International Journal of Legal Medicine

, Volume 133, Issue 4, pp 1191–1205 | Cite as

Automated segmentation of the knee for age assessment in 3D MR images using convolutional neural networks

  • Paul-Louis Pröve
  • Eilin Jopp-van Well
  • Ben Stanczus
  • Michael M. Morlock
  • Jochen Herrmann
  • Michael Groth
  • Dennis Säring
  • Markus Auf der MauerEmail author
Original Article


Age assessment is used to estimate the chronological age of an individual who lacks legal documentation. Recent studies indicate that the ossification degree of the growth plates in the knee joint correlates with chronological age of adolescents and young adults. To verify this hypothesis, a high number of datasets need to be analysed. An approach which enables an automated detection and analysis of the bone structures may be necessary to handle large datasets. The purpose of this study was to develop a fully automatic 2D knee segmentation based on 3D MR images using convolutional neural networks. A total of 76 datasets were available and divided into a training set (74%), a validation set (13%) and a test set (13%). Multiple preprocessing steps were applied to correct image intensity values and to reduce the image size. Image augmentation was employed to virtually increase the dataset size for training. The proposed architecture for the segmentation task resembles the encoder-decoder model type used for the U-Net. The trained network achieved a dice similarity coefficient score of 98% compared to the manual segmentations and an intersection over union of 96%. The precision and recall of the model were balanced, and the error was only 1.2%. No overfitting was observed during training. As a proof of concept, the predicted segmentations were used for the age estimation of 145 subjects. Initial results show the potential of this approach attaining a mean absolute error of 0.48 ± 0.32 years for a test set of 14 subjects. The proposed automated segmentation can contribute to faster, reproducible and potentially more reliable age estimation in the future.


Age estimation MRI Knee Segmentation Convolutional neural networks 


Funding information

This project is funded by the German Research Foundation (DFG), Project (SA 2530/6-1) and (JO 1198/2-1).

Compliance with Ethical Standards

An ethical approval for this study was granted by the Ethics Committee of the Medical Association Hamburg (PV4527). The director of this study had full control of the data and the material submitted for publication.

Conflict of interests

The authors declare that they have no conflict of interest.

Consent for Publication

Written informed consent was obtained from all subjects in this study.


