DCT-based medical image compression using machine learning

Medical images need to be efficiently compressed before transmission and storage, due to the storage capacity and constrained bandwidth issues. An ideal image compression system must yield a high compression ratio with good quality compressed images. Machine learning models are implemented to perform tasks, whereas humans have difficulties in completing. For instance, an optimum compression ratio could be suggested considering the details on an X-ray image. In this paper, machine learning algorithms are trained to relate the medical image contents to their compression ratio. Once trained, the optimum DCT compression ratio of the X-ray images is chosen upon presenting an image to the network. Experimental results showed that the radial basis function neural network learning algorithm can be efficiently used to classify the optimum compression ratio for the X-ray images while maintaining high image quality. The radial basis function neural network learning algorithm can be efficiently used to classify optimum compression ratio, considering optimum compression deviation with various levels of accuracy. The experiments are done using two compression scenarios considering the ratio of training and testing. Two different scenarios are defined and discussed. When proposed scenario 1 is considered, gradient boosting algorithm and support vector machine achieved the highest recognition rate of 79.16%; however, radial basis function neural network achieved the highest recognition rate of 90.625%, whereas when proposed scenario 2 considered with an accuracy rate of 89% as optimum compression deviation 1 is noted.

Automated summary

Machine learning algorithms are trained to relate medical image contents to compression ratios, with the radial basis function neural network achieving the best classification results for optimal compression ratios for X-ray images.