Non-uniform Compression of Magnetic Resonance Brain Images Using Edge-based Active Contours Driven by Maximum Entropy Threshold

As digital imaging data are growing exponentially, compression of medical images is a critical issue for efficient storage and reliable transmission. As a result, researchers are continuously exploring methods for reducing the size of medical images further.

To further improve the compression methods, this paper proposes the maximum entropy-based threshold incorporated into the edge-based active contour method to automate the initialization of the curve for accurate extraction of the diagnostic or pathologically significant parts from unevenly illuminated Magnetic Resonance (MR) brain images. The images are then segmented into informative and background parts, which are further subjected to high-bit-rate and low-bit-rate compressions, respectively. This non-uniform compression results in an improvement in compression rate while preserving the quality of the diagnostic parts of the images.

The evaluation was performed on the dataset of MR brain images, and empirical analysis confirmed that the proposed method is able to outperform other existing methods in terms of segmentation and compression metrics.

The mathematical results of the proposed method indicate that the extracted area of the informative parts was similar to the object of interest in ground truth images. This accurate demarcation of the informative parts results in an improvement in compression rate without compromising the quality of the informative parts.