effective regression learning in high-dimensional output space by appropriately weighting voxels according to their difficulty. We thoroughly conduct an experimental evaluation to validate the components of our proposed method, compare it to a diverse array of state-of-the-art segmentation models, and show that it is architecture-agnostic. The code of our method is made publicly available (\url{https://github.com/Oulu-IMEDS/SiNGR/}).