MAMA-MIA: A Large-Scale Multi-Center Breast Cancer DCE-MRI Benchmark Dataset with Expert Segmentations

• DOI: 10.48550/arxiv.2406.13844
• Publication Date: 2024-06-19
• Authors: Lidia Garrucho; Claire-Anne Reidel; Kaisar Kushibar; Smriti Joshi; Richard Osuala; Apostolia Tsiriko...; Maciej Bobowicz; Javier del Riego; Alessandro Catanese; Katarzyna Gwoździ...; Maria-Laura Cosaka; Pasant M. Abo-Elhoda; Sara W. Tantawy; Shorouq S. Sakrana; Norhan O. Shawky-...; Amr Muhammad Abdo...; Androniki Kozana; Eugen Divjak; Gordana Ivanac; Katerina Nikiforaki; Michail E. Klontzas; Rosa García-Dosdá; Meltem Gulsun-Akp...; Oğuz Lafcı; Ritse M. Mann; Carlos Martín-Isla; Fred Prior; Kostas Marias; Martijn P. A. Sta...; Fredrik Strand; Oliver Díaz; Laura Igual; Karim Lekadir
• Keywords: Benchmark (surveying); Center (category theory)

Current research in breast cancer Magnetic Resonance Imaging (MRI), especially with Artificial Intelligence (AI), faces challenges due to the lack of expert segmentations. To address this, we introduce MAMA-MIA dataset, comprising 1506 multi-center dynamic contrast-enhanced MRI cases segmentations primary tumors and non-mass enhancement areas. These were sourced from four publicly available collections The Cancer Archive (TCIA). Initially, trained a deep learning model automatically segment cases, generating preliminary that significantly reduced segmentation time. Sixteen experts, averaging 9 years experience cancer, then corrected these segmentations, resulting final Additionally, two radiologists conducted visual inspection automatic support future quality control studies. Alongside provide 49 harmonized demographic clinical variables pretrained weights well-known nnUNet architecture using DCE-MRI full-images This dataset aims accelerate development benchmarking models foster innovation diagnostics treatment planning.