Edouard Duchesnay - Data Science, Neuroimaging

Research Scientist in Machine Learning applied to NeuroImaging at NeuroSpin, CEA, Paris-Saclay, France. ORCID:0000-0002-4073-3490

Machine learning to discover neural predictive signature of psychiatric disorder


Machine Learning, Neuroimaging, Scientific computing, Software Engineering, Bioinformatics, High Dimensional Data. Machine Learning in Neuroimaging

Curriculum Vitae in pdf



  • 1999-2001 Ph.D. in Image Processing at LTSI (laboratory of signal and image processing) of Rennes 1 University, France.
  • 1997-1998 Master’s degree in Signal/Image Processing. in Rennes 1 University, France.
  • 1992-1997 Master’s degree in Software Engineering. École Pour l’Informatique et les Techniques Avancées (EPITA), Kremlin Bicêtre, France.

Bio (IEEE style)

Edouard Duchesnay received the engineer’s degree in software engineering from École Pour l’Informatique et les Techniques Avancées (France) in 1997, the M.Sc. degree in signal/image processing from Rennes 1 University (France) in 1998 and the Ph.D. degree in signal and image processing Rennes 1 University (France) in 2001.

Since 2008, E. Duchesnay is a research scientist at NeuroSpin/CEA: an MRI neuroimaging center within the CEA. He developed multivariate machine learning algorithms (ML) classification/regression to capture complex relationships to make inferences at an individual level in the perspective of computer-aided diagnosis/prognosis or biomarkers discovery for brain diseases (Duchesnay et al., NeuroImage, 2011).

To investigate genetic influence on the brain, he proposed multivariate latent variable models (Le Floch et al, NeuroImage, 2012) that integrate sparsity and specific feature selection within learning algorithms to alleviate large dimensionality of both imaging and genetic data.

Given the limitations of state-of-the-art sparse algorithms to produce stable and interpretable predictive signatures, he proposed to push forward the regularization approaches extending classical algorithms with structural constraints issued from the known biological structure (spatial structure of the brain and the linkage disequilibrium or pathways of OMICs data) in order to force the solution to adhere to biological priors, producing more plausible interpretable solutions.

The main outcome is a Python ML library called ParsimonY. This library is dedicated to high dimensional structured input data Hadj-Selem et al., IEEE-TMI, 2018 and supplementary. It has been used to identify neuroimaging functional signature of hallucinations in patients with schizophrenia De Pierrefeu et al., Hum. Brain Mapp., 2018. We extended the popular PCA (Principal Component Analysis) with spatial regularization to identify interpretable patterns of the neuroimaging variability in either functional or anatomical meshes of the cortical surface De Pierrefeu et al., IEEE-TMI, 2018.

Scientific impact


  • Publications: 58(a), 96(b)
  • Total Number of Citations: 4,258(a), 10,962(b)
  • H-Index: 18(a), 22(b)

(a)Web of Science, (b)Google scholar


Five most significant scientific articles

  • A. De Pierrefeu, T. Löfstedt, C. Laidi, F. Hadj-Selem, J. Bourgin, T. Hajek, F. Spaniel, M. Kolenic, P. Ciuciu, N. Hamdani, M. Leboyer, T. Fovet, R. Jardri, J. Houenou, E. Duchesnay “Identifying a neuroanatomical signature of schizophrenia, reproducible across sites and stages, using machine learning with structured sparsity” In: Acta Psychiatrica Scandinavica, Wiley, 2018, 2018, pp.1 - 10 (PDF).

  • A. de Pierrefeu, T. Fovet, F. Hadj-Selem, T. Löfstedt, P. Ciuciu, S. Lefebvre, P. Thomas, R. Lopes, R. Jardri, and E. Duchesnay. “Prediction of activation patterns preceding hallucinations in patients with schizophrenia using machine learning with structured sparsity“. Human Brain Mapping, Wiley, 2018, 39 (4), pp.1777 - 1788 (PDF).

  • F. Hadj-Selem, T. Lofstedt, E. Dohmatob, V. Frouin, M. Dubois, V. Guillemot, and E. Duchesnay. ”Continuation of Nesterov’s Smoothing for Regression with Structured Sparsity in High-Dimensional Neuroimaging“. IEEE Transactions on Medical Imaging (PDF) (April 2018) and supplementary.

