HARP: HARmonizing in-vivo diffusion MRI using Phantom-only training

Hwihun Jeong; Qiang Liu; Kathryn E. Keenan; Elisabeth A. Wilde; Walter Schneider; Sudhir Pathak; Anthony Zuccolotto; Lauren J. O'Donnell

cs.CVMar 5, 2026

HARP: HARmonizing in-vivo diffusion MRI using Phantom-only training

Authors: Hwihun Jeong, Qiang Liu, Kathryn E. Keenan, Elisabeth A. Wilde, Walter Schneider, Sudhir Pathak, Anthony Zuccolotto, Lauren J. O'Donnell, +2 more

Organizations: Department of Psychiatry, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts, United States of America · College of Engineering, Northeastern University, Boston, Massachusetts, United States of America · Physical Measurement Laboratory, National Institute of Standards and Technology, Boulder, Colorado, United States of America · Department of Neurology, University of Utah School of Medicine, Salt Lake City, Utah, United States of America · George E. Wahlen Veterans Affairs Medical Center, Salt Lake City, Utah, United States of America · University of Pittsburgh, Pittsburgh, Pennsylvania , United States of America · Psychology Software Tools, Pittsburgh, Pennsylvania , United States of America · Department of Radiology, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts, United States of America · Harvard-MIT Health Sciences and Technology, Cambridge, Massachusetts, United States of America

Abstract

Purpose: Combining multi-site diffusion MRI (dMRI) data is hindered by inter-scanner variability, which confounds subsequent analysis. Previous harmonization methods require large, matched or traveling human subjects from multiple sites, which are impractical to acquire in many situations. This study aims to develop a deep learning-based dMRI harmonization framework that eliminates the reliance on multi-site in-vivo traveling human data for training. Methods: HARP employs a voxel-wise 1D neural network trained on an easily transportable diffusion phantom. The model learns relationships between spherical harmonics coefficients of different sites without memorizing spatial structures. Results: HARP reduced inter-scanner variability levels significantly in various measures. Quantitatively, it decreased inter-scanner variability as measured by standard error in FA (12%), MD (10%), and GFA (30%) with scan-rescan standard error as the baseline, while preserving fiber orientations and tractography after harmonization. Conclusion: We believe that HARP represents an important first step toward dMRI harmonization using only phantom data, thereby obviating the need for complex, matched in vivo multi-site cohorts. This phantom-only strategy substantially enhances the feasibility and scalability of quantitative dMRI for large-scale clinical studies.

Figures & tables

Figure 1: Conceptual overview of the HARP framework. (a) Comparison between conventional harmonization and the proposed HARP. Unlike traditional methods that necessitate the acquisition of matched or traveling human subject data from multiple sites, HARP utilizes easily accessible diffusion phantom data for network training. (b) Architectural overview of the 1D voxel-wise neural network. The network receives spherical harmonics coefficients from the source site as input and estimates harmonized spherical harmonics coefficients. This non-spatial architecture ensures robust generalizability when transferring the learned relationships from phantom training to in-vivo human inference.

(a) GE $\rightarrow$ Siemens
$b$ value (s/mm 2 )	Number of gradients	TE (ms)	TR (ms)	Slice thickness (mm)
1000	60	59.1	7500	1.5
1500	60	63.4	9200	1.5
2000	60	72.4	10400	1.5
1000	30	59.1	7000	1.5
2000	30	72.4	12000	1.5

Table 1: Diffusion phantom acquisition protocols for multi-site harmonization. The experimental configurations are illustrated for two distinct scenarios: (a) inter-vendor harmonization between a GE Premier (source site) and a Siemens Prisma (target site), and (b) inter-platform harmonization between a Siemens Skyra (source site) and a Siemens Prisma (target site).

Figure 2: Voxel-wise standard error maps for GE $\rightarrow$ Siemens harmonization. The spatial distribution of standard errors for fractional anisotropy (FA), mean diffusivity (MD), and generalized fractional anisotropy (GFA) is displayed. Columns 1 and 2 represent the scan-rescan standard error for the source and target sites, respectively. Prior to harmonization (Column 3), the inter-scanner standard error is notably high across all metrics. The application of HARP (Column 5), which is trained exclusively on phantom data, results in a substantial reduction of standard error, achieving performance closer to the in-vivo-trained LinearRISH baseline (Column 4).

