A region-based Principal Component Analysis (PCA) technique for medical image compression
Principal Component Analysis (PCA) is capable of completely decorrelating input data in the transform domain. However, PCA is limited in image compression because there is a need to encode the eigenvectors of the input data and thereby affects the rate-distortion performance. In an effort to improve...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | http://eprints.utem.edu.my/id/eprint/26876/1/A%20region-based%20principal%20component%20analysis%20%28PCA%29%20technique%20for%20medical%20image%20compression.pdf http://eprints.utem.edu.my/id/eprint/26876/2/A%20region-based%20principal%20component%20analysis%20%28PCA%29%20technique%20for%20medical%20image%20compression.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Principal Component Analysis (PCA) is capable of completely decorrelating input data in the transform domain. However, PCA is limited in image compression because there is a need to encode the eigenvectors of the input data and thereby affects the rate-distortion performance. In an effort to improve rate-distortion performance, this work proposed a block-to-row PCA (BTRPCA) algorithm that employs the eigenvectors from the model image of the same image modality coupled with a row vectorization approach. Region-based compression schemes that reduce storage space while preserving the image quality of the region of interest (ROI) are receiving attention due to the increase in medical imaging data. While PCA is inherently limited by its matrix form, the Arbitrary ROI coding (ARC) proposed in this work models the ROI by means of a factorization approach and the arbitrary-shaped ROI contours and NROI are compressed using BTRPCA. In order to minimize user interaction, an automated brain segmentation technique based on midsagittal plane (MSP) and Absolute Difference Map (ADM) is then incorporated into the proposed Automated Arbitrary PCA (AAPCA). The presented result showed that BTRPCA achieves PSNR improvements of up to 10 dB compared to its PCA counterparts. The ARC outperforms JPEG, Embedded Zerotree Wavelet (EZW) and Embedded Block Coding With Optimized Truncation (EBCOT) at all tested bit rates with an average PSNR improvements of 6 dB, 18 dB and 12 dB respectively. Subjective performance analysis was in agreement with the objective performance analysis in which the AAPCA is capable of extending beyond the compression limits of the conventional PCA algorithm and that the quality of the surroundings of ROI is degrading gracefully at bpp as low as 0.25. The research has successfully developed an improved region-based compression scheme for medical images where lossy and lossless compression is implemented in one PCA architecture. Continuation of this study include using different encoding schemes to boost the rate-distortion performance and extraction of multiple ROI. |
|---|