publications

2026

1. JIIM

   

   Explainability and Clinical Trust of Deep Learning in Glioblastoma Treatment Efficacy Prediction: A Comprehensive Framework

   Romain Andres, Noémie N Moreau, Loïse Dessoude, Loïc Le Henaff , and 5 more authors

   Journal of Imaging Informatics in Medicine, Oct 2026

   Published online 13 October 2026 (open access).

   Abs Bib HTML Code

   A multi-layered explainable AI (XAI) pipeline audits a previously developed deep learning model (ResNet-51q) that predicts glioblastoma overall survival from post-operative MRI: visual focus (Grad-CAM, LRP, LIME), feature dependence through input ablation, and the depth at which class information becomes decodable through linear probing. The audit reveals a high-intensity heuristic: hyper-intense pixels at the skull periphery anchor long-term predictions, while the model defaults to short-term based on tumour enhancement in their absence. In a multi-reader, multi-case clinical study (37 patients, 4 readers, 3 phases, 444 observations), AI assistance improved expert accuracy (OR 1.336 with the AI score, OR 1.598 with heatmaps added) without inflating confidence, and the heatmaps were associated with a change in how experts used the AI score and with fewer patients wrongly called long-term survivors.

   @article{andres_explainability_2026,
     title = {Explainability and Clinical Trust of Deep Learning in Glioblastoma Treatment Efficacy Prediction: A Comprehensive Framework},
     author = {Andres, Romain and Moreau, Noémie N and Dessoude, Loïse and Le Henaff, Loïc and Missohou, Fernand and Stefan, Dinu and Desmonts, Alexis and Herault, Romain and Corroyer-Dulmont, Aurélien},
     year = {2026},
     month = oct,
     journal = {Journal of Imaging Informatics in Medicine},
     doi = {10.1007/s10278-026-02381-5},
     keywords = {Explainability, Treatment efficacy, Interpretability, GBM, MRI, Deep learning},
     note = {Published online 13 October 2026 (open access).}
   }

2025

1. NeuroImage

   

   Development and routine implementation of a deep learning algorithm for automatic brain metastases segmentation on MRI for RANO-BM criteria follow-up

   Loïse Dessoude, Raphaëlle Lemaire, Romain Andres, Thomas Leleu , and 10 more authors

   NeuroImage, Feb 2025

   Dessoude, Lemaire and Andres contributed equally (co-first authors).

   Abs Bib HTML

   This study develops an AI solution that rapidly and accurately segments brain metastases on MRI, providing RANO-BM criteria and volume measurements for efficient clinical evaluation.

   @article{dessoude_development_2025,
     title = {Development and routine implementation of a deep learning algorithm for automatic brain metastases segmentation on {MRI} for {RANO-BM} criteria follow-up},
     author = {Dessoude, Loïse and Lemaire, Raphaëlle and Andres, Romain and Leleu, Thomas and Leclercq, Alexandre G. and Desmonts, Alexis and Corroller, Typhaine and Orou-Guidou, Amirath Fara and Laduree, Luca and Le Henaff, Loïc and Lacroix, Joëlle and Lechervy, Alexis and Stefan, Dinu and Corroyer-Dulmont, Aurélien},
     year = {2025},
     month = feb,
     journal = {NeuroImage},
     volume = {306},
     pages = {121002},
     doi = {10.1016/j.neuroimage.2025.121002},
     note = {Dessoude, Lemaire and Andres contributed equally (co-first authors).}
   }