Optimization of medical image steganography using n-decomposition genetic algorithm
Protecting patients' confidential information is a critical concern in medical image steganography. The Least Significant Bits (LSB) technique has been widely used for secure communication. However, it is susceptible to imperceptibility and security risks due to the direct manipulation of pixel...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | eng eng |
| Published: |
2023
|
| Subjects: | |
| Online Access: | https://etd.uum.edu.my/10713/1/s902694_01.pdf https://etd.uum.edu.my/10713/2/s902694_02.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Protecting patients' confidential information is a critical concern in medical image steganography. The Least Significant Bits (LSB) technique has been widely used for secure communication. However, it is susceptible to imperceptibility and security risks due to the direct manipulation of pixels, and ASCII patterns present limitations. Consequently, sensitive medical information is subject to loss or alteration. Despite attempts to optimize LSB, these issues persist due to (1) the formulation of the optimization suffering from non-valid implicit constraints, causing inflexibility in reaching optimal embedding, (2) lacking convergence in the searching process, where the message length significantly affects the size of the solution space, and (3) issues of application customizability where different data require more flexibility in controlling the embedding process. To overcome these limitations, this study proposes a technique known as an n-decomposition genetic algorithm. This algorithm uses a variable-length search to identify the best location to embed the secret message by incorporating constraints to avoid local minimum traps. The methodology consists of five main phases: (1) initial investigation, (2) formulating an embedding scheme, (3) constructing a decomposition scheme, (4) integrating the schemes' design into the proposed technique, and (5) evaluating the proposed technique's performance based on parameters using medical datasets from kaggle.com. The proposed technique showed resistance to statistical analysis evaluated using Reversible Statistical (RS) analysis and histogram. It also demonstrated its superiority in imperceptibility and security measured by MSE and PSNR to Chest and Retina datasets (0.0557, 0.0550) and (60.6696, 60.7287), respectively. Still, compared to the results obtained by the proposed technique, the benchmark outperforms the Brain dataset due to the homogeneous nature of the images and the extensive black background. This research has contributed to genetic-based decomposition in medical image steganography and provides a technique that offers improved security without compromising efficiency and convergence. However, further validation is required to determine its effectiveness in real-world applications. |
|---|