Top

Published in:

01-04-2020 | Original Article

Bone age determination using only the index finger: a novel approach using a convolutional neural network compared with human radiologists

Authors: Nakul E. Reddy, Jesse C. Rayan, Ananth V. Annapragada, Nadia F. Mahmood, Alan E. Scheslinger, Wei Zhang, J. Herman Kan

Published in: Pediatric Radiology | Issue 4/2020

Abstract

Background

Recently developed convolutional neural network (CNN) models determine bone age more accurately than radiologists.

Objective

The purpose of this study was to determine whether a CNN and radiologists can accurately predict bone age from radiographs using only the index finger rather than the whole hand.

Materials and methods

We used a public anonymized dataset provided by the Radiological Society of North America (RSNA) pediatric bone age challenge. The dataset contains 12,611 hand radiographs for training and 200 radiographs for testing. The index finger was cropped from these images to create a second dataset. Separate CNN models were trained using the whole-hand radiographs and the cropped second-digit dataset using the consensus ground truth provided by the RSNA bone age challenge. Bone age determination using both models was compared with ground truth as provided by the RSNA dataset. Separately, three pediatric radiologists determined bone age from the whole-hand and index-finger radiographs, and the consensus was compared to the ground truth and CNN-model-determined bone ages.

Results

The mean absolute difference between the ground truth and CNN bone age for whole-hand and index-finger was similar (4.7 months vs. 5.1 months, P=0.14), and both values were significantly smaller than that for radiologist bone age determination from the single-finger radiographs (8.0 months, P<0.0001).

Conclusion

CNN-model-determined bone ages from index-finger radiographs are similar to whole-hand bone age interpreted by radiologists in the dataset, as well as a model trained on the whole-hand radiograph. In addition, the index-finger model performed better than the ground truth compared to subspecialty trained pediatric radiologists also using only the index finger to determine bone age. The radiologist interpreting bone age can use the second digit as a reliable starting point in their search pattern.

Available only for authorised users

McConkey MO, Bonasia DE, Amendola A (2011) Pediatric anterior cruciate ligament reconstruction. Curr Rev Musculoskelet Med 4:37CrossRef

Dimeglio A, Canavese F (2013) Progression or not progression? How to deal with adolescent idiopathic scoliosis during puberty. J Child Orthop 7:43–49CrossRef

Thangam P, Mahendiran TV, Thanushkodi K (2012) Skeletal bone age assessment — research directions. J Eng Sci Technol Rev 5:90–96CrossRef

Dahlberg PS, Mosdøl A, Ding KY et al (2017) Agreement between chronological age and bone age based on the Greulich and Pyle atlas for age estimation: a systematic review. Knowledge Centre for the Health Services at the Norwegian Institute of Public Health, Oslo

Larson DB, Chen MC, Lungren MP et al (2017) Performance of a deep-learning neural network model in assessing skeletal maturity on pediatric hand radiographs. Radiology 287:313–322CrossRef

Chartrand G, Cheng PM, Vorontsov E et al (2017) Deep learning: a primer for radiologists. Radiographics 37:2113–2131CrossRef

Andriole K, Erickson B, Flanders A et al (2017) Pediatric bone age challenge. http://rsnachallenges.cloudapp.net/competitions/4. Accessed 10 Aug 2019

Halabi SS, Prevedello LM, Kalpathy-Cramer J et al (2018) The RSNA pediatric bone age machine learning challenge. Radiology 290:498–503CrossRef

Iglovikov VI, Rakhlin A, Kalinin AA, Shvets AA (2018) Paediatric bone age assessment using deep convolutional neural networks. In: Deep learning in medical image analysis and multimodal learning for clinical decision support. Springer, Berlin, pp 300–308

10.

Paszke A, Gross S, Chintala S et al (2017) Automatic differentiation in PyTorch. https://openreview.net/pdf?id=BJJsrmfCZ. Accessed 28 Oct 2019

11.

