BC-MRI-SEG: A Breast Cancer MRI Tumor Segmentation Benchmark

TL;DR

This paper introduces BC-MRI-SEG, a comprehensive benchmark for breast cancer tumor segmentation in MRI images, enabling standardized evaluation and comparison of deep learning methods using publicly available datasets.

Contribution

It provides the first standardized benchmark with multiple datasets for breast cancer MRI segmentation, including zero-shot evaluation, and compares state-of-the-art approaches.

Findings

Benchmark facilitates fair comparison of methods.

Zero-shot evaluation highlights generalization capabilities.

Public datasets and code are openly available.

Abstract

Binary breast cancer tumor segmentation with Magnetic Resonance Imaging (MRI) data is typically trained and evaluated on private medical data, which makes comparing deep learning approaches difficult. We propose a benchmark (BC-MRI-SEG) for binary breast cancer tumor segmentation based on publicly available MRI datasets. The benchmark consists of four datasets in total, where two datasets are used for supervised training and evaluation, and two are used for zero-shot evaluation. Additionally we compare state-of-the-art (SOTA) approaches on our benchmark and provide an exhaustive list of available public breast cancer MRI datasets. The source code has been made available at https://irulenot.github.io/BC_MRI_SEG_Benchmark.