Lossless Medical Image Compression Using Integer Discrete Wavelet Transform With Adaptive Subband Differencing and Context-Adaptive Entropy Coding

Rasha F. Nadhim, College of Electromechanical Engineering, University of Technology-Iraq, Iraq, Baghdad
Ibrahim Adel Ibrahim, College of Artificial Intelligence Engineering, University of Technology-Iraq, Iraq, Baghdad
Ashwaq T. Hashim, College of Artificial Intelligence Engineering, University of Technology-Iraq, Iraq, BaghdadFollow

ORCID

Rasha F. Nadhim: https://orcid.org/0000-0002-2212-4624

Ashwaq T. Hashim: https://orcid.org/0000-0002-3282-6419

Article Type

Original Study

Abstract

From transform-domain decorrelation and adaptive entropy coding, we propose a method for efficient lossless compression of medical images in this work. We implement the Integer Discrete Wavelet Transform (IDWT) to decompose the input image into four subbands of LL, LH, HL, HH, encompassing approximation and directional detail elements, in the initial implementation. It also removes spatial redundancy in image information and decomposes image information into a less correlated and more sparsely distributed set of coefficients. To decrease redundancy further, it proposes a subband-dependent differencing scheme, which decorrelates neighbouring wavelet coefficients with directional prediction. So horizontal differencing is done on LH, vertical differencing on HL, and also horizontal or vertical differencing on LL and HH subbands. This reduces inter-coefficient correlation significantly, and in many instances, particularly in high frequency subbands, tends to move the distribution of the coefficients toward zero mean with a corresponding dip in entropy. Finally, context-adaptive entropy coding (CAEC) is utilized to encode the coefficients obtained. CAEC enables a higher coding efficiency and a better compression thanks to personalisation of the symbol probabilities towards surrounding of the location. By applying IDWT and subband differencing in combination, sparsity is provided and statistical dependency on common pixel-domain methods is substantially decreased. The main novelty is basically in how we ve moved away from these rigid, same-type pipelines by using a specialized scheme that depends on the subband, and it does directional differencing so prediction vectors can follow wavelet geometric routes. it kind of lines everything up in a smoother way, instead of staying uniform or flat. The proposed method attained the highest compression ratios across all benchmark images, ranging from 2.64 to 6.36, with a maximum value of 6.3566 on House.bmp and consistently outperforming Lossless JPEG (1.26–2.08), JPEG-LS (0.72–1.86), PNG (0.92–1.18), TIFF (1.08–1.41). Furthermore, the proposed framework allows exact reconstruction of medical images along with better compression ratios of images into an ideal image for high-precision purposes such as medical archiving and telemedicine.

Keywords

Lossless compression, Medical imaging, Integer Discrete Wavelet Transform (IDWT), Coefficient differencing, context-adaptive entropy coding (CAEC), Data compression, Transform-domain processing

How to Cite This Article

Nadhim, Rasha F.; Ibrahim, Ibrahim Adel; and Hashim, Ashwaq T. (2026) "Lossless Medical Image Compression Using Integer Discrete Wavelet Transform With Adaptive Subband Differencing and Context-Adaptive Entropy Coding," Journal of Intelligent Informatics, Networking, and Cybersecurity: Vol. 2 : Iss. 2 , Article 3.
Available at: https://doi.org/10.65445/3106-1192.1014