Towards an Automatic Estimation of Skeletal Age Using \(k-NN\) Regression with a Reduced Set of Tinny Aligned Regions of Interest

  • José Luis Tonatiúh Banda-Escobar
  • Salvador E. Ayala-RaggiEmail author
  • Aldrin Barreto-Flores
  • Susana Sánchez-Urrieta
  • José Francisco Portillo-Robledo
  • Alinne Michelle Sánchez-Tomay
  • Verónica Edith Bautista-López
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10633)


Human skeletal maturity has been typically estimated from radiographic images of the non-dominant hand through a subjective analysis performed by expert radiologists. In this paper we present a semiautomatic learning approach for estimating bone age. We consider five regions of interest, shortly ROIs, located between metacarpal and phalanges, which are obtained by placing strategic landmarks. ROI images are reshaped in the form of vectors which are merged in order to generate aligned feature vectors of each hand. The method consists of two stages, training and testing, for which radiographic images of female gender were used in a range of 1 to 18 years old. The training stage focuses on structuring the feature vectors of 300 bone-age-labeled images to generate a set of prototypes for a regression classifier. The second step is to approximate the bone age of a novel testing image, by computing its respective feature vector and comparing it with the set of prototypes. The age was determined using regression through a weighted \(k-NN\) classifier. By using a set of 100 testing images, we demonstrate that it is possible to obtain an error comparable with state of the art algorithms by using only five small ROIs within the hand image.


Skeletal maturity recognition Bone age estimation \(k-NN\) regression ROI alignment 


  1. 1.
    Greulich, W., Pyle, S.: Radiographic Atlas of Skeletal Development of Hand and Wrist, 2nd edn. Standford University Press, Palo Alto (1971)Google Scholar
  2. 2.
    Tanner, J., Whitehouse, R., Cameron, N., Marshall, W., Healy, M., Goldstein, H.: Maturity and Prediction of Adult Height (TW2 Method), 2nd edn. Academic Press, London (1975)Google Scholar
  3. 3.
    Molinari, L., Gasser, T., Largo, R.: TW3 bone age: RUS/CB and gender differences of percentiles for score and score increments. Ann. Hum. Biol. 31(4), 421–435 (2004)CrossRefGoogle Scholar
  4. 4.
    Adeshina, S.A., Cootes, T.F., Adams, J.E.: Evaluating different structures for predicting skeletal maturity using statistical appearance models. In: Proceedings of the MIUA (2009)Google Scholar
  5. 5.
    Aja-Fernández, S., de Luis-Garcia, R., Martin-Fernandez, M.A., Alberola-López, C.: A computational TW3 classifier for skeletal maturity assessment. A computing with words approach. J. Biomed. Inf. 37, 99–107 (2004)CrossRefGoogle Scholar
  6. 6.
    Cunha, P., Moura, D.C., López, M.A.G., Guerra, C., Pinto, D., Ramos, I.: Impact of ensemble learning in the assessment of skeletal maturity. J. Med. Syst. 38, 87 (2014)CrossRefGoogle Scholar
  7. 7.
    Liu, H., et al.: Bone age pre-estimation using partial least squares regression analysis with a priori knowledge. In: 2014 IEEE International Symposium on Medical Measurements and Applications, MeMeA 2014, Lisboa, Portugal, 11–12 June 2014, pp. 164–167 (2014)Google Scholar
  8. 8.
    Niemeijer, M., van Ginneken, B., Maas, C., Beek, F., Viergever, M.: Assessing the skeletal age from a hand radiograph: automating the tanner-whitehouse method. In: Sonka, M., Fitzpatrick, J. (eds.) SPIE Medical Imaging, vol. 5032, pp. 1197–1205. SPIE, Bellingham (2003)Google Scholar
  9. 9.
    Hsieh, C.W., Jong, T.L., Chou, Y.H., Tiu, C.M.: Computerized geometric features of carpal bone for bone age estimation. Chin. Med. J. 120(9), 767–770 (2007)Google Scholar
  10. 10.
    Giordano, D., Kavasidis, I., Spampinato, C.: Modeling skeletal bone development with hidden markov models. Comput. Methods Programs Biomed. 124, 138–147 (2016)CrossRefGoogle Scholar
  11. 11.
    Spampinato, C., Palazzo, S., Giordano, D., Aldinucci, M., Leonardi, R.: Deep learning for automated skeletal bone age assessment in x-ray images. Med. Image Anal. 36, 41–51 (2017)CrossRefGoogle Scholar
  12. 12.
    Gertych, A., Zhang, A., Sayre, J., Pospiech-Kurkowska, S., Huang, H.: Bone age assessment of children using a digital hand atlas. Comput. Med. Imaging Graph. 31, 322–331 (2007). Computer-aided Diagnosis (CAD) and Image-guided Decision SupportCrossRefGoogle Scholar
  13. 13.
    Ayala-Raggi, S., Montoya, F., Barreto-Flores, A., Sánchez-Urrieta, S., Portillo-Robledo, J., Bautista-López, V.: A supervised incremental learning technique for automatic recognition of the skeletal maturity, or can a machine learn to assess bone age without radiological training from experts? Int. J. Pattern Recogn. Artif. Intell. (2017)Google Scholar
  14. 14.
    Cootes, T.F., Edwards, G.J., Taylor, C.J.: Active appearance models. IEEE Trans. Pattern Anal. Mach. Intell. 23, 681–685 (2001)CrossRefGoogle Scholar
  15. 15.
    Gilsanz, V., Ratib, O.: Hand Bone Age: A Digital Atlas Of Skeletal Maturity. Springer, Heidelberg (2005)Google Scholar
  16. 16.
    Kashif, M., Deserno, T.M., Haak, D., Jonas, S.: Feature description with SIFT, SURF, BRIEF, BRISK, or FREAK? A general question answered for bone age assessment. Comput. Biol. Med. 68, 67–75 (2016)CrossRefGoogle Scholar
  17. 17.
    Gonzalez, R.C., Woods, R.E.: Digital Image Processing, 3rd edn. Prentice-Hall Inc., Upper Saddle River (2006)Google Scholar

Copyright information

© Springer Nature Switzerland AG 2018

Authors and Affiliations

  • José Luis Tonatiúh Banda-Escobar
    • 1
  • Salvador E. Ayala-Raggi
    • 1
    Email author
  • Aldrin Barreto-Flores
    • 1
  • Susana Sánchez-Urrieta
    • 1
  • José Francisco Portillo-Robledo
    • 1
  • Alinne Michelle Sánchez-Tomay
    • 1
  • Verónica Edith Bautista-López
    • 2
  1. 1.Facultad de Ciencias de la ElectrónicaBenemérita Universidad Autónoma de PueblaPueblaMexico
  2. 2.Facultad de Ciencias de la ComputaciónBenemérita Universidad Autónoma de PueblaPueblaMexico

Personalised recommendations