A multi-scanner neuroimaging data harmonization using RAVEL and ComBat

Abstract

Modern neuroimaging studies frequently combine data collected from multiple scanners and experimental conditions. Such data often contain substantial technical variability associated with image intensity scale (image intensity scales are not the same in different images) and scanner effects (images obtained from different scanners contain substantial technical biases). Here we evaluate and compare results of data analysis methods without any data transformation (RAW), with intensity normalization using RAVEL, with regional harmonization methods using ComBat, and a combination of RAVEL and ComBat. Methods are evaluated on a unique sample of 16 study participants who were scanned on both 1.5T and 3T scanners a few months apart. Neuroradiological evaluation was conducted for 7 different regions of interest (ROI’s) pertinent to Alzheimer’s disease (AD). Cortical measures and results indicate that: (1) RAVEL substantially improved the reproducibility of image intensities; (2) ComBat is preferred over RAVEL and the RAVEL-ComBat combination in terms of regional level harmonization due to more consistent harmonization across subjects and image-derived measures; (3) RAVEL and ComBat substantially reduced bias compared to analysis of RAW images, but RAVEL also resulted in larger variance; and (4) the larger root mean square deviation (RMSD) of RAVEL compared to ComBat is due mainly to its larger variance.

Coordinated Cortical Thickness Alterations across Psychiatric Conditions: A Transdiagnostic ENIGMA Study

“…They thus contain comparatively sparse data points across the cortex and summarize data from broader areas that contain a mosaic of neurobiological regions that may be differentially affected by disease. Moreover, differences in parcel size (86), measurement error, subject motion and scanner/site effects (87,88) may slightly influence spatial covariance analyses.…”

Radiomic Signatures Associated with CD8+ Tumour-Infiltrating Lymphocytes: A Systematic Review and Quality Assessment Study

“…Image acquisition and post-processing standardisation strategies to improve the reproducibility and clinical translatability of radiomic features have been discussed extensively in review articles by Park et al and Vallières et al [ 76 , 77 ]. For instance, the batch effect correction method “ComBat” has recently been demonstrated to substantially reduce inter-scanner biases [ 78 , 79 ], thus allowing for the large-scale harmonisation and pooling of inhomogeneous cohorts [ 80 ]. Furthermore, efforts to simplify radiomics workflows, in particular by automating lesion segmentation and feature processing steps via deep learning, promise to minimise the effect of variable clinical practices on radiomic signatures [ 81 , 82 ].…”

MISPEL: A supervised deep learning harmonization method for multi-scanner neuroimaging data

“…RAVEL increased the similarity of images in their appearance/contrast, GM-WM contrast, and tissue type volumes and segmentation overlap when SPM framework was used. Larger variability and inconsistent behavior of RAVEL in harmonization of images across subjects was reported in (Torbati et al, 2021a), when RAVEL was used for harmonizing paired images of GE 1.5T and Siemens 3T scanners and FreeSurfer was used. Such inconsistency for WS-normalized and RAVEL-harmonized images was also observed as large SD values, when comparing volumetric differences based on FSL segmentation (Table 2 and Figure 8a).…”