  1. 1.
    Britting-Reimer E (2015) Altersbestimmung in Deutschland und im Europäischen VergleichGoogle Scholar
  2. 2.
    Geserick G, Schmeling A (2011) Qualitätssicherung der Forensischen Altersdiagnostik bei Lebenden Personen. Rechtsmedizin 21(1):22–25CrossRefGoogle Scholar
  3. 3.
    Kubilay S (2016) Ablauf des deutschen Asylverfahrens, tech. rep. Bundesamt für Migration und Flüchtlinge (BAMF)Google Scholar
  4. 4.
    European Asylum Support Office (2013) Age assessment practice in EuropeGoogle Scholar
  5. 5.
    World Health Organization (2016) Ionizing radiation, health effects and protective measuresGoogle Scholar
  6. 6.
    Dedouit F, Auriol J, Rousseau H, Rougé D, Crubézy E, Telmon N (2012) Age assessment by magnetic resonance imaging of the knee: a preliminary study. Forensic Sci Int 217(1-3):232CrossRefGoogle Scholar
  7. 7.
    Jopp E, Schröder I, Maas R, Adam G, Püschel K (2010) Proximale Tibiaepiphyse im Magnetresonanztomogramm: Neue Möglichkeit zur Altersbestimmung bei Lebenden?. Rechtsmedizin 20:464–468CrossRefGoogle Scholar
  8. 8.
    Jopp E (2013) Die Abschlussphase des menschlichen Wachstums Longitudinale Ganzkörper- und Unterschenkelmessungen (Knemometrie) an jungen Erwachsenen zur Bestimmung des biologischen Alters und für forensische Zwecke. PhD thesisGoogle Scholar
  9. 9.
    Krämer JA, Schmidt S, Jürgens KU, Lentschig M, Schmeling A, Vieth V (2014) Forensic age estimation in living individuals using 3.0T MRI of the distal femur. Int J Legal Med 128(3):509–514CrossRefGoogle Scholar
  10. 10.
    Laor T, Chun GFH, Dardzinski BJ, Bean JA, Witte DP (2002) Posterior distal femoral and proximal tibial metaphyseal stripes at mr imaging in children and young adults. Radiology 224:669–674CrossRefGoogle Scholar
  11. 11.
    Ekizoglu O, Hocaoglu E, Inci E, Can IO, Aksoy S, Kazimoglu C (2016) Forensic age estimation via 3-T magnetic resonance imaging of ossification of the proximal tibial and distal femoral epiphyses: Use of a T2-weighted fast spin-echo technique, vol 260Google Scholar
  12. 12.
    Fan F, Zhang K, Peng Z, Hui Cui J-h, Hu N, Hua Deng Z-h (2016) Forensic age estimation of living persons from the knee: comparison of MRI with radiographs. Forensic Sci Int 268:145–150CrossRefGoogle Scholar
  13. 13.
    Ottow C, Schulz R, Pfeiffer H, Heindel W, Schmeling A, Vieth V (2017) Forensic age estimation by magnetic resonance imaging of the knee: the definite relevance in bony fusion of the distal femoral- and the proximal tibial epiphyses using closest-to-bone T1 TSE sequence, European Radiology, pp 1–8Google Scholar
  14. 14.
    Jiang J, Trundle P, Ren J (2010) Medical image analysis with artificial neural networks. Comput Med Imaging Graph 34:617–631CrossRefGoogle Scholar
  15. 15.
    Litjens G, Kooi T, Bejnordi BE, Setio AAA, Ciompi F, Ghafoorian M, van der Laak JAWM, van Ginneken B, Sánchez CI (2017) A survey on deep learning in medical image analysis. Med Image Anal 42:60–88CrossRefGoogle Scholar
  16. 16.
    Setiono R, Liu H (1997) Neural-network feature selector. IEEE Trans Neural Netw 8:654–662CrossRefGoogle Scholar
  17. 17.
    Dodin P, Martel-Pelletier J, Pelletier J-PP, Abram F (2011) A fully automated human knee 3D MRI bone segmentation using the ray casting technique. Med Biol Eng Comput 49(12):1413–1424CrossRefGoogle Scholar
  18. 18.
    Dam EB, Lillholm M, Marques J, Nielsen M (2015) Automatic segmentation of high- and low-field knee MRIs using knee image quantification with data from the osteoarthritis initiative, vol 2Google Scholar
  19. 19.
    Ronneberger O, Fischer P, Brox T (2015) U-Net: convolutional networks for biomedical image segmentation, vol 9351Google Scholar
  20. 20.
    Stern D, Ebner T, Bischof H, Grassegger S, Ehammer T, Urschler M (2014) Fully automatic bone age estimation from left hand MR images. Medical Image Computing and Computer-Assisted Intervention - MICCAI 2014 117(2):220–227Google Scholar
  21. 21.
    Stern D, Urschler M (2016) From individual hand bone age estimates to fully automated age estimation via learning-based information fusionGoogle Scholar
  22. 22.
    Stern D, Kainz P, Payer C, Urschler M (2017) Multi-factorial age estimation from skeletal and dental MRI volumes. In: Wang Q, Shi Y, Suk H-I, Suzuki K (eds) Machine learning in medical imaging. Springer International Publishing, Cham, pp 61–69Google Scholar
  23. 23.
    Auf der Mauer M, Säring D, Stanczus B, Herrmann J, Groth M, Jopp-van Well E (2018) A 2-year follow-up MRI study for the evaluation of an age estimation method based on knee bone development. International Journal of Legal MedicineGoogle Scholar
  24. 24.
    Tustison NJ, Avants BB, Cook PA, Zheng Y, Egan A, Yushkevich PA, Gee JC (2010) N4ITK: improved N3 bias correction. IEEE Trans Med Imaging 29:1310–1320CrossRefGoogle Scholar
  25. 25.
    NVIDIA GPU vs CPU? What is GPU Computing?Google Scholar
  26. 26.
    Krizhevsky A, Sutskever I, Hinton GE (2012) Imagenet Classification with Deep Convolutional Neural Networks. Advances in Neural Information Processing Systems 25 (NIPS 2012)Google Scholar