  • A. de Pierrefeu, T. Lofstedt, F. Hadj-Selem, M. Dubois, R. Jardri, T. Fovet, P. Ciuciu, V. Frouin, and E. Duchesnay. “Structured Sparse Principal Components Analysis With the TV-Elastic Net Penalty“. IEEE Transactions on Medical Imaging, 2018, 37 (2), pp.396 - 407 (PDF)).

  • E. Jouvent, E. Duchesnay, F. Hadj-Selem, F. De Guio, J.-F. Mangin, D. Hervé, M. Duering, S. Ropele, R. Schmidt, M. Dichgans, et al. “Prediction of 3-year clinical course in CADASIL“. In: Neurology 87.17 (2016), pp. 1787–1795 (PDF).


PCT/FR2010/050431: Inventors: Duchesnay, Edouard; Paillere, Marie-Laure; Cachia, Arnaud; Martinot, Jean-Luc; Artiges, Eric. ”Method for Developing an Information Prediction Device, Use Thereof, and Corresponding Storage Medium and Apparatus“.


I wrote a course on Machine Learning in Python which is available as (Jupyter notebooks) on github or as pdf. This course is used for three Masters:

  • 2015-Now (35h) - Machine learning in Master 2 - Innovation, marché et science des données (IMSD), Paris-Saclay University. Academic head: Maria-Eugenia Sanin.

  • 2016-2017 (10h) - Data analysis in Master 1 - Mathématiques et applications - option : “Ingénierie mathématique pour les sciences du vivant“, Paris Descartes University. Academic head: Etienne Birmele.

  • 2017-Now (15h) - Biostatistics 3rd year of CentralSupelec, Paris-Saclay University. Academic head: Arthur Tenenhaus.

Supervision experience

  • 2017-now Pauline Favre, Post-doc, together with JF Mangin and J Houenou.

  • 2016-now Amicie de Pierrefeu, PhD.

  • 2016 Pietro Gori, Post-doc together with JF Mangin and J Houenou. P. Gori is now Assistant Professor at Télécom ParisTech, Paris, France.

  • 2013-2015 Fouad Hadj Selem, Post-doc, F. Hadj Selem is now research scientist at the Energy Transition Institute: VeDeCoM, France.

  • 2013-2015 Tommy Lofstedt, Post-doc. T. Lofstedt is now research scientist in Umea University, Sweden.

  • 2013-2014 Mathieu Dubois, Post-doc, M. Dubois is now research engineer at CEA Genoscope, Evry, France.

  • 2014, Clémence Pinaud, Engineer trainee. C Pinaud is now engineer at Dreems, Paris.

  • 2013-2014 Jinpeng Li, Research Engineer. J. Li is now engineer at Qynapse Co., Paris.

  • 2009-2011 Cecilia Damon (PhD defended on 2011) together with JB Poline. C. Damon is now engineer at Qynapse Co., Paris.

  • 2008-2012, Edith Lefloch (PhD defended on 2012). E Lefloch is now research scientist at CEA, CNRGH Evry, France.


  • 2018-2023: R-LiNK (H2020-SC1-2017, 754907). Optimizing response to Li treatment through personalized evaluation of individuals with bipolar I disorder: the R-LiNK initiative. PI: F. Bellivier, WP leader: E Duchesnay and leader for the CEA, Team budget: 800k€.

  • 2014-2018: BIP-Li7 (ANR-14-CE15-0003). Therapeutic Lithium response in Bipolar Disorders and brain Lithium-7 NMR Spectroscopy Imaging at 7 Tesla. PI: F. Bellivier, WP leader:F Boumezbeur, Team budget: 280k€.

  • 2011-2015: MESCOG, (FP6 ERA-NET NEURON 01 EW1207). Mechanisms of Small Vessel Related Brain Damage and Cognitive Impairment: Integrating Imaging Findings from Genetic and Sporadic Disease. WP co-leader: E Duchesnay, Team budget: 195k€.

  • 2012-2016: BRAINOMICS (ANR-10-BINF-04). Methodological and software solutions for the integration of neuroimaging and genomic data. WP leader: E Duchesnay, Team budget: 800k€.

  • 2010-2013: Karamétria (ANR-09-BLAN-0332). A unified framework for feature-based morphometry of the brain. PI: E Duchesnay, Team budget: 200k€.

  • 2007-2010: AGIR (ANR-07-NEUR-0001). AGIR – Autism: Genetic and Imaging Research. WP leader: E Duchesnay, Team budget: 150k€.

  • 2007-Present: Contribution to the CATI platform which is a national platform created by the French Alzheimer plan in 2011 to support multicenter neuroimaging studies (9M€ grant).

Recent Posts