Visual fidelity–driven quality assessment of medical image translation
- 1Faculty of Electrical Engineering, Laboratory of Imaging Technologies
Abstract
Automated and reliable image quality assessment (IQA) is essential for safe use of medical image synthesis in critical applications like adaptive radiotherapy, treatment planning, or missing-modality reconstruction, where unnoticed generative artifacts may adversely affect outcomes. We evaluated image-to-image translation quality by coupling large-scale visual quality assessment with explainable automated IQA modeling. Adversarial diffusion-based framework, SynDiff, was applied to four cross-modality synthesis tasks, including three inter-MR and a CBCT-to-CT translation. Using four-fold cross-validation, ten reference-based and eight no-reference IQA metrics were computed for all synthesized images. Visual IQA ratings were independently collected from thirteen raters using predetermined protocol and specialized image viewer enabling blinded, randomized six-point Likert scoring. Auto-Sklearn was employed to learn ensemble regression models mapping IQA metrics to visual consensus ratings, with separate models trained on reference-based and no-reference metrics. The models closely reproduced distribution and ordering of ratings, typically within ±0.5 Likert points. Reference-based models achieved higher agreement with visual ratings than no-reference models ( R2 0.75 vs. 0.59, resp.), although the latter remained unbiased and informative. Explainability analyses indicated that metrics quantifying structural similarity or fidelity (e.g., anatomical boundary preservation) and intensity-based contrast relationships between tissues were the strongest predictors. Overall, the results demonstrate that ensemble regression models can provide transparent, scalable, and clinically meaningful quality control for generative medical imaging.
Related
- Translation, Cultural Adaptation, and Preliminary Validation of the Persian Version of the Assessment of Medical Education Environment by Teachers (AMEET) Questionnaire
- Quality Assessment and Model Training Impact of Synthetic Images in Medical Diagnostic Imaging: Systematic Review.
- Medical AI: are Osler's principles lost in translation?
- Opportunistic computed tomography assessment of bone, muscle, and adipose tissue: evidence, limitations, and clinical translation
- Deep-Learning-Based Assessment of Breast Arterial Calcification on Mammography: Technical Validation, Cardiovascular Prognostic Value, and Clinical Translation—A Narrative Review
BibTeX
@article{bizjak2026visual,
title = {{Visual fidelity–driven quality assessment of medical image translation}},
author = {Žiga Bizjak and Jan Žagar and Žiga Špiclin},
journal = {Scientific Reports},
year = {2026},
month = jul,
doi = {10.1038/s41598-026-60570-2},
url = {https://doi.org/10.1038/s41598-026-60570-2},
note = {CC BY},
}
RIS
TY - JOUR
TI - Visual fidelity–driven quality assessment of medical image translation
AU - Žiga Bizjak
AU - Jan Žagar
AU - Žiga Špiclin
PY - 2026
DA - 2026/07/01
JO - Scientific Reports
DO - 10.1038/s41598-026-60570-2
UR - https://doi.org/10.1038/s41598-026-60570-2
AB - Automated and reliable image quality assessment (IQA) is essential for safe use of medical image synthesis in critical applications like adaptive radiotherapy, treatment planning, or missing-modality reconstruction, where unnoticed generative artifacts may adversely affect outcomes. We evaluated image-to-image translation quality by coupling large-scale visual quality assessment with explainable automated IQA modeling. Adversarial diffusion-based framework, SynDiff, was applied to four cross-modality synthesis tasks, including three inter-MR and a CBCT-to-CT translation. Using four-fold cross-validation, ten reference-based and eight no-reference IQA metrics were computed for all synthesized images. Visual IQA ratings were independently collected from thirteen raters using predetermined protocol and specialized image viewer enabling blinded, randomized six-point Likert scoring. Auto-Sklearn was employed to learn ensemble regression models mapping IQA metrics to visual consensus ratings, with separate models trained on reference-based and no-reference metrics. The models closely reproduced distribution and ordering of ratings, typically within ±0.5 Likert points. Reference-based models achieved higher agreement with visual ratings than no-reference models ( R^2 0.75 vs. 0.59, resp.), although the latter remained unbiased and informative. Explainability analyses indicated that metrics quantifying structural similarity or fidelity (e.g., anatomical boundary preservation) and intensity-based contrast relationships between tissues were the strongest predictors. Overall, the results demonstrate that ensemble regression models can provide transparent, scalable, and clinically meaningful quality control for generative medical imaging.
SN - 2045-2322
ER -