Preclinical images are a fundamental tool for many areas of medical research including cancer detection and drug development. Usually, Micro Computed Tomography (MicroCT) images are acquired to evaluate skeleton malformations after a genetic alteration of an animal. Such images have very high resolution and need to be compressed to reduce hard disk space and transmission time. However, like other CT images, the acquisition process introduces noise in the resulting image, which prevents most modern coding systems, including JPEG2000, from obtaining good coding performance. In this work, we do not address noise filtering as such, but we focus on an air filtering approach - inspired by the biological nature of the captured MicroCT images - aimed to improve the coding performance. Extensive experimental tests using different images suggest that air filtering does not penalize the diagnosis process of experts (measured in Mean Opinion Score), nor the visual quality, while allowing substantial improvements in coding performance.