Machine Learning for Large-Scale Quality Control of 3D Shape Models in Neuroimaging

TL;DR

This paper develops machine learning models to automate quality control of 3D brain shape models in neuroimaging, significantly reducing human effort while maintaining high accuracy across diverse datasets.

Contribution

It introduces a scalable machine learning approach using shape features and classifiers to predict MRI data quality, generalizing across multiple cohorts and diseases.

Findings

Models reduce human workload by 30-70%.

High recall rates approaching inter-rater reliability.

Effective across diverse datasets and conditions.

Abstract

As very large studies of complex neuroimaging phenotypes become more common, human quality assessment of MRI-derived data remains one of the last major bottlenecks. Few attempts have so far been made to address this issue with machine learning. In this work, we optimize predictive models of quality for meshes representing deep brain structure shapes. We use standard vertex-wise and global shape features computed homologously across 19 cohorts and over 7500 human-rated subjects, training kernelized Support Vector Machine and Gradient Boosted Decision Trees classifiers to detect meshes of failing quality. Our models generalize across datasets and diseases, reducing human workload by 30-70\%, or equivalently hundreds of human rater hours for datasets of comparable size, with recall rates approaching inter-rater reliability.