Robust medical image compression based on wavelet transform and vector quantization

Abstract

Medical imaging, such as Magnetic Resonance Imaging (MRI), Computed Tomography (CT), and Ultrasound, serves as a precursor for determining a patient's expedition towards therapeutics or surgery. The rising pervasiveness of chronic illnesses worldwide has resulted in a great increase in the number of diagnostic imaging techniques being executed annually. This, in turn, has given rise to more advanced imaging technologies and software to expedite accurate diagnosis. For the purposes of patient medical history, these images are stored for very long periods. Also, future research and medical developments renders such records very sensitive, emphasizing their need to be stored. Storing, however, poses a great challenge since there is limited storage capacity to preserve these ever growing medical images. Any technology that improves medical image compression is welcome, since indirectly, it also promotes applications such as telemedicine that require fewer bits to transmit imagery over a computer network. In this study is proposed a DWT-VQ (Discrete Wavelet Transform – Vector Quantization) technique to compress images, and to preserve their perceptual quality at a medically tolerant level. In this hybrid technique, speckle and salt and pepper noises in ultrasound imagery are significantly reduced. If the image is not ultrasound, the process has a negligible effect, but the edge is preserved. The images are then filtered using DWT. A threshold approach is applied to generate coefficients by efficient means. The result is then vector quantized. Finally, the quantized coefficients are Huffman encoded. The resulting bits that represent the compressed image are then stored, and retrieved when needed. The result of the proposed technique is promising, as it outperforms other recent techniques.