Adaptive Compression of Medical Images through Self Organizing Feature Map

摘要

Raw Medical images are large in size and required to be compressed before storage or transmission. This paper proposes an adaptive technique for compressing medical images through self organizing feature map. Here, the input image is split into small subblocks and are fed to the input layer of SOFM Neural Network. The network is trained to produce codebook whose length depends upon the number of neurons in the output layer. The codebook forms the representation of the image and is used for reconstruction of the original image. Obtained results are analyzed and compared with JPEG - the widely used standard for medical images.