Computer Vision News - January 2017

32 Computer Vision News Research Research Computer assisted detection of abnormal airway variation in CT scans related to paediatric tuberculosis: LA-PDM Every month, Computer Vision News reviews a research from our field. This month we have chosen to review Computer assisted detection of abnormal airway variation in CT scans related to paediatric tuberculosis , a research paper presenting a local airway point distribution model ( LA-PDM ) to automatically analyze regions of the airway tree in CT scans, in order to identify abnormal airway deformation. We are indebted to the authors ( Benjamin J. Irving, Pierre Goussard, Savvas Andronikou, Robert Gie, Tania S. Douglas, Andrew Todd- Pokropek and Paul Taylor ) for giving us unpublished images to illustrate this review and precious unpublished data to help review it. The full paper is here . Background: Aim and Motivation - Childhood tuberculosis is still prevalent in developing countries and it represents a large proportion of TB cases. However, for physiological reasons pediatric TB is more difficult to diagnose and therefore a combination of factors is used to diagnose it. One of the key indicators of the disease is lymph nodes involvement. Bronchoscopy is the ‘‘gold standard’’ for determining airway involvement but it is invasive; moreover, general anesthesia is often required and the external cause of the airway involvement cannot be seen. Challenge - Recent studies suggest that CT with volume rendering can be used as an alternative to bronchoscopy. As a consequence, there is considerable value in automatically detecting deformation of the airways caused by lymph-node distention (lymphadenopathy) to assist in the detection and assessment of pediatric TB cases (and potentially of other diseases). Novelty - Airway analysis from CT scans can help determine abnormal compression or deformation of the airways and this is key to detecting disease and visualizing affected airway regions. This paper introduces a method called LA-PDM (Local Airway Point Distribution Model) which identifies abnormal airway deformation. The LA-PDM presents a high accuracy differentiate airway involvement (pediatric pulmonary TB) from normal airways by examining regions of the airway likely to be affected by lymphadenopathy. Method: LA-PDM model has four key main steps (1) Segmentation and branch labelling; (2) Surface point projection; (3) Thin-plate-spline warp; (4) Airway shape features. Follows a detailed description of each of these four steps: 1. Segmentation and branch labelling - The airway tree is segmented from each CT scan and each branch is labelled. The segmentation is based on the algorithm of Lo et al. 2012. Next, the centerline is extracted using the Palagyi et al. 2006 method, iteratively removing voxels from the segmentation as long as they do