浏览全部资源
扫码关注微信
收稿：2020-11-24，
修回：2021-3-22，
录用：2021-3-29，
纸质出版：2022-03-16
移动端阅览
目的
2
在骨龄智能评估研究中，如何准确地提取手腕参照骨的兴趣区域(region of interest
ROI)是保证骨龄精确评估的关键。基于传统深度学习的方法用于手腕骨ROI提取，存在个别参照骨漏判、误判等情况，导致平均提取准确率较低。本文结合目标检测强大的定位和识别能力，以准确提取所有手腕骨ROI为目的，提出了一种参照骨自动匹配与修正方法。
方法
2
针对不同参照骨形状、位置等特征表现出的规律性和关联性，本文采集了大量不同性别、不同年龄段的人手腕图谱作为参照骨样本匹配，然后分多个阶段提取参照骨ROI：1)基于目标检测算法初步提取出所有参照骨候选ROI，并根据一定的阈值剔除置信度较低的区域；2)结合参照骨的大数据样本构建位置点匹配模型，对剔除区域进行自动匹配与填补，保证ROI提取的完整性；3)通过多尺度滑动窗口以及ROI分类模型，对填补得到的ROI位置进行滑动修正，进一步保证提取准确率。
结果
2
实验结果表明，本文结合目标检测与匹配修正的方法优于现有绝大多数方法。其中，匹配修正方法在目标检测算法的提取结果基础上，提升了约1.42%的平均准确率，当结合Faster R-CNN(region-convolutional neural network)算法时能达到最高98.45%的交并比(intersection-over-union
IoU)准确率。
结论
2
本文方法利用手腕骨的位置特征，对个别提取困难的参照骨类型进行重新匹配与修正，有效地弥补了传统方法泛化能力不足的缺点。本文方法能够同时面向人手腕中所有参照骨ROI提取，具备良好的扩展性和易操作性。
Objective
2
The regions of interest (ROI) extrancted correction in hand-wrist reference bones is the essential method to target accurate bone age. Current image processing methods have high time complexity and operation difficulty
which cannot meet the requirement of large-scale clinical use. Deep learning technology has its feature extraction priority but the missing and misjudgment of individual reference bones will lead to low average extraction accuracy when it is used to extract ROIs of hand-wrist reference bones. The intensive positioning and classification capability of object detection algorithms is analyzed to demonstrate an automatic reference bone matching and correction method for the purpose of extracting all hand-wrist ROIs accurately.
Method
2
The amount of reference bones in the human hand-wrist is clear in common. The identified illustration strict regularity and correlation for the time-scaled bone age. Based on this
a large quantity of human hand-wrist X-ray images have been collected as the benchmark in genders and age scales. As for the evaluation
the reference bone sample orientation can be easily identified to calibrate the reference bone ROI and improve the extraction accuracy further based on matching the position similarity between the sample and the standard atlas. This demonstration is mainly divided into 3 steps: 1) All reference bone candidate ROIs are extracted in preliminary based on the object detection algorithm. Due to the insufficient generalization capability of most deep learning models
the accuracy rate is low when extracting some complicated reference bones. Based on the results of the reference bone candidate ROIs generated
this research adds a series of post-processing procedures
including replicated ROIs of the same reference bone category deduction
and filter regions alteration with low confidence in terms of the threshold set by the algorithm; 2) In order to guarantee the overall capability of the reference bone ROI extraction
it is necessary to match automatically and refill the deducted regions of the sample. First
while the skeletal development process of different human hand-wrists is similar
the distribution of reference bones in the same skeletal region is also strictly regular. It is of great value to use the large data atlas of reference bones to evaluate the sample's ROI. Hence
the human hand-wrist is segmented into 4 individual skeletal regions in terms of the bone growth characteristics. Different skeletal regions are matched and corrected independently. Next
a location-point matching model sorted out the most similar changeable features
such as the shape and position of the deducted region based on the big data standardized atlas of reference bones and choose it as the reference bone ROI that is missing from the sample; 3) The generated ROI cannot guarantee high-coordinated accuracy
it is required to optimize the location coordinates for a higher extraction accuracy further based on the refilled ROI result. On the basis of the refilled ROI coordinates
all possible regions of the reference bone category in the hand-wrist are categorized via a multi-scale sliding window and construct the ROI classification model to generate the possibility of the current reference bone in the window region. The region with the highest confidence is altered as the revised reference bone ROI. The basic network structure of AlexNet is demonstrated in the ROI classification model
which includes five convolutional layers and three fully connected layers.Moreover
the image augmentation method is applied to generate all positive and negative samples of the reference bone ROIs for model training. At the end
the trained model is applied in to verify its classification effect in the testing set.