Lin T-Y, Goyal P, Girshick R et al (2018) Focal loss for dense object detection. IEEE Trans Pattern Anal Mach Intell. https://doi.org/10.1109/TPAMI.2018.2858826

12.

Zeiler MD, Fergus R (2014) Visualizing and understanding convolutional networks. In: European conference on computer vision. Springer, Berlin, pp 818–833

13.

Chollet F (2017) Xception: deep learning with depthwise separable convolutions. IEEE Xplore. https://doi.org/10.1109/CVPR.2017.195

14.

Bland JM, Altman D (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 327:307–310CrossRef

15.

Springenberg JT, Dosovitskiy A, Brox T, Riedmiller M (2014) Striving for simplicity: the all convolutional net. arXiv, Cornell University, Ithaca

16.

Vignolo M, Milani S, DiBattista E et al (1990) Modified Greulich-Pyle, Tanner-Whitehouse, and Roche-Wainer-Thissen (knee) methods for skeletal age assessment in a group of Italian children and adolescents. Eur J Pediatr 149:314–317CrossRef

17.

Berst MJ, Dolan L, Bogdanowicz MM et al (2001) Effect of knowledge of chronologic age on the variability of pediatric bone age determined using the Greulich and Pyle standards. AJR Am J Roentgenol 176:507–510CrossRef

18.

Paxton ML, Lamont AC, Stillwell AP (2013) The reliability of the Greulich–Pyle method in bone age determination among Australian children. J Med Imaging Radiat Oncol 57:21–24CrossRef

19.

Lynnerup N, Belard E, Buch-Olsen K et al (2008) Intra- and interobserver error of the Greulich–Pyle method as used on a Danish forensic sample. Forensic Sci Int 179:e1–e6

20.

Mutasa S, Chang PD, Ruzal-Shapiro C, Ayyala R (2018) MABAL: a novel deep-learning architecture for machine-assisted bone age labeling. J Digit Imaging 31:513–519CrossRef

21.

Spampinato C, Palazzo S, Giordano D et al (2017) Deep learning for automated skeletal bone age assessment in X-ray images. Med Image Anal 36:41–51CrossRef

22.

Thodberg HH, Kreiborg S, Juul A, Pedersen KD (2008) The BoneXpert method for automated determination of skeletal maturity. IEEE Trans Med Imaging 28:52–66CrossRef

Title: Bone age determination using only the index finger: a novel approach using a convolutional neural network compared with human radiologists
Authors: Nakul E. Reddy
Jesse C. Rayan
Ananth V. Annapragada
Nadia F. Mahmood
Alan E. Scheslinger
Wei Zhang
J. Herman Kan
Publication date: 01-04-2020
Publisher: Springer Berlin Heidelberg
Published in: Pediatric Radiology / Issue 4/2020
Print ISSN: 0301-0449
Electronic ISSN: 1432-1998
DOI: https://doi.org/10.1007/s00247-019-04587-y

Keynote webinar | Spotlight on medication adherence

Springer Medicine

Bone age determination using only the index finger: a novel approach using a convolutional neural network compared with human radiologists

Abstract

Background

Objective

Materials and methods

Results

Conclusion

Keynote webinar | Spotlight on medication adherence

Springer Medicine

Abstract

Background

Objective

Materials and methods

Results

Conclusion

Please log in to get access to this content

Other articles of this Issue 4/2020

Measurements of cervical lymph nodes in children on computed tomography

The importance of fostering workplace wellness in pediatric radiology

Quantification of skeletal muscle mass: sarcopenia as a marker of overall health in children and adults

Ultrasound diagnosis of medial clavicular epiphysis avulsion fracture in a neonate

18F-2-fluoro-2-deoxyglucose uptake in white adipose tissue on pediatric oncologic positron emission tomography (PET)/computed tomography (CT)

Morphological changes of the dorsal contour of the corpus callosum during the first two years of life