A pediatric bone age assessment method for hand bone X-ray images based on dual-path network

Shuang Wang1, Shuyan Jin2, Kun Xu3,4, Jiayan She5, Jipeng Fan6,7, Mingji He8, Liao Shaoyi Stephen9, Zhongjun Gao8, Xiaobo Liu7, Keqin Yao1
1Shenzhen Health Development Research and Data Management Center, Shenzhen, China
2Shenzhen Maternity & Child Healthcare Hospital, Southern Medical University, Shenzhen, China
3Shenzhen Institute for Advanced Study, UESTC, Shenzhen, China
4Shenzhen Hospital of Southern Medical University, Shenzhen, China
5West China Second University Hospital, Sichuan University, Chengdu, China
6DICOM Standard National and Local Collaborated Engineering Laboratory, Chengdu, China
7Chengdu Chengdian Goldisc Health Data Technology Co., Ltd, Chengdu, China
8Shenzhen Chengdian Goldisc Health Data Technology Co., Ltd., Shenzhen, China
9City University of Hong Kong, Hong Kong, China

Tóm tắt

Bone age assessment is a common diagnostic method used for abnormal growth and development in children. Despite recent significant advancements in convolutional neural network (CNN)-based intelligent bone age assessment in children, there remains room for improvement in assessment accuracy. Studies have indicated that a dual-path network (DPN) can incorporate different features of a CNN and improve the potential of the model to extract critical features compared to a single structural CNN. Attention mechanisms can also contribute to the enhanced ability of the model to extract channel and spatial features. Therefore, we propose a dual attention dual-path network (DADPN) to improve the accuracy of pediatric bone age assessment. DPN serves as a backbone network in DADPN by incorporating residual and dense connections. DPN was modified using two different attention mechanisms while containing gender information to compensate for physiological differences in bone age between males and females. Experiments were performed using this method with the RSNA Pediatric Bone Age Challenge dataset and compared to nine representative bone age assessment methods. This method achieved an optimal assessment accuracy with a mean absolute error (MAE) of 4.76 months. The experimental results suggest that the DADPN can extract the effective features of pediatric hand bone X-ray images and improve bone age assessment accuracy more than other deep learning methods.

