Owing to a large amount of images, image compression is requisite for minimizing the redundancies in image, and it offers efficient transmission and archiving of images. This paper presents a novel medical image compression model using intelligent techniques. The adopted medical image compression comprises of three major steps such as, Segmentation, Image compression, and Image decompression. Initially, the Region of Interest (ROI) and Non-ROI regions of the image are split by means of a Segmentation procedure using Modified Region Growing (MRG) algorithm. Moreover, the image compression process begins which is varied for both ROI and Non-ROI regions. On considering the ROI regions, the compression is carried out by Discrete Cosine Transform (DCT) model and SPIHT encoding method, whereas the compression of Non-ROI region is carried out by Discrete Wavelet Transform (DWT) and Merge-based Huffman encoding (MHE) methods. As a main contribution, this paper intends to deploy the optimized filter coefficients in both DCT and DWT techniques. Here, the optimization of both filter coefficients is performed using Modified Rider Optimization Algorithm (ROA) called Improvised Steering angle and Gear-based ROA (ISG-ROA). In the final step, decompression is done by implementing the reverse concept of compression process with similar optimized coefficients. The filter coefficients are tuned in such a way that the Compression Ratio (CR) should be minimum. In addition, the comparative analysis over the state-of-the-art models proves the superior performance of the proposed model.