Towards an informational model for tuberculosis lesion discrimination on X-ray CT images

摘要

Proper characterization of Tuberculosis (TB) as a continuous spectrum between latent and active stages is urgently needed as part of the efforts to control this devastating pandemic. Radiological imaging has revealed as an indispensable tool for this purpose and has raised the need to automate the identification of TB lesions and the characterization of disease progression on X-ray computer tomography (CT) scans of appropriate animal models. The contribution of the present work is to introduce a methodology for the above-mentioned task. The TB lesion detection is achieved using statistical region merging and their characterization is based on the extraction of texture features and the optimization of a random forest estimator. Our results demonstrate that the proposed methodology selects a simple but powerful model that achieves a proper classification of abnormal tissue.