基于YOLOv8模型辅助诊断斜肌功能异常

doi:10.6040/j.issn.1673-3770.0.2024.329

摘要/Abstract

摘要： 目的评估使用YOLOv8模型辅助诊断斜肌功能的准确性和可行性。方法研究收集深圳市眼科医院斜肌功能异常及正常受试者的眼底照相图片。使用YOLOv8模型进行训练,标记识别眼底照相图片的黄斑区及视盘区,定量检测黄斑中心凹-视盘中心夹角(disc-fovea angle, DFA)的大小以评估斜肌功能。结果斜肌功能异常组与正常组受试者的性别(χ2=0.478,P=0.489)和年龄(U=35891.5,P=0.770)差异均无统计学意义。概率密度曲线结果表明斜肌功能异常组与正常组的DFA分布存在差异。YOLOv8模型针对视盘类别的识别准确率为100%,对黄斑类别的识别准确率为95%,其在不同置信度阈值下对视盘类别的识别性能优于黄斑类别(对于整体类别,在置信度0.307时,模型的F1分数最高,为0.92)。独立样本t检验、Mann-Whitney U检验、Kolmogorov-Smirnov检验和Levene's检验均表明基于YOLOv8模型识别斜肌功能异常组和正常组的DFA存在显著差异(P均<0.001)。当DFA<-17.286或DFA>6.278时,DFA分布结果属于斜肌功能异常类别的概率>80.00%。结论应用YOLOv8模型可辅助临床评估斜肌功能。当DFA<-17.286°或DFA>6.278°时,斜肌功能异常的概率>80%。

关键词: 斜肌功能, 眼底照相, 黄斑中心凹-视盘中心夹角, YOLOv8模型, 机器学习

Abstract: Objective To evaluate the accuracy and feasibility of using YOLOv8 model to assist in diagnosing oblique muscle function. Methods Fundus photographs of abnormal oblique muscle function and normal subjects were collected from Shenzhen Eye Hospital. The YOLOv8 model was used for training, and the macular areas and optic disc area in fundus photographs were marked and identified, and the disc-fovea angle(DFA)of the macular fovea was quantitatively detected to evaluate the oblique muscle function. Results There were no significant differences in gender(χ2=0.478, P=0.489)and age(U=358 91.5, P=0.770)between the abnormal oblique muscle function group and the normal group. The results of the probability density curve showed that there were differences in DFA distribution between the abnormal oblique muscle function group and the normal one. The YOLOv8 model demonstrated a 100% recognition accuracy for the optic disc category and a 95% recognition accuracy for the macular category. The model exhibited superior recognition performance for the optic disc category in comparison to the macular category across diverse confidence thresholds(for the overall category, the F1 score of the model was the highest, at 0.92, when the confidence was 0.307). Independent sample t test, Mann-Whitney U test, Kolmogorov-Smirnov test and Levene's test all showed that there were significant differences in DFA between abnormal and normal oblique muscle function groups based on YOLOv8 model(all P<0.001). When DFA<-17.286 or DFA>6.278, the probability of DFA distribution results belonging to the category of oblique muscle dysfunction is greater than 80%. Conclusion The application of YOLOv8 model can assist clinical evaluation of oblique muscle function. When DFA is less than-17.286° or greater than 6.278°, the probability of abnormal oblique muscle function is greater than 80%.

Key words: Oblique muscle function, Fundus photography, Disc-fovea angle, YOLOv8 model, Machine learning

中图分类号:

R779.7

姚雪,陆小凤,张梦芮,胡馨雅,赵嘉洛,赖思思,李玄,刘子潇,沈超凡,范梓欣,张寅升,张国明. 基于YOLOv8模型辅助诊断斜肌功能异常[J]. 山东大学耳鼻喉眼学报, 2025, 39(5): 76-82.

YAO Xue, LU Xiaofeng, ZHANG Mengrui, HU Xinya, ZHAO Jialuo, LAI Sisi, LI Xuan, LIU Zixiao, SHEN Chaofan, FAN Zixin, ZHANG Yinsheng, ZHANG Guoming. Research on auxiliary diagnosis of ocular oblique muscle dysfunction based on the YOLOv8 model[J]. Journal of Otolaryngology and Ophthalmology of Shandong University, 2025, 39(5): 76-82.