Tài liệu tham khảo

Wan J, Zhao Y, Feng QQ et al (2021) Statistical confirmation of a method of US determination of bone age. Radiology 1:176–183. https://doi.org/10.1148/radiol.2021204353 Greulich WW, Idell PS (1959) Radio graphic atlas of skeletal development of the hand and wrist. Am J Med Sci 238(3):393. https://doi.org/10.1097/00000441-195909000-00030 Tanner JM, Whitehouse RH (1976) Clinical longitudinal standards for height, weight, height velocity, weight velocity, and stages of puberty. Arch Dis Child 51(3):170–179. https://doi.org/10.1136/adc.51.3.170 Thodberg HH, Kreiborg S, Juul A et al (2009) The BoneXpert method for automated determination of skeletal maturity. IEEE Trans Med Imaging 28(1):52–66. https://doi.org/10.1109/tmi.2008.926067 Nguyen QH, Nguyen BP, Nguyen MT et al (2022) Bone age assessment and sex determination using transfer learning. Expert Syst Appl 200:1–11. https://doi.org/10.1016/j.eswa.2022.116926 Pietka E, Pospiech KS, Gertych A et al (2003) Integration of computer assisted bone age assessment with clinical PACS. Comput Med Imaging Graph 27(2–3):217–228. https://doi.org/10.1016/S0895-6111(02)00076-9 Gertych A, Zhang A, Sayre J et al (2007) Bone age assessment of children using a digital hand atlas. Comput Med Imageing and Graphics 31(4):322–331. https://doi.org/10.1016/j.compmedimag.2007.02.012 Yang J, Xu Z, Wu W et al (2021) Erratum to: Graph Synergy: a network-inspired deep learning model for anticancer drug combination prediction. J Am Med Inform Assoc 28(1):2336–2345. https://doi.org/10.1093/jamia/ocab162 Wang Y, Zhang J, Cavichini M et al (2021) Robust Content-adaptive global registration for multimodal retinal images using weakly supervised deep-learning framework. IEEE Trans Image Process 30:3167–3178. https://doi.org/10.1109/TIP.2021.3058570 Halabi ss, Prevedello LM, Kalpathy CJ, et al (2018) The RSNA pediatric bone age machine learning challenge. Radiology 290(2):498–503. https://doi.org/10.1148/radiol.2018180736 Lee H, Tajmir S, Lee J et al (2017) Fully automated deep learning system for bone age assessment. J Digit Imaging 30(4):427–441. https://doi.org/10.1007/s10278-017-9955-8 Lee JH, Kim YJ, Kim KG (2020) Bone age estimation using deep learning and hand X-ray images. Biomed Eng Lett 10(3):323–331. https://doi.org/10.1007/s13534-020-00151-y Saric R, Kevric J, Ĉustovic E, et al (2019) Evaluation of skeletal gender and maturity for hand radiographs using deep convolutional neural networks. In: Proceedings of the 2019 6th International Conference on Control, Decision and Information Technologies(Co DIT), pp 1115–1120. https://doi.org/10.1109/CoDIT.2019.8820451 WU E, KONG B, WANG X, et al (2019) Residual attention based network for hand bone age assessment. In: 2019 IEEE 16th International Symposium on Biomedical Imaging, pp 1158–1161. https://doi.org/10.1109/ISBI.2019.8759332 Gao YY, Zhu T, Xu XH (2020) Bone age assessment based on deep convolution neural network incorporated with segmentation. Int J Comput Assist Radiol Surg 15(12):1951–1962. https://doi.org/10.1007/s11548-020-02266-0 Liu B, Zhang Y, Chu MC et al (2019) Bone age assessment based on Rank-Monotonicity enhanced ranking CNN. IEEE Access 7:120976–120983. https://doi.org/10.1109/ACCESS.2019.2937341 Salim I, Hamza AB (2021) Ridge regression neural network for pediatric bone age assessment. Multimed Tools Appl 80(20):461–478. https://doi.org/10.1007/s11042-021-10935-8 Hao P, Ye T, Xie X, et al (2020) Radiographs and texts fusion learning based deep networks for skeletal bone age assessment. Multimed Tools Appl.https://doi.org/10.1007/s11042-020-08943-1. Hao PY, Xie XH, Han TX, et al (2021) Overlap classification mechanism for skeletal bone age assessment. In: Proc of the 2nd ACM International Conference on Multimedia in Asia, Association for Computing Machinery, pp 1–7. https://doi.org/10.1145/3444685.3446286 Ren X, Li T, Yang X, et al (2018) Regression convolutional neural network for automated pediatric bone age assessment from hand radiograph. IEEE J Biomed Health Inf, 23(5): 2030–2038. https://doi.org/10.1109/jbhi.2018.2876916. Zhang X, Zhang JH, Zhang S (2022) Pediatric bone age assessment method combined with dual attention network. Appl Res Comput 39(10):1–8. https://doi.org/10.19734/j.issn.1001-3695.2022.03.0110 CHEN YP, LI JN, XIAO HX, et al (2017) Dual path networks. In: The 31st International Conference on Neural Information Processing Systems, pp 4470–4478. https://doi.org/10.48550/arXiv.1707.01629 LI Q, WANG X, GUAN X (2022) A Dual-path Network Chest Film Disease Classification Method Combined with a Triple Attention Mechanism. Journal of Electronics & Information Technology, 44:1–14. https://doi.org/10.11999/JEIT220172. Rajkomar A, Dean J, KOHANE I, (2019) Machine learning in medicine Reply. N Engl J Med 380(26):2589–2590. https://doi.org/10.1056/NEJMc1906060 Spampinato C, Palazzo S, Giordano D et al (2017) Deep learning for automated skeletal bone age assessment in X-ray images. Med Image Anal 36:41–51. https://doi.org/10.1016/j.media.2016.10.010 Chandran J J G, Karthick R, Rajagopal R, et al (2023) Dual-channel capsule generative adversarial network optimized with golden