1.中国中医科学院西苑医院综合内科,北京 100091
2.中国科学院微电子研究所,北京 100029
徐钰莹,女,31岁,博士。研究方向:中医药现代化研究。
宋婷婷,E-mail:songtingting@ime.ac.cn
李秋艳,E-mail:liuqiuyan1968@sohu.com
纸质出版日期:2024-08-25,
收稿日期:2023-11-11,
移动端阅览
徐钰莹,胡广洋,张继伟,等.Mask R-CNN神经网络模型对舌象裂纹严重程度的评价效果[J].北京中医药,2024,43(8):942-946.
XU Yuying,HU Guangyang,ZHANG Jiwei,et al.Evaluation results of Mask R-CNN neural network model on the severity of fissured tongue[J]. Beijing Journal of Traditional Chinese Medicine,2024,43(08):942-946.
徐钰莹,胡广洋,张继伟,等.Mask R-CNN神经网络模型对舌象裂纹严重程度的评价效果[J].北京中医药,2024,43(8):942-946. DOI: 10.16025/j.1674-1307.2024.08.023.
XU Yuying,HU Guangyang,ZHANG Jiwei,et al.Evaluation results of Mask R-CNN neural network model on the severity of fissured tongue[J]. Beijing Journal of Traditional Chinese Medicine,2024,43(08):942-946. DOI: 10.16025/j.1674-1307.2024.08.023.
目的
2
基于掩膜区域的卷积神经网络(Mask R-CNN)算法,在裂纹舌识别与提取的基础上探索裂纹舌严重程度的客观评价方法。
方法
2
从中国中医科学院西苑医院收集200例裂纹舌与200例非裂纹舌的舌象图片,建立基于神经网络的裂纹舌识别模型,以准确率、精确率、召回率对模型裂纹舌识别效果进行评价。由3名中医专业主任医师对200张裂纹舌图片按照轻度裂纹、中度裂纹、重度裂纹进行严重程度分级标注,通过裂纹识别模型进行裂纹舌的识别与特征提取,选择裂纹面积比(
<math id="M1"><mi>x</mi></math>
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=65746779&type=
2.28600001
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=65746781&type=
1.52400005
)、裂纹方向(
<math id="M2"><mi>z</mi></math>
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=65746796&type=
2.28600001
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=65746797&type=
1.18533325
)、裂纹条数(
<math id="M3"><mi>n</mi></math>
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=65746813&type=
2.28600001
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=65746814&type=
1.60866666
)、主裂纹长度(
<math id="M4"><mi>l</mi></math>
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=65746800&type=
2.28600001
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=65746818&type=
0.93133330
)作为裂纹评价指标,以医生标注结果作为分级标准,根据分级结果对各指标进行权重赋值,裂纹严重程度综合权重计算公式:
<math id="M5"><mi mathvariant="normal">W</mi><mo>=</mo><mfenced separators="|"><mrow><mo>∑</mo><msub><mrow><mi mathvariant="normal">w</mi></mrow><mrow><mi mathvariant="normal">i</mi></mrow></msub></mrow></mfenced><mo>/</mo><mn mathvariant="normal">4</mn><mo stretchy="false">(</mo><mi mathvariant="normal">i</mi><mo>=</mo><mi>x</mi><mo>
</mo><mi>z</mi><mo>
</mo><mi>n</mi><mo>
</mo><mi>l</mi><mo stretchy="false">)</mo></math>
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=65746823&type=
3.80999994
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=65746825&type=
39.70866776
,计算裂纹多维度指标的分布区间,评价裂纹的严重程度。
结果
2
模型识别裂纹舌,准确率为0.945,精确率为0.949,召回率为0.940。舌裂纹的严重程度评价结果:W∈[1,3]为轻度裂纹,W∈(3,6]为中度裂纹,W∈(6,10]为重度裂纹。经验证,裂纹舌总体评价准确率为88.3%,其中轻度裂纹的评价准确率为88.9%,中度裂纹的评价准确率为91.7%,重度裂纹评价准确率为83.3%。
结论
2
选择裂纹面积比、裂纹方向、裂纹条数、主裂纹长度作为评价裂纹舌严重程度的指标,可较好地完成辨识任务,实现舌象裂纹程度的定量化评价。
裂纹舌掩膜区域的卷积神经网络算法舌象特征提取严重程度舌象诊断中医客观化
赵志红,郝斌,钟鸣,等.裂纹舌新解[J].中华中医药杂志,2018,33(11):5103-5104.
朱文峰.中医诊断学[M].北京:中国中医药出版社,2002:75.
黄淑琼,张云龙,周静,等.浅谈中医舌象客观化、定量化、标准化研究[J].中华中医药杂志,2017,32(4):1625-1627.
施展,周昌乐.舌象裂纹提取及特征分析[J].计算机技术与发展,2007,17(5):245–253.
陈小芬,李翠华,杜晓凤.自适应阈值的舌象裂纹检测[J].计算机技术与发展,2009,19(1):17-20
颜建军,徐姿,郭睿等.基于Mask R-CNN的舌图像分割研究[J].世界科学技术-中医药现代化,2020,22(5):1532-1538.
HE K, GKIOXARI G, DOLLAR P, et al. Mask R-CNN[J]. IEEE Trans Pattern Anal Mach Intell,2020,42(2):386-397.
李晶.中医诊断学[M].2版.北京:科学出版社,2011.
JUN L, JING BH, TAO J, et al. A multi-step approach for tongue image classification in patients with diabetes[J]. Comput Biol Med, 2022,149(10):105935-105945.
ZHANG NN, JIANG ZX, LI JX, et al. Multiple color representation and fusion for diabetes mellitus diagnosis based on back tongue images[J]. Comput Biol Med, 2023,155(4):106652.
SHARMA M, SHARMA VK. Recurrent facial palsy and fissured tongue[J]. Eur J Intern Med, 2021,89(3):104-105.
PICCIANI BLS, TEIXEIRA-SOUZA T, PESSÔA TM, et al. Fissured tongue in patients with psoriasis[J]. J Am Acad Dermatol, 2018,78(2):413-414.
0
浏览量
0
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构