Identification of Paediatric Tuberculosis from Airway Shape Features

  • Benjamin Irving
  • Pierre Goussard
  • Robert Gie
  • Andrew Todd-Pokropek
  • Paul Taylor
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6893)


Clinical signs of paediatric pulmonary tuberculosis (TB) include stenosis and deformation of the airways. This paper presents two methods to analyse airway shape and detect airway pathology from CT images. Features were extracted using (1) the principal components of the airway surface mesh and (2) branch radius and orientation features. These methods were applied to a dataset of 61 TB and non-TB paediatric patients. Nested cross-validation of the support vector classifier found the sensitivity of detecting TB to be 86% and a specificity of 91% for the first 10 PCA modes while radius based features had a sensitivity of 86% and a specificity of 94%. These methods show the potential of computer assisted detection of TB and other airway pathology from airway shape deformation.


Landmark Point Leave Main Bronchus Paediatric Tuberculosis Point Distribution Model Airway Pathology 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    Andronikou, S., Wieselthaler, N.: Modern imaging of tuberculosis in children: thoracic, central nervous system and abdominal tuberculosis. Pediatr. Radiol. 34, 861–875 (2004)CrossRefGoogle Scholar
  2. 2.
    Bookstein, F.: Principal warps: Thin-plate splines and the decomposition of deformations. IEEE Trans. Pattern Anal. Mach. Intell. 11, 567–585 (1989)CrossRefzbMATHGoogle Scholar
  3. 3.
    Cootes, T., Taylor, C., Cooper, D., Graham, J., et al.: Active shape models-their training and application. Comput. Vis. Image. Und. 61, 38–59 (1995)CrossRefGoogle Scholar
  4. 4.
    Deligianni, F., Chung, A., Yang, G.: Nonrigid 2-D/3-D registration for patient specific bronchoscopy simulation with statistical shape modeling: Phantom validation. IEEE Trans. Med. Imag. 25, 1462–1471 (2006)CrossRefGoogle Scholar
  5. 5.
    Hutton, T., Buxton, B., Hammond, P., Potts, H.: Estimating average growth trajectories in shape-space using kernel smoothing. IEEE Trans. Med. Imag. 22, 747–753 (2003)CrossRefGoogle Scholar
  6. 6.
    Irving, B., Taylor, P., Todd-Pokropek, A.: 3D segmentation of the airway tree using a morphology based method. In: Second International Workshop on Pulmonary Image Analysis, MICCAI (2009)Google Scholar
  7. 7.
    Kaus, M., Pekar, V., Lorenz, C., Truyen, R., Lobregt, S., Weese, J.: Automated 3-D PDM construction from segmented images using deformable models. IEEE Trans. Med. Imag. 22, 1005–1013 (2003)CrossRefGoogle Scholar
  8. 8.
    Masters, I., Ware, R., Zimmerman, P., Lovell, B., Wootton, R., Francis, P., Chang, A.: Airway sizes and proportions in children quantified by a video-bronchoscopic technique. BMC Pulmonary Medicine 6, 5–13 (2006)CrossRefGoogle Scholar
  9. 9.
    Palágyi, K., Tschirren, J., Hoffman, E., Sonka, M.: Quantitative analysis of pulmonary airway tree structures. Comput. Biol. Med. 36, 974–996 (2006)CrossRefGoogle Scholar
  10. 10.
    du Plessis, J., Goussard, P., Andronikou, S., Gie, R., George, R.: Comparing three-dimensional volume-rendered CT images with fibreoptic tracheobronchoscopy in the evaluation of airway compression caused by tuberculous lymphadenopathy in children. Pediatr. Radiol. 39, 694–702 (2009)CrossRefGoogle Scholar
  11. 11.
    Varma, S., Simon, R.: Bias in error estimation when using cross-validation for model selection. BMC Bioinformatics 7, 1–8 (2006)CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Benjamin Irving
    • 1
  • Pierre Goussard
    • 2
  • Robert Gie
    • 2
  • Andrew Todd-Pokropek
    • 1
  • Paul Taylor
    • 1
  1. 1.University College LondonLondonUK
  2. 2.Tygerberg HospitalStellenbosch UniversityWestern CapeSouth Africa

Personalised recommendations