参考文献

[1] 张伟, 谢芳. 旋转性斜视的诊断和治疗[J]. 中华眼科杂志, 2020, 56(3): 235-240. doi:10.3760/cma.j.issn.0412-4081.2020.03.016 ZHANG Wei, XIE Fang. Chinese Journal of Ophthalmology, 2020, 56(3): 235-240. doi:10.3760/cma.j.issn.0412-4081.2020.03.016
[2] Guyton DL. Ocular torsion reveals the mechanisms of cyclovertical strabismus: the Weisenfeld lecture[J]. Invest Ophthalmol Vis Sci, 2008, 49(3): 847-857, 846. doi:10.1167/iovs.07-0739
[3] Felius J, Locke KG, Hussein MA, et al. Photographic assessment of changes in torsional strabismus[J]. J AAPOS, 2009, 13(6): 593-595. doi:10.1016/j.jaapos.2009.09.008
[4] 陈遐. 眼底照相法在旋转斜视诊断和疗效评估中的应用[D]. 天津: 天津医科大学, 2006
[5] 唐凯, 毕宏生, 宋继科, 等. 基于马氏杆三棱镜和三棱镜交替遮盖试验的急性共同性内斜视手术疗效比较[J]. 中华眼视光学与视觉科学杂志, 2023, 25(4): 285-290. doi:10.3760/cma.j.cn115909-20221102-00424 TANG Kai, BI Hongsheng, SONG Jike, et al. Comparison of the surgical effects based on maddox prism test and prism and alternative cover test for acute acquired comitant esotropia[J]. Chinese Journal of Optometry Ophthalmology and Visual Science, 2023, 25(4): 285-290. doi:10.3760/cma.j.cn115909-20221102-00424
[6] Jethani J, Dave P. The subjectivity of objective evaluation of torsion on fundus photographs by practicing strabismologists[J]. Indian J Ophthalmol, 2018, 66(9): 1301-1303. doi:10.4103/ijo.IJO_182_17
[7] Spierer A. Measurement of cyclotorsion[J]. Am J Ophthalmol, 1996, 122(6): 911-912. doi:10.1016/s0002-9394(14)70402-7
[8] Freedman SF, Gearinger MD, Enyedi LB, et al. Measurement of ocular torsion after macular translocation: disc fovea angle and Maddox rod[J]. J AAPOS, 2003, 7(2): 103-107. doi:10.1016/mpa.2003.S1091853103000107
[9] Brodsky MC, Klaehn L, Goddard SM, et al. Heidelberg Spectralis infrared video imaging: a clinical tool for diagnosing ocular torsional instability[J]. J AAPOS, 2014, 18(3): 306-307. doi:10.1016/j.jaapos.2014.01.009
[10] Kang H, Lee SJ, Shin HJ, et al. Measuring ocular torsion and its variations using different nonmydriatic fundus photographic methods[J]. PLoS One, 2020, 15(12): e0244230. doi:10.1371/journal.pone.0244230
[11] 陈遐, 赵堪兴, 郭新, 等. 眼底照相法在下斜肌亢进诊断和疗效评估中的应用[J]. 眼视光学杂志, 2008(3): 222-224. doi:10.3760/cma.j.issn.1674-845X.2008.03.018 CHEN Xia, ZHAO Kanxing, GUO Xin, et al. Application of fundus photography in diagnosis and curative effect evaluation of inferior oblique hyperactivity[J]. Chinese Journal of Optometry & Ophthalmology, 2008(3): 222-224. doi:10.3760/cma.j.issn.1674-845X.2008.03.018
[12] Piedrahita-Alonso E, Valverde-Megias A, Gomez-de-Liano R. Rotation of retinal vascular Arcades and comparison with disc-fovea angle in the assessment of cycloposition[J]. Br J Ophthalmol, 2014, 98(1): 115-119. doi:10.1136/bjophthalmol-2013-303680
[13] Zheng B, Shen YF, Luo YX, et al. Automated measurement of the disc-fovea angle based on DeepLabv3[J]. Front Neurol, 2022, 13: 949805. doi:10.3389/fneur.2022.949805
[14] Simiera J, Ordon AJ, Loba P. Objective cyclodeviation measurement in normal subjects by means of Cyclocheck^® application[J]. Eur J Ophthalmol, 2021, 31(2): 704-708. doi:10.1177/1120672120905312
[15] Resch H, Pereira I, Hienert J, et al. Influence of disc-fovea angle and retinal blood vessels on interindividual variability of circumpapillary retinal nerve fibre layer[J]. Br J Ophthalmol, 2016, 100(4): 531-536. doi:10.1136/bjophthalmol-2015-307020
