Zero-Reference Structural–Frequency Constrained Deep Learning Framework for Diagnostic-Safe Contrast Enhancement of Clinical Medical Images

Priyal Chaturvedi; Saurabh Srivastava

Zero-Reference Structural–Frequency Constrained Deep Learning Framework for Diagnostic-Safe Contrast Enhancement of Clinical Medical Images

Author(s)	Priyal Chaturvedi, Prof. Saurabh Srivastava
Country	India
Abstract	We introduce SFC-ZRNet, a zero-reference deep learning framework designed specifically for safe and reliable contrast enhancement in medical imaging. Unlike traditional histogram-based techniques, which may unintentionally alter anatomical structures, our method is developed with a strong emphasis on preserving diagnostic integrity. The framework explicitly measures and restricts structural distortion, ensuring that critical anatomical boundaries and lesion characteristics remain intact during enhancement. The proposed network is built on a multi-scale convolutional neural architecture and is guided by carefully designed objective functions. These include a structural distortion loss that combines edge-preservation and boundary-displacement constraints, as well as a modality-aware spectral loss to maintain frequency consistency across different imaging types. This design allows the model to enhance contrast while minimizing structural alterations that could compromise clinical interpretation.Extensive evaluation on ultrasound, MRI, CT, and X-ray datasets demonstrates that SFC-ZRNet consistently improves image quality. The method achieves higher PSNR, SSIM, and entropy values compared to conventional approaches, indicating superior contrast and fidelity. Importantly, it records the lowest Boundary Displacement Measure (BDM) among the compared methods, with approximately a 30% reduction relative to CLAHE, confirming improved structural preservation. The model is computationally efficient, requiring approximately 31 milliseconds to process a 256×256 image on a GPU, and contains only around 2.9 million parameters, making it lightweight and suitable for real-world clinical deployment. As a zero-reference framework, SFC-ZRNet eliminates the need for paired ground-truth enhanced images, enabling practical implementation across diverse clinical imaging protocols without additional annotation requirements.
Keywords	SFC-ZRNet, CNN, ultrasound, MRI, CT, X-ray, SSIM, PSNR, and CLAHE.
Field	Computer Applications
Published In	Volume 7, Issue 1, January-February 2026
Published On	2026-02-04

View / Download PDF File

E-ISSN 3048-7641

doi

CrossRef DOI is assigned to each research paper published in our journal.

AIJFR DOI prefix is
10.63363/aijfr

Downloads

Research Paper Format Copyright Permission Form and Undertaking Form Cover Page Vol 7 Isu 1 Cover Page Vol 6 Isu 6 Cover Page Vol 6 Isu 5

All research papers published on this website are licensed under Creative Commons Attribution-ShareAlike 4.0 International License, and all rights belong to their respective authors/researchers.

CC-BY-SA

About AIJFR 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		+91-9898-79-39-59	editor@aijfr.com

Advanced International Journal for Research

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

Zero-Reference Structural–Frequency Constrained Deep Learning Framework for Diagnostic-Safe Contrast Enhancement of Clinical Medical Images

Share this