  27. 27.
    He K, Zhang X, Ren S, Sun J (2016) Deep residual learning for image recognition. In: The IEEE conference on computer vision and pattern recognition (CVPR), pp 770–778Google Scholar
  28. 28.
    Iandola FN, Han S, Moskewicz MW, Ashraf K, Dally WJ, Keutzer K (2016) SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and < 0.5MB model size. pp 1–13Google Scholar
  29. 29.
    Chen L-C, Papandreou G, Schroff F, Adam H (2017) Rethinking atrous convolution for semantic image segmentationGoogle Scholar
  30. 30.
    Zhao H, Shi J, Qi X, Wang X, Jia J (2016) Pyramid scene parsing networkGoogle Scholar
  31. 31.
    Lin G, Milan A, Shen C, Reid I (2016) RefineNet: multi-path refinement networks for high-resolution semantic segmentationGoogle Scholar
  32. 32.
    Chen L-C, Papandreou G, Kokkinos I, Murphy K, Yuille AL (2016) DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs. pp 1–14Google Scholar
  33. 33.
    Nekrasov V, Ju J, Choi J (2016) Global Deconvolutional Networks for Semantic segmentationGoogle Scholar
  34. 34.
    Milletari F, Navab N, Ahmadi S-A (2016) V-Net: fully convolutional neural networks for volumetric medical image segmentation. In: 2016 Fourth International conference on 3D vision (3DV), IEEE, pp 565–571Google Scholar
  35. 35.
    Çiçek Ö, Abdulkadir A, Lienkamp SS, Brox T, Ronneberger O (2016) 3D U-Net: learning dense volumetric segmentation from sparse annotation. In: Medical image computing and computer-assisted intervention – MICCAI 2016, vol 9901, pp 424–432Google Scholar
  36. 36.
    Badrinarayanan V, Kendall A, Cipolla R (2015) SegNet: A deep convolutional encoder-decoder architecture for image segmentation. pp 1–14Google Scholar
  37. 37.
    Nair V, Hinton GE (2010) Rectified linear units improve restricted boltzmann machines. In: Proceedings of the 27th international conference on machine learning (ICML-10), pp 807–814Google Scholar
  38. 38.
    Clevert D-A, Unterthiner T, Hochreiter S (2015) Fast and accurate deep network learning by exponential linear units (ELUs). pp 1–14Google Scholar
  39. 39.
    Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R (2014) Dropout: a simple way to prevent neural networks from overfitting. J Mach Learn Res 15:1929–1958Google Scholar
  40. 40.
    Simonyan K, Zisserman A (2014) Very deep convolutional networks for large-scale image recognition. ICLR 2015, pp 1–14Google Scholar
  41. 41.
    Ioffe S, Szegedy C (2015) Batch normalization: accelerating deep network training by reducing internal covariate shift. In: Proceedings of machine learning research, vol 37, pp 448–456Google Scholar
  42. 42.
    Kingma DP, Ba J (2015) Adam: a method for stochastic optimization. In: 3rd international conference for learning representations, pp 1–15Google Scholar
  43. 43.
    Chollet F (2017) Deep Learning With python, vol 1. Manning PublicationsGoogle Scholar
  44. 44.
    Prasoon A, Petersen K, Igel C, Lauze F, Dam E, Nielsen M (2013) Deep feature learning for knee cartilage segmentation using a triplanar convolutional neural network. Medical Image Computing and Computer-Assisted Intervention – MICCAI 2013 8150:246–253Google Scholar
  45. 45.
    Säring D, Auf der Mauer M, Jopp E (2014) Klassifikation des Verschlussgrades der Epiphyse der proximalen Tibia zur Altersbestimmung. In: Informatik aktuell. Springer, Berlin, pp 60–65Google Scholar
  46. 46.
    Saint-Martin P, Rérolle C, Dedouit F, Rousseau H, Rougé D, Telmon N (2014) Evaluation of an automatic method for forensic age estimation by magnetic resonance imaging of the distal tibial epiphysis—a preliminary study focusing on the 18-year threshold. Int J Legal Med 128:675–683CrossRefGoogle Scholar
  47. 47.
    Jopp E (2007) Methoden zur Alters- und Geschlechtsbestimmung auf dem Pruefstand - eine rechtsmedizinische empirische Studie. KovacGoogle Scholar
  48. 48.
    Krämer JA, Schmidt S, Jürgens KU, Lentschig M, Schmeling A, Vieth V (2014) The use of magnetic resonance imaging to examine ossification of the proximal tibial epiphysis for forensic age estimation in living individuals. Forensic Sci Med Pathol 10(3):306–313CrossRefGoogle Scholar
  49. 49.
    Saint-Martin P, Rérolle C, Pucheux J, Dedouit F, Telmon N (2015) Contribution of distal femur MRI to the determination of the 18-year limit in forensic age estimation. Int J Legal Med 129(3):619–620CrossRefGoogle Scholar
  50. 50.
    Pelt DM, Sethian JA (2018) A mixed-scale dense convolutional neural network for image analysis. Proc Natl Acad Sci 115(2):254–259CrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Department of Medical and Industrial Image ProcessingUniversity of Applied Sciences of WedelWedelGermany
  2. 2.Department of Legal MedicineUniversity Medical Center Hamburg-Eppendorf (UKE)HamburgGermany
  3. 3.Institute of Biomechanics M3Hamburg University of Technology (TUHH)HamburgGermany
  4. 4.Pediatric Radiology DepartmentUniversity Medical Center Hamburg-Eppendorf (UKE)HamburgGermany

Personalised recommendations