		Scan-rescan standard error		Inter-scanner standard error
		Source	Target	No harmonization	LinearRISH	HARP
\multirow 3*GE $\rightarrow$ Siemens	FA ( $\times 10^{-3}$ )	42.2 $\pm$ 24.5	26.6 $\pm$ 15.9	46.5 (10%) $\pm$ 23.1	39.9 (-5%) $\pm$ 19.1	41.6 (-1%) $\pm$ 18.4
	MD ( $\times 10^{-6}$ )	60.6 $\pm$ 60.9	36.9 $\pm$ 36.5	65.7 (8%) $\pm$ 63.2	59.3 (-2%) $\pm$ 57.3	55.2 (-9%) $\pm$ 55.1
	GFA ( $\times 10^{-3}$ )	42.7 $\pm$ 26.5	23.5 $\pm$ 16.1	81.9 (92%) $\pm$ 37.9	34.4 (-19%) $\pm$ 20.0	56.9 (33%) $\pm$ 28.6
\multirow 2*Skyra $\rightarrow$ Prisma	FA ( $\times 10^{-3}$ )	28.4 $\pm$ 23.8	25.1 $\pm$ 14.0	44.4 (56%) $\pm$ 30.2	37.8 (33%) $\pm$ 20.3	40.3 (42%) $\pm$ 24.0
	MD ( $\times 10^{-6}$ )	48.0 $\pm$ 56.1	48.0 $\pm$ 48.4	80.7 (68%) $\pm$ 93.8	74.9 (56%) $\pm$ 93.1	77.0 (61%) $\pm$ 92.7

Table 2: Average voxel-wise standard error values for FA, MD, and GFA across three harmonization scenarios. The inter-scanner standard error is reported for the No harmonization data, LinearRISH, and the proposed HARP. Values in parentheses represent the percentage difference between inter-scanner standard error and source site scan-rescan standard error: (

\frac{\sigma_{e,Inter-scanner}-\sigma_{e,Scan-rescan}}{\sigma_{e,Scan-rescan}}

Figure 3: Box plots illustrating the standard errors for averaged FA, MD, and GFA values across selected white matter regions of interest. The variability is categorized into source site scan-rescan (red), target site scan-rescan (yellow), inter-scanner without harmonization (blue), inter-scanner with LinearRISH (green), and inter-scanner with HARP (purple). Statistical significance from paired t-tests is indicated by *, **, and *** for p-values less than 0.05, 0.01, and 0.001, respectively. HARP significantly reduces the standard error across most tracts, with the exception of FA in the GE $\rightarrow$ Siemens scenario and MD in the Skyra $\rightarrow$ Prisma scenario at b = 1000 s/mm 2 .

Figure 4: Box plots illustrating the standard errors for averaged FA, MD, and GFA values across selected gray matter regions of interest. The color coding follows the same scheme as in Figure 3 : source site scan-rescan (red), target site scan-rescan (yellow), inter-scanner without harmonization (blue), inter-scanner with LinearRISH (green), and inter-scanner with HARP (purple). Significance levels are denoted by *, **, and *** for $p<$ 0.05, 0.01, and 0.001, respectively. In gray matter regions, HARP consistently and significantly reduces the inter-scanner standard error across all cases, demonstrating robust harmonization that aligns with the performance of the in-vivo-trained baseline.

			Scan-rescan standard error		Inter-scanner standard error
			Source	Target	No harmonization	LinearRISH	HARP (proposed)
\multirow 6*GE $\rightarrow$ Siemens	\multirow 2*FA ( $\times 10^{-3}$ )	WM	3.33 $\pm$ 0.75	3.04 $\pm$ 0.76	12.2 $\pm$ 4.58	4.64 $\pm$ 0.92	11.0 $\pm$ 3.27
		GM	4.09 $\pm$ 1.37	2.21 $\pm$ 1.13	28.8 $\pm$ 3.55	7.31 $\pm$ 1.21	15.1 $\pm$ 2.56
	\multirow 2*MD ( $\times 10^{-6}$ )	WM	6.47 $\pm$ 1.77	2.90 $\pm$ 1.18	22.1 $\pm$ 8.03	8.07 $\pm$ 2.38	15.9 $\pm$ 5.90
		GM	12.1 $\pm$ 3.07	8.95 $\pm$ 4.61	29.9 $\pm$ 15.5	17.6 $\pm$ 7.44	19.7 $\pm$ 9.34
	\multirow 2*GFA ( $\times 10^{-3}$ )	WM	5.45 $\pm$ 2.87	3.16 $\pm$ 1.30	58.1 $\pm$ 10.0	6.73 $\pm$ 0.96	32.5 $\pm$ 8.28

Table 3: Quantitative summary of mean standard error values for FA, MD, and GFA averaged across white matter (WM) and gray matter (GM) regions of interest.

