Classifying sex with volume-matched brain MRI
Sex differences in the size of specific brain structures have been extensively studied, but careful and reproducible statistical hypothesis testing to identify them produced overall small effect sizes and differences in brains of males and females. On the other hand, multivariate statistical or mach...
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Elsevier
2023-09-01
|
Series: | Neuroimage: Reports |
Subjects: | |
Online Access: | http://www.sciencedirect.com/science/article/pii/S2666956023000260 |
_version_ | 1797683006782046208 |
---|---|
author | Matthis Ebel Martin Domin Nicola Neumann Carsten Oliver Schmidt Martin Lotze Mario Stanke |
author_facet | Matthis Ebel Martin Domin Nicola Neumann Carsten Oliver Schmidt Martin Lotze Mario Stanke |
author_sort | Matthis Ebel |
collection | DOAJ |
description | Sex differences in the size of specific brain structures have been extensively studied, but careful and reproducible statistical hypothesis testing to identify them produced overall small effect sizes and differences in brains of males and females. On the other hand, multivariate statistical or machine learning methods that analyze MR images of the whole brain have reported respectable accuracies for the task of distinguishing brains of males from brains of females. However, most existing studies lacked a careful control for brain volume differences between sexes and, if done, their accuracy often declined to 70% or below. This raises questions about the relevance of accuracies achieved without careful control of overall volume.We examined how accurately sex can be classified from gray matter properties of the human brain when matching on overall brain volume. We tested, how robust machine learning classifiers are when predicting cross-cohort, i.e. when they are used on a different cohort than they were trained on. Furthermore, we studied how their accuracy depends on the size of the training set and attempted to identify brain regions relevant for successful classification. MRI data was used from two population-based data sets of 3298 mostly older adults from the Study of Health in Pomerania (SHIP) and 399 mostly younger adults from the Human Connectome Project (HCP), respectively. We benchmarked two multivariate methods, logistic regression and a 3D convolutional neural network.We show that male and female brains of the same intracranial volume can be distinguished with >92% accuracy with logistic regression on a dataset of 1166 matched individuals. The same model also reached 85% accuracy on a different cohort without retraining. The accuracy for both methods increased with the training cohort size up to and beyond 3000 individuals, suggesting that classifiers trained on smaller cohorts likely have an accuracy disadvantage. We found no single outstanding brain region necessary for successful classification, but important features appear rather distributed across the brain. |
first_indexed | 2024-03-12T00:08:10Z |
format | Article |
id | doaj.art-6dae452243b44f279b185cfcae400c15 |
institution | Directory Open Access Journal |
issn | 2666-9560 |
language | English |
last_indexed | 2024-03-12T00:08:10Z |
publishDate | 2023-09-01 |
publisher | Elsevier |
record_format | Article |
series | Neuroimage: Reports |
spelling | doaj.art-6dae452243b44f279b185cfcae400c152023-09-16T05:31:54ZengElsevierNeuroimage: Reports2666-95602023-09-0133100181Classifying sex with volume-matched brain MRIMatthis Ebel0Martin Domin1Nicola Neumann2Carsten Oliver Schmidt3Martin Lotze4Mario Stanke5University of Greifswald, Institute of Mathematics and Computer Science, Greifswald, 17489, GermanyUniversity Medicine Greifswald, Functional Imaging, Institute of Diagnostic Radiology and Neuroradiology, Greifswald, 17489, GermanyUniversity Medicine Greifswald, Functional Imaging, Institute of Diagnostic Radiology and Neuroradiology, Greifswald, 17489, GermanyUniversity Medicine Greifswald, Institute for Community Medicine, Greifswald, 17475, GermanyUniversity Medicine Greifswald, Functional Imaging, Institute of Diagnostic Radiology and Neuroradiology, Greifswald, 17489, GermanyUniversity of Greifswald, Institute of Mathematics and Computer Science, Greifswald, 17489, Germany; Corresponding author.Sex differences in the size of specific brain structures have been extensively studied, but careful and reproducible statistical hypothesis testing to identify them produced overall small effect sizes and differences in brains of males and females. On the other hand, multivariate statistical or machine learning methods that analyze MR images of the whole brain have reported respectable accuracies for the task of distinguishing brains of males from brains of females. However, most existing studies lacked a careful control for brain volume differences between sexes and, if done, their accuracy often declined to 70% or below. This raises questions about the relevance of accuracies achieved without careful control of overall volume.We examined how accurately sex can be classified from gray matter properties of the human brain when matching on overall brain volume. We tested, how robust machine learning classifiers are when predicting cross-cohort, i.e. when they are used on a different cohort than they were trained on. Furthermore, we studied how their accuracy depends on the size of the training set and attempted to identify brain regions relevant for successful classification. MRI data was used from two population-based data sets of 3298 mostly older adults from the Study of Health in Pomerania (SHIP) and 399 mostly younger adults from the Human Connectome Project (HCP), respectively. We benchmarked two multivariate methods, logistic regression and a 3D convolutional neural network.We show that male and female brains of the same intracranial volume can be distinguished with >92% accuracy with logistic regression on a dataset of 1166 matched individuals. The same model also reached 85% accuracy on a different cohort without retraining. The accuracy for both methods increased with the training cohort size up to and beyond 3000 individuals, suggesting that classifiers trained on smaller cohorts likely have an accuracy disadvantage. We found no single outstanding brain region necessary for successful classification, but important features appear rather distributed across the brain.http://www.sciencedirect.com/science/article/pii/S2666956023000260Sex discriminationMachine learningConvolutional neural networkPopulation based dataVoxel based morphometry |
spellingShingle | Matthis Ebel Martin Domin Nicola Neumann Carsten Oliver Schmidt Martin Lotze Mario Stanke Classifying sex with volume-matched brain MRI Neuroimage: Reports Sex discrimination Machine learning Convolutional neural network Population based data Voxel based morphometry |
title | Classifying sex with volume-matched brain MRI |
title_full | Classifying sex with volume-matched brain MRI |
title_fullStr | Classifying sex with volume-matched brain MRI |
title_full_unstemmed | Classifying sex with volume-matched brain MRI |
title_short | Classifying sex with volume-matched brain MRI |
title_sort | classifying sex with volume matched brain mri |
topic | Sex discrimination Machine learning Convolutional neural network Population based data Voxel based morphometry |
url | http://www.sciencedirect.com/science/article/pii/S2666956023000260 |
work_keys_str_mv | AT matthisebel classifyingsexwithvolumematchedbrainmri AT martindomin classifyingsexwithvolumematchedbrainmri AT nicolaneumann classifyingsexwithvolumematchedbrainmri AT carstenoliverschmidt classifyingsexwithvolumematchedbrainmri AT martinlotze classifyingsexwithvolumematchedbrainmri AT mariostanke classifyingsexwithvolumematchedbrainmri |