Published in Bioinformatics, 2022
This paper presents BIODICA, a computational environment for the application of Independent Component Anlysis to omics data.
Recommended citation: Captier N, Merlevede J, Molkenov A, Seisenova A, Zhubanchaliyev A, Nazarov PV, Barillot E, Kairov U, Zinovyev A. BIODICA: a computational environment for Independent Component Analysis of omics data. Bioinformatics. 2022.
Published in Journal of Nuclear Medicine, 2024
This paper presents an independent evaluation of two new imaging biomarkers for Positron Emission Tomography (PET), related to the distance between the tumor hotspot of the radiotracer uptake and its centroid or surface.
Recommended citation: Hovhannisyan-Baghdasarian N, Luporsi M, Captier N, Nioche C, Cuplov V, Woff E, Hegarat N, Livartowski A, Girard N, Buvat I, Orlhac F. Promising Candidate Prognostic Biomarkers in [18F]FDG PET Images: Evaluation in Independent Cohorts of Non-Small Cell Lung Cancer Patients. J Nucl Med. 2024.
Published in Journal of Nuclear Medicine, 2024
This paper presents RadShap, a model- and modality-agnostic tool, based on Shapley values, that explains the predictions of multiregion radiomic models by highlighting the contribution of each individual region.
Recommended citation: Captier N, Orlhac F, Hovhannisyan-Baghdasarian N, Luporsi M, Girard N, Buvat I. RadShap: An Explanation Tool for Highlighting the Contributions of Multiple Regions of Interest to the Prediction of Radiomic Models. J Nucl Med. 2024.
2024
This paper presents the conclusions of an extensive benchmark of machine learning pipelines (i.e, learning algorithm, feature selection strategy, hyperparameter tuning) to predict survival outcomes in medical datasets.
Recommended citation: Rebaud L, Capobianco N, Captier N, Escobar T, Spottiswoode B, Buvat I. Similar performance of 8 machine learning models on 71 censored medical datasets: a case for simplicity. medRxiv. 2024.
Published in Nature Communications, 2025
This paper presents the conclusions of our extensive benchmark of multimodal machine learning approaches to predict immunotherapy outcome in non-small cell lung cancer (NSCLC). This analysis was performed on an original cohort of metastatic NSCLC patients treated with first line immunotherapy, gathering data modalities such as Positron Emission Tomography scans, bulk RNA transcriptomic data from biopsy tissues, or pathological slides.
Recommended citation: Captier, N., Lerousseau, M., Orlhac, F. et al. Integration of clinical, pathological, radiological, and transcriptomic data improves prediction for first-line immunotherapy outcome in metastatic non-small cell lung cancer. Nat Commun 16, 614 (2025).
Python package which provides a stabilized algorithm for Independent Component Analysis (ICA). 
Multimodal PyTorch models with attention mechanism for the prediction of immunotherapy response in non-small cell lung cancer
Python code for multimdoal machine learning strategies (early and late fusion) for classification and survival tasks.
Python toolbox to complement stabilized-ica and apply ICA to omics data. 
Python package for highlighting the contribution of different regions of interest (ROIs) to the predictions of multiregion models with Shapley values. 
Published:
Published:
Training School Radiomics and AI in Molecular Imaging, Medical University Vienna, 2021
Download Download Slides