Generative Adversarial Networks in Medical Imaging: A Scoping Review of Applications, Translational Readiness, and Evidence Gaps

This scoping review maps the recent applications of generative adversarial network (GAN)-based methods in medical imaging from 2021 to 2025. The aim was to provide an evidence map of recent original studies, focusing on methodological categories, imaging modalities, anatomical targets, application domains, translational-readiness evidence, and trustworthiness gaps. The scoping review was conducted and reported in accordance with the PRISMA extension for Scoping Reviews (PRISMA-ScR). PubMed, Scopus, and IEEE Xplore were searched, supplemented by manual screening of reference lists and related articles. Original full-text articles published in English between 2021 and 2025 were eligible if they used adversarially trained GAN-based methods for medical imaging tasks. A total of 112 studies were included after duplicate removal and two-stage screening. GAN-based methods were most often applied to disease diagnosis support, image reconstruction, image quality enhancement, cross-modality translation, anatomical segmentation, and training data expansion. Evidence reporting differed substantially across the five assessed domains. External validation was clearly reported in 17 studies (15.2%), prospective evaluation in 2 (1.8%), and clinical workflow integration in 3 (2.7%). Privacy assessment was not clearly reported in any study, with only one (0.9%) partially addressing this domain. In contrast, hallucination or failure analysis was clearly reported in 32 studies (28.6%) and partially reported in 74 (66.1%). These findings indicate widespread technical experimentation but uneven documentation of evidence relevant to clinical translation. Further external and prospective evaluation, explicit privacy assessment, more comprehensive failure analysis, and clinician-centered workflow studies are needed to support the responsible clinical implementation of GAN-based medical imaging methods.

Authors

Institutions

Publication Details

Journal
Journal of Imaging Informatics in Medicine
Published
2026-11-17
DOI
https://doi.org/10.1007/s10278-026-02544-4
Primary Topic
Artificial Intelligence in Healthcare and Education
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Generative Adversarial Networks in Medical Imaging: A Scoping Review of Applications, Translational Readiness, and Evidence Gaps

Se‐woon Choe, Woo Young Jang, Subrata Bhattacharjee, Jangsun Hwang
2 links
Journal of Imaging Informatics in Medicine
Artificial Intelligence in Healthcare and Education
article

Generative Adversarial Networks in Medical Imaging: A Scoping Review of Applications, Translational Readiness, and Evidence Gaps

Se‐woon Choe, Woo Young Jang, Subrata Bhattacharjee, Jangsun Hwang
article en

Abstract

This scoping review maps the recent applications of generative adversarial network (GAN)-based methods in medical imaging from 2021 to 2025. The aim was to provide an evidence map of recent original studies, focusing on methodological categories, imaging modalities, anatomical targets, application domains, translational-readiness evidence, and trustworthiness gaps. The scoping review was conducted and reported in accordance with the PRISMA extension for Scoping Reviews (PRISMA-ScR). PubMed, Scopus, and IEEE Xplore were searched, supplemented by manual screening of reference lists and related articles. Original full-text articles published in English between 2021 and 2025 were eligible if they used adversarially trained GAN-based methods for medical imaging tasks. A total of 112 studies were included after duplicate removal and two-stage screening. GAN-based methods were most often applied to disease diagnosis support, image reconstruction, image quality enhancement, cross-modality translation, anatomical segmentation, and training data expansion. Evidence reporting differed substantially across the five assessed domains. External validation was clearly reported in 17 studies (15.2%), prospective evaluation in 2 (1.8%), and clinical workflow integration in 3 (2.7%). Privacy assessment was not clearly reported in any study, with only one (0.9%) partially addressing this domain. In contrast, hallucination or failure analysis was clearly reported in 32 studies (28.6%) and partially reported in 74 (66.1%). These findings indicate widespread technical experimentation but uneven documentation of evidence relevant to clinical translation. Further external and prospective evaluation, explicit privacy assessment, more comprehensive failure analysis, and clinician-centered workflow studies are needed to support the responsible clinical implementation of GAN-based medical imaging methods.

Journal of Imaging Informatics in Medicine
Kumoh National Institute of Technology (KR), Korea University (KR)
Good health and well-being
Artificial Intelligence in Healthcare and Education
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon