Effectiveness of Reconstruction Methods in Compressive Sensing for Biomedical Images

Abstract

For the effectual storage and transmission of signal in telemedicine, compression of medical images is one of the indispensable operations. Acquisition speed is always an issue in medical images like magnetic resonance imaging and computed tomography images. Compressed sensing came up as an inkling that achieves sparse signal with under sampled Nyquist rate. Compressed sensing is always astounding because only few samples can perfectly recover the entire signal is indeed a big achievement. In this paper different performance parameters peak signal to noise ratio, compression ratio, structural similarity in dearer evaluated for medical images by reconstruction algorithms like basic pursuit (l₁), least square (l₂), or thogonal matching pursuit. From these recovery algorithms, it is pointed thatl₁norm minimization is most established convex optimization approach to achieve better quality image. Performance metric speak signal to noise ratio and root mean square error are observed at different measurement samples and it is seen that peak signal to noise ratio increases with increased measurement and root mean square error decreases.

Keywords:Compressed Sensing, Sparsity, Compression Ratio, Structural Similarity Index, Root Mean Square Error, Performance Root Mean Square Error Difference.