TY - JOUR
T1 - Using structural MRI to identify bipolar disorders – 13 site machine learning study in 3020 individuals from the ENIGMA Bipolar Disorders Working Group
AU - Nunes, Abraham
AU - Schnack, Hugo G.
AU - Ching, Christopher R.K.
AU - Agartz, Ingrid
AU - Akudjedu, Theophilus N.
AU - Alda, Martin
AU - Alnæs, Dag
AU - Alonso-Lana, Silvia
AU - Bauer, Jochen
AU - Baune, Bernhard T.
AU - Bøen, Erlend
AU - Bonnin, Caterina del Mar
AU - Busatto, Geraldo F.
AU - Canales-Rodríguez, Erick J.
AU - Cannon, Dara M.
AU - Caseras, Xavier
AU - Chaim-Avancini, Tiffany M.
AU - Dannlowski, Udo
AU - Díaz-Zuluaga, Ana M.
AU - Dietsche, Bruno
AU - Doan, Nhat Trung
AU - Duchesnay, Edouard
AU - Elvsåshagen, Torbjørn
AU - Emden, Daniel
AU - Eyler, Lisa T.
AU - Fatjó-Vilas, Mar
AU - Favre, Pauline
AU - Foley, Sonya F.
AU - Fullerton, Janice M.
AU - Glahn, David C.
AU - Goikolea, Jose M.
AU - Grotegerd, Dominik
AU - Hahn, Tim
AU - Henry, Chantal
AU - Hibar, Derrek P.
AU - Houenou, Josselin
AU - Howells, Fleur M.
AU - Jahanshad, Neda
AU - Kaufmann, Tobias
AU - Kenney, Joanne
AU - Kircher, Tilo T.J.
AU - Krug, Axel
AU - Lagerberg, Trine V.
AU - Lenroot, Rhoshel K.
AU - López-Jaramillo, Carlos
AU - Machado-Vieira, Rodrigo
AU - Malt, Ulrik F.
AU - McDonald, Colm
AU - Mitchell, Philip B.
AU - van Haren, Neeltje E.M.
N1 - Publisher Copyright:
© 2018, The Author(s).
Copyright:
Copyright 2020 Elsevier B.V., All rights reserved.
PY - 2020/9
Y1 - 2020/9
N2 - Bipolar disorders (BDs) are among the leading causes of morbidity and disability. Objective biological markers, such as those based on brain imaging, could aid in clinical management of BD. Machine learning (ML) brings neuroimaging analyses to individual subject level and may potentially allow for their diagnostic use. However, fair and optimal application of ML requires large, multi-site datasets. We applied ML (support vector machines) to MRI data (regional cortical thickness, surface area, subcortical volumes) from 853 BD and 2167 control participants from 13 cohorts in the ENIGMA consortium. We attempted to differentiate BD from control participants, investigated different data handling strategies and studied the neuroimaging/clinical features most important for classification. Individual site accuracies ranged from 45.23% to 81.07%. Aggregate subject-level analyses yielded the highest accuracy (65.23%, 95% CI = 63.47–67.00, ROC-AUC = 71.49%, 95% CI = 69.39–73.59), followed by leave-one-site-out cross-validation (accuracy = 58.67%, 95% CI = 56.70–60.63). Meta-analysis of individual site accuracies did not provide above chance results. There was substantial agreement between the regions that contributed to identification of BD participants in the best performing site and in the aggregate dataset (Cohen’s Kappa = 0.83, 95% CI = 0.829–0.831). Treatment with anticonvulsants and age were associated with greater odds of correct classification. Although short of the 80% clinically relevant accuracy threshold, the results are promising and provide a fair and realistic estimate of classification performance, which can be achieved in a large, ecologically valid, multi-site sample of BD participants based on regional neurostructural measures. Furthermore, the significant classification in different samples was based on plausible and similar neuroanatomical features. Future multi-site studies should move towards sharing of raw/voxelwise neuroimaging data.
AB - Bipolar disorders (BDs) are among the leading causes of morbidity and disability. Objective biological markers, such as those based on brain imaging, could aid in clinical management of BD. Machine learning (ML) brings neuroimaging analyses to individual subject level and may potentially allow for their diagnostic use. However, fair and optimal application of ML requires large, multi-site datasets. We applied ML (support vector machines) to MRI data (regional cortical thickness, surface area, subcortical volumes) from 853 BD and 2167 control participants from 13 cohorts in the ENIGMA consortium. We attempted to differentiate BD from control participants, investigated different data handling strategies and studied the neuroimaging/clinical features most important for classification. Individual site accuracies ranged from 45.23% to 81.07%. Aggregate subject-level analyses yielded the highest accuracy (65.23%, 95% CI = 63.47–67.00, ROC-AUC = 71.49%, 95% CI = 69.39–73.59), followed by leave-one-site-out cross-validation (accuracy = 58.67%, 95% CI = 56.70–60.63). Meta-analysis of individual site accuracies did not provide above chance results. There was substantial agreement between the regions that contributed to identification of BD participants in the best performing site and in the aggregate dataset (Cohen’s Kappa = 0.83, 95% CI = 0.829–0.831). Treatment with anticonvulsants and age were associated with greater odds of correct classification. Although short of the 80% clinically relevant accuracy threshold, the results are promising and provide a fair and realistic estimate of classification performance, which can be achieved in a large, ecologically valid, multi-site sample of BD participants based on regional neurostructural measures. Furthermore, the significant classification in different samples was based on plausible and similar neuroanatomical features. Future multi-site studies should move towards sharing of raw/voxelwise neuroimaging data.
UR - http://www.scopus.com/inward/record.url?scp=85053320480&partnerID=8YFLogxK
U2 - 10.1038/s41380-018-0228-9
DO - 10.1038/s41380-018-0228-9
M3 - Article
C2 - 30171211
AN - SCOPUS:85053320480
SN - 1359-4184
VL - 25
SP - 2130
EP - 2143
JO - Molecular Psychiatry
JF - Molecular Psychiatry
IS - 9
ER -