Multi-Site MRI Data Harmonization with an Adversarial Learning Approach: Implementation to the Study of Brain Connectivity in Autism Spectrum Disorders

Magnetic resonance imaging (MRI) nowadays plays an important role in the identification of brain underpinnings in a wide range of neuropsychiatric disorders, including Autism Spectrum Disorders (ASD). Characterizing the hallmarks in these pathologies is not a straightforward task and machine learning (ML) is certainly one of the most promising tools for addressing complex and non-linear problems. ML algorithms and, in particular, deep neural networks (DNNs), need large datasets in order to be properly trained and thus ensure generalization capabilities on new data. Large datasets can be obtained by collecting images from different centers, thus bringing unavoidable biases in the analysis due to differences in hardware and scanning protocols between different centers. In this work, we dealt with the issue of multicenter MRI data harmonization by comparing two different approaches: the analytical ComBat-GAM procedure, whose effectiveness is already documented in the literature, and an originally developed site-adversarial deep neural network (ad-DNN). The latter aims to perform a classification task while simultaneously searching for site-relevant patterns in order to make predictions free from site-related biases. As a case study, we implemented DNN and ad-DNN classifiers to distinguish subjects with ASD with respect to typical developing controls based on functional connectivity measures derived from data of the multicenter ABIDE collection. The classification performance of the proposed ad-DNN, measured in terms of the area under the ROC curve (AUC), achieved the value of AUC = 0.70±0.03, which is comparable to that obtained by a DNN on data harmonized according to the analytical procedure (AUC = 0.71±0.01). The relevant functional connectivity alterations identified by both procedures showed an agreement between each other and with the patterns of neuroanatomical alterations previously detected in the same cohort of subjects.

[1]  Hugo A. Ferreira,et al.  Normative model detects abnormal functional connectivity in psychiatric disorders , 2023, Frontiers in Psychiatry.

[2]  C. Correll,et al.  Candidate diagnostic biomarkers for neurodevelopmental disorders in children and adolescents: a systematic review , 2023, World psychiatry : official journal of the World Psychiatric Association.

[3]  Li Kang,et al.  Autism spectrum disorder recognition based on multi-view ensemble learning with multi-site fMRI , 2022, Cognitive Neurodynamics.

[4]  R. Bellotti,et al.  Multi-site harmonization of MRI data uncovers machine-learning discrimination capability in barely separable populations: An example from the ABIDE dataset , 2022, NeuroImage: Clinical.

[5]  E. Fombonne,et al.  Global prevalence of autism: A systematic review update , 2022, Autism research : official journal of the International Society for Autism Research.

[6]  S. Calderoni Sex/gender differences in children with autism spectrum disorder: A brief overview on epidemiology, symptom profile, and neuroanatomy , 2022, Journal of neuroscience research.

[7]  D. Rangaprakash,et al.  Functional Connectivity-Based Prediction of Autism on Site Harmonized ABIDE Dataset , 2021, IEEE Transactions on Biomedical Engineering.

[8]  Dinggang Shen,et al.  Multi-site MRI harmonization via attention-guided deep domain adaptation for brain disorder identification , 2021, Medical Image Anal..

[9]  M. B. Nebel,et al.  Altered Inferior Parietal Functional Connectivity is Correlated with Praxis and Social Skill Performance in Children with Autism Spectrum Disorder. , 2020, Cerebral cortex.

[10]  Nicola Amoroso,et al.  Extensive Evaluation of Morphological Statistical Harmonization for Brain Age Prediction , 2020, Brain sciences.

[11]  Christos Davatzikos,et al.  Harmonization of large MRI datasets for the analysis of brain imaging patterns throughout the lifespan , 2019, NeuroImage.

[12]  Qingmao Hu,et al.  Specific Functional Connectivity Patterns of Middle Temporal Gyrus Subregions in Children and Adults with Autism Spectrum Disorder , 2019, Autism research : official journal of the International Society for Autism Research.

[13]  Piernicola Oliva,et al.  Evaluation of Altered Functional Connections in Male Children With Autism Spectrum Disorders on Multiple-Site Data Optimized With Machine Learning , 2019, Front. Psychiatry.

[14]  Rebecca C. Knickmeyer,et al.  ENIGMA and global neuroscience: A decade of large-scale studies of the brain in health and disease across more than 40 countries , 2019, Biological Psychiatry.

[15]  F. Pollick,et al.  Biological motion perception in autism spectrum disorder: a meta-analysis , 2019, Molecular Autism.

[16]  W. K. Lau,et al.  Resting-state abnormalities in Autism Spectrum Disorders: A meta-analysis , 2019, Scientific Reports.

[17]  S. Rose,et al.  A systematic review of structural MRI biomarkers in autism spectrum disorder: A machine learning perspective , 2018, International Journal of Developmental Neuroscience.

[18]  M. Weissman,et al.  Statistical harmonization corrects site effects in functional connectivity measurements from multi‐site fMRI data , 2018, Human brain mapping.

[19]  Jennifer Fedor,et al.  Cortical and subcortical brain morphometry differences between patients with autism spectrum disorders (ASD) and healthy individuals across the lifespan: results from the ENIGMA-ASD working group , 2017 .

[20]  Ragini Verma,et al.  Harmonization of multi-site diffusion tensor imaging data , 2017, NeuroImage.

[21]  Daniel P. Kennedy,et al.  Enhancing studies of the connectome in autism using the autism brain imaging data exchange II , 2017, Scientific Data.

[22]  Letitia R. Naigles,et al.  Language comprehension and brain function in individuals with an optimal outcome from autism , 2015, NeuroImage: Clinical.

[23]  François Laviolette,et al.  Domain-Adversarial Training of Neural Networks , 2015, J. Mach. Learn. Res..

[24]  Daniel P. Kennedy,et al.  The Autism Brain Imaging Data Exchange: Towards Large-Scale Evaluation of the Intrinsic Brain Architecture in Autism , 2013, Molecular Psychiatry.

[25]  Charles J. Lynch,et al.  Underconnectivity between voice-selective cortex and reward circuitry in children with autism , 2013, Proceedings of the National Academy of Sciences.

[26]  R. Schultz,et al.  The social motivation theory of autism , 2012, Trends in Cognitive Sciences.

[27]  M. Allard,et al.  The Integration of Prosodic Speech in High Functioning Autism: A Preliminary fMRI Study , 2010, PloS one.

[28]  R. Poldrack,et al.  Reward processing in autism , 2010, Autism research : official journal of the International Society for Autism Research.

[29]  Tom Fawcett,et al.  An introduction to ROC analysis , 2006, Pattern Recognit. Lett..

[30]  Ning Zhang,et al.  A Deep Neural Network Study of the ABIDE Repository on Autism Spectrum Classification , 2020 .

[31]  Cheng Li,et al.  Adjusting batch effects in microarray expression data using empirical Bayes methods. , 2007, Biostatistics.