Abstract:Objectives: To develop a deep learning-based harmonization framework, assessing whether it can improve performance of radiomics models given different kernels in different clinical tasks and additionally generalize to mitigate the effects of new/unobserved kernels on radiomics features. Methods: Patient data with 2 reconstruction kernels and phantom data with 22 reconstruction kernels were included. Eighty-five patients were studied for lymph node metastasis (LNM) prediction, and 164 patients for differential diagnosis between lung cancer (LC) and pulmonary tuberculosis (TB). Two convolutional neural network (CNN) models were developed to convert images (i) from B70f to B30f (CNNa) and (ii) from B30f to B70f (CNNb). Model performance between the two kernels was evaluated using AUC and compared with other well-known harmonization methods. Patient-normalized feature difference (PNFD) was used to identify the incompatible kernels (i.e., kernel with median PNFD > 1) with baseline (B30f/B70f), and measure the ability of the CNN models to convert the non-comparable kernels. Results: For LC versus pulmonary TB diagnosis, AUCs of CNNa vs. others were 0.85 vs. 0.54-0.74 (p = 0.0001-0.0003), and for CNNb vs. others: 0.87 vs. 0.54-0.86 (p = 0.0001-0.55). For LNM prediction, AUCs of CNNa vs. others were 0.68 vs. 0.56-0.61 (p = 0.10-0.39), and for CNNb vs. others: 0.78 vs. 0.70-0.73 (p = 0.07-0.40). After CNN harmonization, 17 of 20 (85%) of investigated unknown kernels produced comparable radiomics feature values relative to baseline (median PNFD from 1.10-2.31 to 0.23-1.13). Conclusion: The CNN harmonization effectively improved performance of radiomics models between reconstruction kernels in different clinical tasks, and reduced feature differences between unknown kernels vs. baseline. Key points: • The soft (B30f) and sharp (B70f) kernels strongly affect radiomics reproducibility and generalizability. • The convolutional neural network (CNN) harmonization methods performed better than location-scale (ComBat and centering-scaling) and matrix factorization harmonization methods (based on singular value decomposition (SVD) and independent component analysis (ICA)) in both clinical tasks. • The CNN harmonization methods improve feature reproducibility not only between specific kernels (B30f and B70f) from the same scanner, but also between unobserved kernels from different scanners of different vendors.

DeepResBat: deep residual batch harmonization accounting for covariate distribution differences

DeepComBat: A statistically motivated, hyperparameter-robust, deep learning approach to harmonization of neuroimaging data

Investigation of ComBat Harmonization on Radiomic and Deep Features from Multi-Center Abdominal MRI Data

ComBat models for harmonization of resting-state EEG features in multisite studies

ComBatLS: A location- and scale-preserving method for multi-site image harmonization

Empirical assessment of the assumptions of ComBat with diffusion tensor imaging

Harmonizing multisite neonatal diffusion-weighted brain MRI data for developmental neuroscience

Distributed Harmonization: Federated Clustered Batch Effect Adjustment and Generalization

Harmonization of Diffusion MRI-based Structural Connectomes for Multi-site Studies

Statistical harmonization corrects site effects in functional connectivity measurements from multi‐site fMRI data

Deep feature batch correction using ComBat for machine learning applications in computational pathology

Deep Causal Feature Extraction and Inference with Neuroimaging Genetic Data.

Bundle Analytics based Data Harmonization for Multi-Site Diffusion MRI Tractometry

Evaluation of Mean Shift, ComBat, and CycleGAN for Harmonizing Brain Connectivity Matrices Across Sites

Unpaired Volumetric Harmonization of Brain MRI with Conditional Latent Diffusion

Diffusion based multi-domain neuroimaging harmonization method with preservation of anatomical details

Harmonization of Structural Brain Connectivity through Distribution Matching

Harmonization techniques for machine learning studies using multi-site functional MRI data

SAN: Mitigating spatial covariance heterogeneity in cortical thickness data collected from multiple scanners or sites

A comparison of methods to harmonize cortical thickness measurements across scanners and sites

Deep learning-based harmonization of CT reconstruction kernels towards improved clinical task performance