Extended Applications of Compressed Sensing Algorithm in Biomedical Signal and Image Compression

With rapid development of real-time and dynamic applications, Compressed Sensing (CS) has been used for signal and image compression in the last decades. Storing the medical data and images remains a critical task for the health care sectors owing to the large storage needs. An extended applications of CS algorithm are suggested here for compression of biomedical signals and images for minimizing the storage space without compromising the quality. Electroencephalogram (EEG) signals and Digital Imaging and Communications in Medicine (DICOM) images are considered as medical signal and image as sample data for this proposed work. EEG signals using 16 different electrodes are collected from medical center and combined into a unique composite signal based on their statistical properties and then the composite signal is used as an input to the CS algorithm. The composite signal is converted into frequency domain for calculation of relative power in different frequency bands for identification of Alzheimer’s disease. An application of CS on medical DICOM image compression is also proposed in this paper. Large dimensional DICOM images are splitted into number of blocks and each block is compressed using CS. The simulation result shows that suggested techniques perform better than the industry standard compression algorithms, in terms of Compression Ratio (CR), Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM) and Reconstruction Time (RT) with reduced complexity of operation and storage requirements. Specifically, the proposed technique offers CR up to 50:1 in case of biomedical signal compression. Dicom Image compression using the suggested technique offers SSIM improvement approximately by 15%, PSNR improvement by 4% and reduction in RT by 94% than the standard CS-based compression.