# Efficacy of MRI data harmonization in the age of machine learning. A multicenter study across 36 datasets

arXiv: https://arxiv.org/abs/2211.04125 · HTML: https://mlsanity.com/paper/2211.04125/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 3 of its 113 references that are in the map.
- First posted on arXiv 8 Nov 2022.

## Builds on (most cited first, 3 of 3)

- [XGBoost: A Scalable Tree Boosting System](https://mlsanity.com/paper/1603.02754/) (2016)
- [Detect and Correct Bias in Multi-Site Neuroimaging Datasets](https://mlsanity.com/paper/2002.05049/) (2020)
- [Inflation of test accuracy due to data leakage in deep learning-based classification of OCT images](https://mlsanity.com/paper/2202.12267/) (2022)

## Built on by (most cited first, 6 of 6)

- [Impact of Leakage on Data Harmonization in Machine Learning Pipelines in Class Imbalance Across Sites](https://mlsanity.com/paper/2410.19643/) (2024)
- [Skull-stripping induces shortcut learning in MRI-based Alzheimer's disease classification](https://mlsanity.com/paper/2501.15831/) (2025)
- [ComBAT Harmonization for diffusion MRI: Challenges and Best Practices](https://mlsanity.com/paper/2505.14722/) (2025)
- [Benchmarking External Generalization of SPD Matrix Learning for Resting-State fMRI Connectome Prediction](https://mlsanity.com/paper/2608.30418/) (2026)
- [A generalizable structural brain MRI foundation model built through dual-priority federated pretraining](https://mlsanity.com/paper/2609.27611/) (2026)
- [Frozen Site-Probe Accuracy Is Not a Standalone Harmonization Score: An Audit of Evaluation Practice for Multi-Site MRI](https://mlsanity.com/paper/2610.14140/) (2026)

## Abstract (arXiv)

Pooling publicly-available MRI data from multiple sites allows to assemble extensive groups of subjects, increase statistical power, and promote data reuse with machine learning techniques. The harmonization of multicenter data is necessary to reduce the confounding effect associated with non-biological sources of variability in the data. However, when applied to the entire dataset before machine learning, the harmonization leads to data leakage, because information outside the training set may affect model building, and potentially falsely overestimate performance. We propose a 1) measurement of the efficacy of data harmonization; 2) harmonizer transformer, i.e., an implementation of the ComBat harmonization allowing its encapsulation among the preprocessing steps of a machine learning pipeline, avoiding data leakage. We tested these tools using brain T1-weighted MRI data from 1740 healthy subjects acquired at 36 sites. After harmonization, the site effect was removed or reduced, and we showed the data leakage effect in predicting individual age from MRI data, highlighting that introducing the harmonizer transformer into a machine learning pipeline allows for avoiding data leakage.