Privacy Preserving Structurally Constrained GAN for Artifact Reduction in Liver MRIs

Raghul Sachin R; Anusha T

Privacy Preserving Structurally Constrained GAN for Artifact Reduction in Liver MRIs

Author(s)	Mr. Raghul Sachin R, Dr. Anusha T
Country	India
Abstract	Magnetic Resonance Imaging (MRI) is one of the major tools used for diagnosing the liver; nevertheless, liver images are prone to motion and breathing artifacts, thus resulting in low-quality images. Although Generative Adversarial Networks (GANs) have been used in restoring images, traditional GAN models suffer from generating hallucinations and changes in essential anatomical structures. In addition, using medical imaging data poses great privacy risks. This work presents the Privacy-Preserving Structurally Constrained GAN for artifact removal in 3D Liver MRIs. First, 2D slices are extracted from 3D NIfTI images. Then, an extremely structured generator is trained via Structural Similarity (SSIM) and perceptual loss functions in order to preserve liver parenchyma geometry and vascular anatomy. The model introduces the differentially private discrimination gradient as a privacy-preserving strategy to safeguard patient identities. The model can successfully eliminate artifacts while maintaining the image’s structure, according to experiments conducted on paired and unpaired artifact-filled NIfTI images. The framework effectively prevents membership inference attacks without significantly impairing restorative performance by mathematically bounding the impact of any single patient scan during the optimization phase. The ensuing high-fidelity restorations reduce the possibility of incorrect diagnosis or the need for expensive patient rescans by giving radiologists clear, diagnostically reliable cross-sections. In the end, this approach opens the door for safe implementation in actual clinical settings by bridging the crucial gap between sophisticated deep learning capabilities and stringent healthcare data regulations.
Keywords	Generative Adversarial Networks (GAN), Liver MRI, Artifact Reduction, Privacy-Preserving ML, NIfTI, Image Restoration, Structural Constraints.
Field	Computer > Artificial Intelligence / Simulation / Virtual Reality
Published In	Volume 8, Issue 5, September-October 2026
Published On	2026-09-30

View / Download PDF File

About IJFMR Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us		+91-9687-828-838	editor@ijfmr.com

International Journal For Multidisciplinary Research

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Bi-monthly Scholarly International Journal

Privacy Preserving Structurally Constrained GAN for Artifact Reduction in Liver MRIs

Share this