Result
2
The demonstrated results illustrated that the intergrated method of object detection and matching correction algorithms has its priority. The ROI classification model application can achieve a classification accuracy of more than 99% for all 14 reference bones. When the location-point matching model is used to correct a single reference bone ROI in different skeletal regions
it can reach an average of 92.56% accuracy. The integrated matching correction method combining with proposed models can improve the average accuracy of 1.42% based on the extraction results from the object detection algorithms. The average accuracy of faster region-convolutional neural network (Faster R-CNN)
you only look once v3(YOLOv3)
and single shot multibox detector(SSD) has increased by 0.54%
2.48% and 1.23%
respectively. Based on the Faster R-CNN algorithm intergration
it can reach the highest 98.45% intersection-over-union (IoU) accuracy rate. The calculated IoU value is greater than or equal to 0.75
the currently predicted candidate region is assessed as the correct region.
Conclusion
2
In terms of the location features of the hand-wrist reference bones
this method can re-match and correct the individual reference bone categories that are difficult to extract. Following sufficient data atlas of other reference bone categories
this method can be further applied to multiple bone age scoring methods including the TW3 and RUS-CHN method.
Aja-Fernández S, De Luis-garcía R, Martín-Fernández M Á and Alberola-López C. 2004. A computational TW3 classifier for skeletal maturity assessment. A computing with words approach. Journal of Biomedical Informatics, 37(2): 99-107 [DOI: 10.1016/j.jbi.2004.01.002]
Berst M J, Dolan L, Bogdanowicz M M, Stevens M A, Chow S and Brandser E A. 2001. Effect of knowledge of chronologic age on the variability of pediatric bone age determined using the greulich and Pyle standards. American Journal of Roentgenology, 176(2): 507-510 [DOI: 10.2214/ajr.176.2.1760507]
Cao R, Xiong KY, Huang W H, Yao L H, He H and He Y Y. 2018. An algorithm to recognize ossification centers of carpal bones based on RUS-CHN standard. Journal of Beijing Sport University, 41(4): 75-81
曹润, 熊开宇, 黄伟航, 姚乐辉, 何辉, 贺莹莹. 2018. 基于RUS-CHN标准的手腕骨骨化中心识别分割算法. 北京体育大学学报, 41(4): 75-81 [DOI: 10.19582/j.cnki.11-3785/g8.2018.04.011]
Chen L J, Zhou X N, Wang M H, Qiu J F, Liu S Y and Mao K J. 2019. ARU-Net: research and application for wrist reference bone segmentation//International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob). Barcelona, Spain: IEEE: 1-5 [ DOI: 10.1109/WiMOB.2019.8923394 http://dx.doi.org/10.1109/WiMOB.2019.8923394 ]
Cheng L L, Tang P, Luo F L, Xiong Y H, Zhang B and Liu J H. 2019. Progress in skeletal age estimation with radiological technology. Chinese Journal of Forensic Medicine, 34(1): 62-66
程龙龙, 汤鹏, 罗方亮, 熊妍荷, 张彪, 刘技辉. 2019. 应用医学影像学技术评定骨龄的研究进展. 中国法医学杂志, 34(1): 62-66 [DOI: 10.13618/j.issn.1001-5728.2019.01.14]
Ding W L, Ding X, Chi K K, Chen Q and Mao K J. 2020. An adaptive extraction method of carpal feature region based on neural network. Journal of Zhejiang University of Technology, 48(5): 562-569
丁维龙, 丁潇, 池凯凯, 陈琦, 毛科技. 2020. 基于神经网络的手骨特征区域自适应提取方法. 浙江工业大学学报, 48(5): 562-569 [DOI: 10.3969/j.issn.1006-4303.2020.05.016]
Girshick R, Donahue J, Darrell T and Malik J. 2014. Rich feature hierarchies for accurate object detection and semantic segmentation//Proceedings of 2014 IEEE Conference on Computer Vision and Pattern Recognition. Columbus, USA: IEEE: 580-587 [ DOI: 10.1109/CVPR.2014.81 http://dx.doi.org/10.1109/CVPR.2014.81 ]
Greulich W W and Pyle S I. 1959. Radiographic Atlas of Skeletal Development of the Hand and Wrist. Stanford: Stanford University Press.
