学术文献库

Compressing and expanding (companding) HDR computerized tomography (CT) images to a single LDR image is still a relevant challenge. Besides the general need for simplification of the window setting method for the purpose of diagnosis, there are specific clinical needs for the resection of malignant lesions in the mandible, for example, where the lesions may exist in both the soft tissue and the bone. A successful medical algorithm has to take into account the requirement to variously expose the same range of specific gray levels, when they are presented by different body tissues. As a solution, we propose an adaptive multi-scale contrast companding (AMCC) algorithm that is implemented by using soft threshold separation to two channels (bone and soft tissue). This separation is determined by the HU intensity values (that are already computed in the algorithm) in the specific resolution that is best fitted to the teeth. Each channel contains different set of parameters. The AMCC algorithm successfully and adaptively compands a large variety of mandibular CT HDR images as well as natural images. Two collaborating physicians who evaluated tumor boundaries, by the 'single presentation' method, reported that 92% of the algorithm output images were at least as useful for diagnosis as those in the window method, while 50% of the algorithm output images were better. When each slice was evaluated simultaneously, by the window setting method (bone and soft tissue) and algorithm output images, 93% of the evaluations declared a preference for the output images of the algorithm. We describe here a low-cost method for companding the HDR images with the ability to facilitate resections of mandibular lesions by providing optimal boundary definition.