[16] Piedrahita-Alonso E, Valverde-Megias A, Martin-Garcia B, et al. Minimal detectable change of the disc-fovea angle for ocular torsion assessment[J]. Ophthalmic Physiol Opt, 2022, 42(1): 133-139. doi:10.1111/opo.12897
[17] 石争浩, 周亮, 李成建, 等. 深度学习方法在睡眠呼吸暂停检测中的研究进展[J]. 山东大学耳鼻喉眼学报, 2023, 37(6): 46-61. doi:10.6040/j.issn.1673-3770.0.2023.190 SHI Zhenghao, ZHOU Liang, LI Chengjian, et al. Research progress of deep learning methods in sleep apnea detection[J]. Journal of Otolaryngology and Ophthalmology of Shandong University, 2023, 37(6): 46-61. doi:10.6040/j.issn.1673-3770.0.2023.190
[18] 杜曰山一, 王鲜, 张国明. 人工智能辅助早产儿视网膜病变诊疗新进展[J]. 山东大学耳鼻喉眼学报, 2023, 37(3): 157-162. doi:10.6040/j.issn.1673-3770.0.2022.374 DU Yueshanyi, WANG Xian, ZHANG Guoming. Progress in the diagnosis and treatment of retinopathy of prematurity using artificial intelligence[J]. Journal of Otolaryngology and Ophthalmology of Shandong University, 2023, 37(3): 157-162. doi:10.6040/j.issn.1673-3770.0.2022.374
[19] 刘佳钰, 樊慧明, 邹游, 等. 人工智能在鼻咽癌诊断与治疗中的应用研究进展[J]. 山东大学耳鼻喉眼学报, 2023, 37(2): 135-142. doi:10.6040/j.issn.1673-3770.0.2022.089 LIU Jiayu, FAN Huiming, ZOU You, et al. Research progress on the application of artificial intelligence in the diagnosis and treatment of nasopharyn-geal carcinoma[J]. Journal of Otolaryngology and Ophthalmology of Shandong University, 2023, 37(2): 135-142. doi:10.6040/j.issn.1673-3770.0.2022.089
[20] Liu S, Qi L, Qin H, et al. Path aggregation network for instance segmentation 2018 IEEE/CVF conference on computer vision and pattern recognition[C]. Salt Lake: IEEE, 2018. doi: 10.1109/CVPR.2018.00913
[21] Zhu MJ, Han K, Yu CB, et al. Dynamic feature pyramid networks for object detection[EB/OL]. 2020: 2012.00779. https://arxiv.org/abs/2012.00779v2.
[22] 麦光焕. 眼外肌功能亢进与不足程度的分级方法[J]. 中华眼科杂志, 2005, 41(7): 663-666. doi:10.3760/j: issn: 0412-4081.2005.07.021 MAI Guanghuan. Chinese Journal of Ophthalmology, 2005, 41(7): 663-666. doi:10.3760/j: issn: 0412-4081.2005.07.021
[23] Del Monte MA, Parks MM. Denervation and extirpation of the inferior oblique. An improved weakening procedure for marked overaction[J]. Ophthalmology, 1983, 90(10): 1178-1185. doi:10.1016/s0161-6420(83)34409-2
[24] Wilson ME, Parks MM. Primary inferior oblique overaction in congenital esotropia, accommodative esotropia, and intermittent exotropia[J]. Ophthalmology, 1989, 96(7): 950-5; discussion956-7. doi:10.1016/s0161-6420(89)32774-6
[25] Ohba M, Nakagawa T. Treatment for “a” and “V” exotropia by slanting muscle insertions[J]. Jpn J Ophthalmol, 2000, 44(4): 433-438. doi:10.1016/S0021-5155(00)00182-9
[26] Lemos J, Eggenberger E. Clinical utility and assessment of cyclodeviation[J]. Curr Opin Ophthalmol, 2013, 24(6): 558-565. doi:10.1097/ICU.0000000000000003
[27] Jung JH, Holmes JM. Quantitative intraoperative torsional forced duction test[J]. Ophthalmology, 2015, 122(9): 1932-1938. doi:10.1016/j.ophtha.2015.05.025
[28] Jethani J, Dave P. A technique for standardizing disk foveal angle measurement[J]. J AAPOS, 2015, 19(1): 77-78. doi:10.1016/j.jaapos.2014.08.015
[29] Shin KH, Lee HJ, Lim HT. Ocular torsion among patients with intermittent exotropia: relationships with disease severity factors[J]. Am J Ophthalmol, 2013, 155(1): 177-182. doi:10.1016/j.ajo.2012.07.011

多维度评价

Viewed

Full text

Abstract

Cited

Shared

Discussed