MicroCT Image Coding Based on Air Filtering

Preclinical imaging is a key enabling technology for medical research, such as in drug discovery, cancer detection, or osteoporosis screening. Europe is investing a significant amount of resources in a distributed phenotype study, carried out using Micro Computed Tomography (CT) images, which are acquired at very high resolutions to ease the detection of skeleton malformations. Unfortunately, like other CT images, MicroCT images commonly contain a notable amount of noise emitted by the acquisition device, which hinders the encoding of such images. In order to improve coding performance, we propose to use the Hounsfield Scale to establish a relationship between CT values and biological tissue together with an air-filtering approach, which modifies only samples located outside the biological area without penalizing the usefulness of the images to clinical experts, nor the visual perception of images. The proposed filter sets all samples having value lower than a specific threshold T to a constant figure.