Helm S. 1979. Skeletal maturity in Danish schoolchildren assessed by the TW2 method. American Journal of Physical Anthropology, 51(3): 345-351 [DOI: 10.1002/ajpa.1330510305]
Hue T T M, Song M G, Kim J Y and Choi S H. 2011. Hand bone image segmentation using watershed transform with multistage merging. The Journal of Korean Institute of Information Technology, 9: 59-66
Ji Y F and Wu X S. 2010. Study of segmentation of hand X-ray regions of interest. Computer Engineering and Applications, 46(19): 246-248
计玉芳, 吴锡生. 2010. 手腕骨X线片兴趣区域图像分割的研究. 计算机工程与应用, 46(19): 246-248 [DOI: 10.3778/j.issn.1002-8331.2010.19.071]
Li J and Zhao H Y. 2013. A 3D segmentation technique for carpal bone images. China Medical Devices, 28(10): 151-152
李晶, 赵海燕. 2013. 手腕骨三维图像分割方法. 中国医疗设备, 28(10): 151-152 [DOI: 10.3969/j.issn.1674-1633.2013.10.056]
Li R, Zhang S J, Huang A Y and Chen H. 2020. Bone age assessment of hand bone for adolescents based on deep learning. Computer Technology and Development, 30(1): 124-128, 134
李睿, 张世杰, 黄奥云, 陈虎. 2020. 基于深度学习的青少年手腕骨骨龄评价. 计算机技术与发展, 30(1): 124-128, 134 [DOI: 10.3969/j.issn.1673-629X.2020.01.022]
Liang B Y, Zhai Y K, Tong C, Zhao J, Li J, He X Y and Ma Q Q. 2019. A deep automated skeletal bone age assessment model via region-based convolutional neural network. Future Generation Computer Systems, 98: 54-59 [DOI: 10.1016/j.future.2019.01.057]
Liu M Q, Lan J,Chen X, Yu G J and Yang X J. 2018. Bone age assessment model based on multi-dimensional feature fusion using deep learning. Academic Journal of Second Military Medical University, 39(8): 909-916
刘鸣谦, 兰钧, 陈旭, 于广军, 杨秀军. 2018. 基于多维度特征融合的深度学习骨龄评估模型. 第二军医大学学报, 39(8): 909-916 [DOI: 10.16781/j.0258-879x.2018.08.0909]
Liu R, Jia Y Y, He X Q, Li Z, Cai J H and Li H. 2019. Hand bone segmentation of skeletal age X-ray image based on Softmax regression model. Journal of Chongqing University, 42(9): 73-83
刘蕊, 贾媛媛, 贺向前, 李哲, 蔡金华, 李昊. 2019. 基于Softmax回归模型的骨龄X射线图像手骨分割. 重庆大学学报, 42(9): 73-83 [DOI: 10.11835/j.issn.1000-582X.2019.09.009]
Liu W, Anguelov D, Erhan D, Szegedy C, Reed S, Fu C Y and Berg A C. 2016. SSD: single shot MultiBox detector//Proceedings of the 14th European Conference on Computer Vision. Amsterdam, the Netherlands: Springer: 21-37 [ DOI: 10.1007/978-3-319-46448-0_2 http://dx.doi.org/10.1007/978-3-319-46448-0_2 ]
Pietka E, Gertych A, Pospiech S, Cao F, Huang H K and Gilsanz V. 2001. Computer-assisted bone age assessment: image preprocessing and epiphyseal/metaphyseal ROI extraction. IEEE Transactions on Medical Imaging, 20(8): 715-729 [DOI: 10.1109/42.938240]
Ran L K, Zhou L H and Chen Z. 2011. Novel location algorithm research for phalange and carpal ROI based on k -cosine and shape information. Computer Engineering and Applications, 47(12): 169-172
冉隆科, 周丽华, 陈忠. 2011. 一种新的指腕骨ROI的定位算法研究——基于 k 余弦和形状信息. 计算机工程与应用, 47(12): 169-172 [DOI: 10.3778/j.issn.1002-8331.2011.12.048]
Redmon J, Divvala S, Girshick R and Farhadi A. 2016. You only look once: unified, real-time object detection//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA: IEEE: 779-788 [ DOI: 10.1109/CVPR.2016.91 http://dx.doi.org/10.1109/CVPR.2016.91 ]
