Assessing and tuning brain decoders: cross-validation, caveats, and guidelines

TL;DR

This paper reviews cross-validation methods for brain decoder evaluation, highlighting the limitations of leave-one-out, advocating for repeated random splits, and providing guidelines for tuning decoders in neuroimaging.

Contribution

It offers a comprehensive review and empirical analysis of cross-validation procedures, proposing best practices for decoding in neuroimaging.

Findings

Leave-one-out cross-validation is unstable and biased.

Repeated random splits improve estimation stability.

Nested cross-validation helps tune parameters without bias.

Abstract

Decoding, ie prediction from brain images or signals, calls for empirical evaluation of its predictive power. Such evaluation is achieved via cross-validation, a method also used to tune decoders' hyper-parameters. This paper is a review on cross-validation procedures for decoding in neuroimaging. It includes a didactic overview of the relevant theoretical considerations. Practical aspects are highlighted with an extensive empirical study of the common decoders in within-and across-subject predictions, on multiple datasets --anatomical and functional MRI and MEG-- and simulations. Theory and experiments outline that the popular " leave-one-out " strategy leads to unstable and biased estimates, and a repeated random splits method should be preferred. Experiments outline the large error bars of cross-validation in neuroimaging settings: typical confidence intervals of 10%. Nested cross-validation can tune decoders' parameters while avoiding circularity bias. However we find that it can be more favorable to use sane defaults, in particular for non-sparse decoders.