A brain magnetic resonance image compression technique using wavelet-based SPIHT algorithm and capsule autoencoder

Medical imaging plays a starring role in diagnosis and treatment. It provides clinically meaningful information and thus reduces uncertainty in diagnosis. E-services related to the health sector, like telehealth, telemedicine, etc., require the transmission and storage of these digital data to be cost-effective and reliable. This paper discusses a selective image compression model for brain magnetic resonance (MR) images to tackle the telehealth sector’s increasing bandwidth and storage requirements. The brain MR image undergoes a sequence of filtering operations as part of its preprocessing stage. Optimized Fuzzy C-Means clustering is utilized to segment the medically significant region of the pre-processed image from the background information. A Wavelet transform-based SPIHT algorithm compresses the medically significant part, and a capsule autoencoder-based compression technique is used for the background data. The suggested algorithm has a higher compression ratio (28.66) and a higher Peak Signal-To-Noise Ratio (PSNR) (43.6dB) than existing techniques.