DeepISLES: a clinically validated ischemic stroke segmentation model from the ISLES'22 challenge

Identificador:  imarina:9464098

Handle: https://hdl.handle.net/20.500.11797/imarina9464098

Autors:  de la Rosa, E; Reyes, M; Liew, SL; Hutton, A; Wiest, R; Kaesmacher, J; Hanning, U; Hakim, A; Zubal, R; Valenzuela, W; Robben, D; Sima, DM; Anania, V; Brys, A; Meakin, JA; Mickan, A; Broocks, G; Heitkamp, C; Gao, SB; Liang, KM; Zhang, ZJ; Siddiquee, MMR; Myronenko, A; Ashtari, P; Van Huffel, S; Jeong, H; Yoon, C; Kim, C; Huo, JY; Ourselin, S; Sparks, R; Clèrigues, A; Oliver, A; Lladó, X; Chalcroft, L; Pappas, I; Bertels, J; Heylen, E; Moreau, J; Hatami, N; Frindel, C; Qayyum, A; Mazher, M; Puig, D; Lin, SC; Juan, CJ; Hu, TX; Boone, L; Goubran, M; Liu, YJ; Wegener, S; Kofler, F; Ezhov, I; Shit, S; Petzsche, MRH; Müller, M; Menze, B; Kirschke, JS; Wiestler, B

Resum:
Diffusion-weighted MRI is critical for diagnosing and managing ischemic stroke, but variability in images and disease presentation limits the generalizability of AI algorithms. We present DeepISLES, a robust ensemble algorithm developed from top submissions to the 2022 Ischemic Stroke Lesion Segmentation challenge we organized. By combining the strengths of best-performing methods from leading research groups, DeepISLES achieves superior accuracy in detecting and segmenting ischemic lesions, generalizing well across diverse axes. Validation on a large external dataset (N = 1685) confirms its robustness, outperforming previous state-of-the-art models by 7.4% in Dice score and 12.6% in F1 score. It also excels at extracting clinical biomarkers and correlates strongly with clinical stroke scores, closely matching expert performance. Neuroradiologists prefer DeepISLES' segmentations over manual annotations in a Turing-like test. Our work demonstrates DeepISLES' clinical relevance and highlights the value of biomedical challenges in developing real-world, generalizable AI tools. DeepISLES is freely available at https://github.com/ezequieldlrosa/DeepIsles.