Medical imaging has now become an everyday part of hospital care systems, thus helping doctors look inside the human body without having to perform any surgery. However, as imaging technologies like CT, MRI, and PET have shown advancements, the amount of data produced has been growing enormously. Such type of images are now larger, more detailed, and also much harder to manage. This scenario has created a clear and urgent need for smarter ways to process and analyze them quickly without losing crucial and fine details that matter most in diagnosis. In this work, a complete framework that combined the methods of image fusion, image compression, and texture-based feature extraction to identify whether the tissue regions are benign or malignant has been proposed. The approach has been built on our earlier research work where images from multiple modalities were fused using an Undecimated Discrete Wavelet Transform (UDWT) and compressed with Context-Based Adaptive Binary Arithmetic Coding (CABAC) without loss of image quality. After the compression process, we extract texture features using an improved Gray Level Co-occurrence Matrix (GLCM) model along with deep learning-based descriptors so as to capture the subtle variations in the tissue texture. Experiments that we have were carried out on medical image datasets show that proposed system is able to maintain the strong diagnostic quality and it has achieved an accuracy of 94.3% even after image compression. The results have highlighted its potential for faster, more reliable diagnosis in telemedicine and smart healthcare environments.