NeuroImage (Dec 2022)
Sample size requirement for achieving multisite harmonization using structural brain MRI features
- Pravesh Parekh,
- Gaurav Vivek Bhalerao,
- John P. John,
- G. Venkatasubramanian,
- Biju Viswanath,
- Naren P. Rao,
- Janardhanan C. Narayanaswamy,
- Palanimuthu T. Sivakumar,
- Arun Kandasamy,
- Muralidharan Kesavan,
- Urvakhsh Meherwan Mehta,
- Odity Mukherjee,
- Meera Purushottam,
- Bhupesh Mehta,
- Thennarasu Kandavel,
- B. Binukumar,
- Jitender Saini,
- Deepak Jayarajan,
- A. Shyamsundar,
- Sydney Moirangthem,
- K.G. Vijay Kumar,
- Jayant Mahadevan,
- Bharath Holla,
- Jagadisha Thirthalli,
- Bangalore N. Gangadhar,
- Pratima Murthy,
- Mitradas M. Panicker,
- Upinder S. Bhalla,
- Sumantra Chattarji,
- Vivek Benegal,
- Mathew Varghese,
- Janardhan Y.C. Reddy,
- Padinjat Raghu,
- Mahendra Rao,
- Sanjeev Jain
Affiliations
- Pravesh Parekh
- NORMENT, Division of Mental Health and Addiction, Oslo University Hospital & Institute of Clinical Medicine, University of Oslo, Oslo, Norway; ADBS Neuroimaging Centre, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, India; Department of Psychiatry, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, India
- Gaurav Vivek Bhalerao
- Translational Psychiatry Lab, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, India; ADBS Neuroimaging Centre, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, India; Department of Psychiatry, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, India; Department of Psychiatry, University of Oxford, United Kingdom
- John P. John
- NORMENT, Division of Mental Health and Addiction, Oslo University Hospital & Institute of Clinical Medicine, University of Oslo, Oslo, Norway; ADBS Neuroimaging Centre, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, India; Department of Psychiatry, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, India; Corresponding authors: Dr. John P. John, Multimodal Brain Image Analysis Laboratory, Department of Psychiatry, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, 560029, India.
- G. Venkatasubramanian
- Translational Psychiatry Lab, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, India; ADBS Neuroimaging Centre, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, India; Department of Psychiatry, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, India; Corresponding authors: Dr. Ganesan Venkatasubramanian, Translational Psychiatry Lab, Department of Psychiatry, National Institute of Mental Health and Neurosciences (NIMHANS), Bangalore, 560029, India.
- Biju Viswanath
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Naren P. Rao
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Janardhanan C. Narayanaswamy
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Palanimuthu T. Sivakumar
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Arun Kandasamy
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Muralidharan Kesavan
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Urvakhsh Meherwan Mehta
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Odity Mukherjee
- Institute for Stem Cell Biology and Regenerative Medicine (InStem)
- Meera Purushottam
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Bhupesh Mehta
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Thennarasu Kandavel
- National Institute of Mental Health and Neurosciences (NIMHANS)
- B. Binukumar
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Jitender Saini
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Deepak Jayarajan
- National Institute of Mental Health and Neurosciences (NIMHANS)
- A. Shyamsundar
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Sydney Moirangthem
- National Institute of Mental Health and Neurosciences (NIMHANS)
- K.G. Vijay Kumar
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Jayant Mahadevan
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Bharath Holla
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Jagadisha Thirthalli
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Bangalore N. Gangadhar
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Pratima Murthy
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Mitradas M. Panicker
- National Center for Biological Sciences (NCBS)
- Upinder S. Bhalla
- National Center for Biological Sciences (NCBS)
- Sumantra Chattarji
- National Center for Biological Sciences (NCBS)
- Vivek Benegal
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Mathew Varghese
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Janardhan Y.C. Reddy
- National Institute of Mental Health and Neurosciences (NIMHANS)
- Padinjat Raghu
- National Center for Biological Sciences (NCBS)
- Mahendra Rao
- National Center for Biological Sciences (NCBS)
- Sanjeev Jain
- National Center for Biological Sciences (NCBS)
- Journal volume & issue
-
Vol. 264
p. 119768
Abstract
When data is pooled across multiple sites, the extracted features are confounded by site effects. Harmonization methods attempt to correct these site effects while preserving the biological variability within the features. However, little is known about the sample size requirement for effectively learning the harmonization parameters and their relationship with the increasing number of sites. In this study, we performed experiments to find the minimum sample size required to achieve multisite harmonization (using neuroHarmonize) using volumetric and surface features by leveraging the concept of learning curves. Our first two experiments show that site-effects are effectively removed in a univariate and multivariate manner; however, it is essential to regress the effect of covariates from the harmonized data additionally. Our following two experiments with actual and simulated data showed that the minimum sample size required for achieving harmonization grows with the increasing average Mahalanobis distances between the sites and their reference distribution. We conclude by positing a general framework to understand the site effects using the Mahalanobis distance. Further, we provide insights on the various factors in a cross-validation design to achieve optimal inter-site harmonization.