eagle optimization for pediatric bone age assessment from hand x-ray image. Int J Pattern Recog Artif Intell, 37(02). https://doi.org/10.1142/S0218001423540010 Palaniswamy T (2022) Hyperparameter optimization based deep convolution neural network model for automated bone age assessment and classification. Displays: Technology and Applications, 73: 206–211. https://doi.org/10.1016/j.displa.2022.102206 Pang S, Du A, Orgun MA et al (2021) Tumor attention networks: Better feature selection, better tumor segmentation. Neural Netw 140(1):203–222. https://doi.org/10.1016/j.neunet.2021.03.006 Wang S, Zhu Y, Lee S et al (2022) Global-Local attention network with multi-task uncertainty loss for abnormal lymph node detection in MR images. Med Image Anal 77(10):2345. https://doi.org/10.1016/j.media.2021.102345 Zhao Z, Liu Q, Wang S (2021) Learning deep global multi-scale and local attention features for facial expression recognition in the Wild. IEEE Trans Image Process Publi IEEE Signal Process Society 30:6544–6556. https://doi.org/10.1109/TIP.2021.3093397 Obeso AM, Benois PJ, Vazquez M et al (2022) Visual vs internal attention mechanisms in deep neural networks for image classification and object detection. Pattern Recogn 123:14. https://doi.org/10.1016/j.patcog.2021.108411 Arco JE, Ortiz A, Nicolás J et al (2023) Enhancing Multimodal Patterns in Neuroimaging by Siamese Neural Networks with Self-Attention Mechanism. Int J Neural Syst 33(04):19–36. https://doi.org/10.1142/S0129065723500193 Ding X, Nie W, Liu X et al (2023) Compact convolutional neural network with multi-headed attention mechanism for seizure prediction. Int J Neural Syst 33(03):14–31. https://doi.org/10.1142/S0129065723500144 Gayathri M, Malathy C [2022] A Deep Learning Framework for Intrusion Detection and Multimodal Biometric Image Authentication. J Mob Multimed, 18(2): 393–419. https://doi.org/10.13052/jmm1550-4646.18212 Shen K, Shi Q, Wang H (2021) Multimodal visibility deep learning model based on visible-infrared image pair. J Comput-Aided Des Comput Graph 33(6):939–946. https://doi.org/10.3724/SP.J.1089.2021.18420 Lu Z, Lu Y (2023) A balanced triplet loss for person re-identification. Int J Pattern Recog Artif Intell, 37(01): 2256022. https://doi.org/10.1142/S0218001422560225 Mikki S (2023) Artificial general intelligence and noncomputability: a dynamical framework. J Artifi Intell Conscious, 10(01): 71–101.https://doi.org/10.1142/S2705078522500163 Xuelong L (2023) Multi-modal cognitive computing.SCIENTIA SINICA Informationis, 2023, 53(1): 1–32. https://doi.org/10.1360/SSI-2022-0226 Liu T, Huang J, Liao T et al (2022) A hybrid deep learning model for predicting molecular subtypes of human breast cancer using multimodal data. Inno Res Biomed Eng: IRBM 1:62–74. https://doi.org/10.1016/j.irbm.2020.12.002 HE KM, ZHANG XY, REN SQ, et al (2016) Deep residual learning for image recognition. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) IEEE, 2016. pp 770–778. https://doi.org/10.1109/CVPR.2016.90 HUANG G, LIU Z, VAN DML, et al (2017) Densely connected convolutional networks. In: 2017 IEEE Conference on Computer Vision and Pattern Recognition, pp 2261–2269. https://doi.org/10.1109/CVPR.2017.243 Woo S, Park J, Lee JY, et al (2018) CBAM: Convolutional block attention module. In: The 15th European Conference on Computer Vision, 3–19. https://doi.org/10.1007/978-3-030-01234-2_1 Teixeira V, Braz L, Pedrini H, et al (2020) DuaLAnet: Dual lesion attention network for thoracic disease classification in chest X-rays. In: 2020 International Conference on Systems, Signals and Image Processing, pp 69–74. https://doi.org/10.1109/IWSSIP48289.2020.9145037 He KM, Gkioxari G, Dollar P, Girshick R (2020) Mask R-CNN. IEEE Trans Pattern Anal Mach Intell 42(2):386–397. https://doi.org/10.1109/TPAMI.2018.2844175 Ren SQ, He KM, Girshick R, Sun J (2017) Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Trans Pattern Anal Mach Intell 39(6):1137–1149. https://doi.org/10.1109/TPAMI.2016.2577031 Martin DD, Wit JM, Hochberg Z et al (2011) The use of bone age in clinical practice-part 1. Hormone Res Paediatrics 76(1):1–9. https://doi.org/10.1159/000329372 King DG, Steventon DM, Osullivan MP et al (1994) Reproducibility of bone ages when performed by radiology registrars: an audit of Tanner and Whitehouse II versus Greulich and Pyle methods. Br J Radiol 67(801):848–851. https://doi.org/10.1259/0007-1285-67-801-848 Kingma D, Jimmy B. Adam: a method for stochastic optimization. ICLR 2015:1–15. https://doi.org/10.48550/arXiv.1412.6980 Tan Mx, Le QV (2019) EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks. In: 36th International Conference on Machine Learning (ICML), pp 10691–10700. https://doi.org/10.48550/arXiv.1905.11946 Szegedy C, Vanhoucke V, Ioffe S et al (2016) Rethinking the inception architecture for computer vision. IEEE Conf Comput Vision Pattern Recognition: CVPR 2016:2276–3033. https://doi.org/10.1109/CVPR.2016.308 Chollet F (2017) Xception: deep learning with depthwise separable convolutions. In: Proceedings of the IEEE computer society conference on computer vision and pattern recognition. pp 1251–1258. https://doi.org/10.1109/CVPR.2017.195