Figure 5: Visual comparison of (a) tractography (arcuate fasciculus and corticospinal tract) and (b) fiber orientation maps before (top) and after (bottom) HARP harmonization. The high consistency between rows confirms that the 1D voxel-wise framework preserves underlying fiber information.

	\multirow 2* GE $\rightarrow$ Siemens	Skyra $\rightarrow$ Prisma	Skyra $\rightarrow$ Prisma
		$b=1000$ s/mm 2	$b=2000$ s/mm 2
wDICE	0.872	0.874	0.910
Angular error	6.49 ∘	6.83 ∘	4.93 ∘

Table 4: Weighted DICE (wDICE) and angular error values comparing dMRI data before and after HARP harmonization. Across all scenarios, high wDICE scores and low angular errors demonstrate that HARP preserves the underlying fiber information.

Explore similar work

IHF-Harmony: Multi-Modality Magnetic Resonance Images Harmonization using Invertible Hierarchy Flow Model
Feb 25, 2026Pengli Zhu, Yitao Zhu, Haowen Pang +1Magnetic Resonance Imaging Unpaired Image-to-Image Translation
Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hong Kong · School of Integrated Circuits and Electronics, Beijing Institute of Technology, China · Mental Health Research Center, The Hong Kong Polytechnic University, Hong Kong +1
SA-CycleGAN-2.5D: Self-Attention CycleGAN with Tri-Planar Context for Multi-Site MRI Harmonization
Mar 17, 2026Ishrith Gowda, Chunwei LiuMagnetic Resonance Imaging Unsupervised Domain Adaptation
Department of Electrical Engineering and Computer Sciences University of California, Berkeley Berkeley, CA 94720 · Department of Computer Science Purdue University West Lafayette, IN 47907
Robust-ComBat: Mitigating Outlier Effects in Diffusion MRI Data Harmonization
Mar 18, 2026Yoan David, Pierre-Marc Jodoin, Alzheimer's Disease Neuroimaging Initiative +1Medical Image Analysis Diffusion MRI
VitaLab, Dep of Computer Science, University of Sherbrooke, Sherbrooke, QC, J1K 2R1, Canada · Imeka Solutions Inc., Sherbrooke, J1H 4A7, QC, Canada · A comprehensive list of authors and affiliations of this consortium appear at the end of the paper.

Feb 25, 2026J/K move · Enter open · S save

IHF-Harmony: Multi-Modality Magnetic Resonance Images Harmonization using Invertible Hierarchy Flow Model

Pengli Zhu, Yitao Zhu, Haowen Pang, Anqi Qiu

Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hong Kong · School of Integrated Circuits and Electronics, Beijing Institute of Technology, China · Mental Health Research Center, The Hong Kong Polytechnic University, Hong Kong+1

Magnetic Resonance Imaging Unpaired Image-to-Image Translation Normalizing Flows

Open paper AskPDF arXiv

Retrospective MRI harmonization is limited by poor scalability across modalities and reliance on traveling subject datasets. To address these challenges, we introduce IHF-Harmony, a unified invertible hierarchy flow framework for multi-modality harmonization using unpaired data. By decomposing the translation process into reversible feature transformations, IHF-Harmony guarantees bijective mapping and lossless reconstruction to prevent anatomical distortion. Specifically, an invertible hierarchy flow (IHF) performs hierarchical subtractive coupling to progressively remove artefact-related features, while an artefact-aware normalization (AAN) employs anatomy-fixed feature modulation to accurately transfer target characteristics. Combined with anatomy and artefact consistency loss objectives, IHF-Harmony achieves high-fidelity harmonization that retains source anatomy. Experiments across multiple MRI modalities demonstrate that IHF-Harmony outperforms existing methods in both anatomical fidelity and downstream task performance, facilitating robust harmonization for large-scale multi-site imaging studies. Code is available at https://github.com/Idea89560041/IHF-Harmony.

Abstract

Explore similar work

IHF-Harmony: Multi-Modality Magnetic Resonance Images Harmonization using Invertible Hierarchy Flow Model

SA-CycleGAN-2.5D: Self-Attention CycleGAN with Tri-Planar Context for Multi-Site MRI Harmonization

Robust-ComBat: Mitigating Outlier Effects in Diffusion MRI Data Harmonization