Vol. 17, no. 4, 2025
РусскийEnglish
MEDICAL PHYSICS
An Adaptive Multiscale Total Variation Filter for Effective Elimination of Salt and Pepper Noise in CT Images
Hussein M. Hussein
Ninevah University, https://uoninevah.edu.iq/en/
Mosul 41002, Iraq
Email: hussein.hussein@uoninevah.edu.iq
Shahad Mahgoob Nafl
University of Baghdad, https://en.uobaghdad.edu.iq/
Baghdad 10071, Iraq
Email: shahad.m@comed.uobaghdad.edu.iq
Abdullah Thair Al-obaidi
University of Diyala, http://www.uodiyala.edu.iq/
Baqubah, 32001, Diyala, Iraq
Email: abdullah.thair@uodiyala.edu.iq
Received January 03, 2025, peer-reviewed January 10, 2025, accepted january 13, 2025, published August 14, 2025
Abstract: The quality of computed tomography (CT) images is severely deteriorated by salt and pepper noise, which affects both automatic analysis and visual interpretation. In order to efficiently eliminate salt and pepper noise while maintaining delicate anatomical structures, this research suggests a New Multiscale Edge-Adaptive Total Variation (MEATV) technique. By combining edge-aware total variation filtering with multiscale image decomposition, the suggested framework improves impulse noise suppression at different resolutions. The NIH Chest CT dataset was subjected to extensive testing at different noise intensities (30%, 50%, 70%, and 90%). The results show that the suggested approach maintains a substantially shorter execution time while consistently achieving higher PSNR and lower MSE when compared to cutting-edge filtering, adaptive, and deep learning techniques. The approach is effective, scalable, and appropriate for clinical real-time applications.
Keywords: impulse noise reduction, computed tomography images, total variation filtering, multiscale image decomposition, real-time filtering
UDC 004.032.26, 621.397.3
RENSIT, 2025, 17(4):493-500e DOI: 10.17725/j.rensit.2025.17.493
Full-text electronic version of this article - web site http://en.rensit.ru/vypuski/article/689/17(4)493-500e.pdf