Harmonization Of Brain Diffusion MRI: Concepts And Methods.pdf

This review article by Pinto et al. (2020) explores the challenges and solutions in harmonizing brain diffusion MRI data across different sites and time points, aiming to improve comparability and statistical power in multi-center studies.

Key Points:

- Inter- and Intra-Site Variability: Diffusion MRI data exhibits significant variability due to hardware differences (scanners, coils), reconstruction algorithms, acquisition settings, and image quality, hindering multi-site and longitudinal studies.

- Importance of Harmonization: Robust harmonization methods are crucial to reduce site-related variations, enabling reliable comparisons and joint analyses of diffusion data from diverse sources.

- Categorization of Methods: The authors classify harmonization techniques into two main groups:
1. Diffusion Parametric Map Harmonization (DPMH): Focuses on harmonizing calculated diffusion metrics like FA, MD, AD, RD, KA, MK, AK, and RK.
2. Diffusion Weighted Image Harmonization (DWIH): Aims to harmonize raw diffusion-weighted images.

- Challenges: Establishing a gold standard harmonization technique remains difficult due to the complexity of brain tissue and various contributing factors.

- Tools for Selection: The article provides two classification tools:
- Feature Table: Helps readers identify suitable harmonization methods based on data characteristics.
- Flowchart: Guides selection of appropriate techniques according to study requirements.

- Conclusion: Harmonization is essential for reproducible and reliable multi-site diffusion MRI studies, and the authors offer practical tools to aid researchers in choosing suitable harmonization methods.