Ren S Q, He K M, Girshick R and Sun J. 2017. Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(6): 1137-1149 [DOI: 10.1109/TPAMI.2016.2577031]
Simu S, Lal S, Nagarsekar P and Naik A. 2017. Fully automatic ROI extraction and edge-based segmentation of radius and ulna bones from hand radiographs. Biocybernetics and Biomedical Engineering, 37(4): 718-732 [DOI: 10.1016/j.bbe.2017.07.004]
Song R G. 2014. On hand bones roi location for X ray image of bone age. Computer Applications and Software, 31(7): 226-228, 232
宋日刚. 2014. 骨龄X射线图像的手骨兴趣区域定位研究. 计算机应用与软件, 31(7): 226-228, 232 [DOI: 10.3969/j.issn.1000-386x.2014.07.057]
Su L, Fu X J, Zhang X D, Cheng X G, Ma Y M, Gan Y G and Hu Q M. 2018. Delineation of carpal bones from hand X-ray images through prior model, and integration of region-based and boundary-based segmentations. IEEE Access, 6: 19993-20008 [DOI: 10.1109/ACCESS.2018.2815031]
Tanner J, Oshman D, Bahhage F and Healy M. 1997. Tanner-Whitehouse bone age reference values for North American children. The Journal of Pediatrics, 131(1): 34-40 [DOI: 10.1016/S0022-3476(97)90000-7]
Tanner J M, Whitehouse R H, Marshall W A, Healy M J R and Goldstein H. 1975. Assessment of Skeletal Maturity and Prediction of Adult Height (TW2 Method). London: Academic Press
Thangam P, Thanushkodi K and Mahendiran T V. 2012. PSO for graph-based segmentation of wrist bones in bone age assessment. International Journal of Computers Communications and Control, 8(1): 153-160 [DOI: 10.15837/ijccc.2013.1.179]
Thodberg H H, Kreiborg S, Juul A and Pedersen K D. 2009. The BoneXpert method for automated determination of skeletal maturity. IEEE Transactions on Medical Imaging, 28(1): 52-66 [DOI: 10.1109/TMI.2008.926067]
Yang L, Peng L M and Wang Y Q. 2009. Research of radius' segmentation and extraction in bone age automatic evaluation system. Modern Electronics Technique, 32(16): 113-116, 119
杨莉, 彭利民, 王运琼. 2009. 骨龄自动评测系统中桡骨的分割与提取研究. 现代电子技术, 32(16): 113-116, 119 [DOI: 10.3969/j.issn.1004-373X.2009.16.035]
Zhang S Y, Yang S Z, Shao W D, Huang G P, Ou J H, Jin S G, Huang W J, Ji Q K and Wu H R. 1993. The standards of skeletal development of hand and wrist for Chinese——CHN method. China Sport Science, (6): 33-39, 93
张绍岩, 杨士增, 邵伟东, 黄耕培, 欧建华, 金寿光, 黄惟军, 纪庆科, 吴惠茹. 1993. 中国人手腕骨发育标准——CHN法. 体育科学, (6): 33-39, 93 [DOI: CNKI:SUN:TYKX.0.1993-06-008]
Zhong Y P. 2013. Research on automatic segmentation method of wrist bone image based on PCNN. Software Guide, 12(7): 166-167
Zhou X N. 2020. Research on wrist region of interest extraction algorithm for CHN intelligent bone age assessment. Hangzhou: Zhejiang University of Technology
周贤年. 2020. 面向CHN法智能骨龄评估的手腕部兴趣区域提取算法研究. 杭州: 浙江工业大学[ DOI: https://d.wanfangdata.com.cn/thesis/D02218786 http://dx.doi.org/https://d.wanfangdata.com.cn/thesis/D02218786 ]
Zhu X Y, Ouyang B, Luo D M and Liu J. 2016. Design of computerized assessment system for radius bone age grades according to CHN 05. China Sport Science, 36(9): 59-65, 72
朱翔宇, 欧阳斌, 罗冬梅, 刘杰. 2016. 基于中华05骨龄标准的桡骨骨龄等级计算机评定系统的设计. 体育科学, 36(9): 59-65, 72 [DOI: 10.16469/j.css.201609009]
相关作者
相关机构
京公网安备11010802024621
