{"title":"“通过机器学习对人工智能驱动的视频分析进行微调;面瘫自动评估工具的开发","authors":"Takeichiro Kimura, Keigo Narita, Kohei Oyamada, Masahiko Ogura, Tomoyasu Ito, Takashi Okada, Akihiko Takushima","doi":"10.1097/PRS.0000000000011924","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>Establishment of a quantitative, objective evaluation tool for facial palsy has been a challenging issue for clinicians and researchers, and artificial intelligence-driven video analysis can be considered a reasonable solution. The authors introduced facial keypoint detection, which detects facial landmarks with 68 points, but existing models had been organized almost solely with images of healthy individuals, and low accuracy was presumed in the prediction of asymmetric faces of patients with facial palsy. The accuracy of the existing model was assessed by applying it to videos of 30 patients with facial palsy. Qualitative review clearly showed its insufficiency. The model was prone to detect patients' faces as symmetric, and was unable to detect eye closure. Thus, the authors enhanced the model through the machine-learning process of annotation (ie, fine-tuning).</p><p><strong>Methods: </strong>A total of 1181 images extracted from the videos of 196 patients were enrolled in the training, and these images underwent manual correction of 68 keypoints. The annotated data were integrated into the previous model with a stack of 2 hourglass networks combined with channel aggregation block.</p><p><strong>Results: </strong>The postannotation model showed improvement in normalized mean error from 0.026 to 0.018, and qualitative keypoint detection on each facial unit revealed improvements.</p><p><strong>Conclusions: </strong>Strict control of inter- and intra-annotator variability successfully fine-tuned the presented model. The new model is a promising solution for objective assessment of facial palsy.</p>","PeriodicalId":20128,"journal":{"name":"Plastic and reconstructive surgery","volume":" ","pages":"1071e-1081e"},"PeriodicalIF":3.2000,"publicationDate":"2025-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Fine-Tuning on AI-Driven Video Analysis through Machine Learning: Development of an Automated Evaluation Tool of Facial Palsy.\",\"authors\":\"Takeichiro Kimura, Keigo Narita, Kohei Oyamada, Masahiko Ogura, Tomoyasu Ito, Takashi Okada, Akihiko Takushima\",\"doi\":\"10.1097/PRS.0000000000011924\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Background: </strong>Establishment of a quantitative, objective evaluation tool for facial palsy has been a challenging issue for clinicians and researchers, and artificial intelligence-driven video analysis can be considered a reasonable solution. The authors introduced facial keypoint detection, which detects facial landmarks with 68 points, but existing models had been organized almost solely with images of healthy individuals, and low accuracy was presumed in the prediction of asymmetric faces of patients with facial palsy. The accuracy of the existing model was assessed by applying it to videos of 30 patients with facial palsy. Qualitative review clearly showed its insufficiency. The model was prone to detect patients' faces as symmetric, and was unable to detect eye closure. Thus, the authors enhanced the model through the machine-learning process of annotation (ie, fine-tuning).</p><p><strong>Methods: </strong>A total of 1181 images extracted from the videos of 196 patients were enrolled in the training, and these images underwent manual correction of 68 keypoints. The annotated data were integrated into the previous model with a stack of 2 hourglass networks combined with channel aggregation block.</p><p><strong>Results: </strong>The postannotation model showed improvement in normalized mean error from 0.026 to 0.018, and qualitative keypoint detection on each facial unit revealed improvements.</p><p><strong>Conclusions: </strong>Strict control of inter- and intra-annotator variability successfully fine-tuned the presented model. The new model is a promising solution for objective assessment of facial palsy.</p>\",\"PeriodicalId\":20128,\"journal\":{\"name\":\"Plastic and reconstructive surgery\",\"volume\":\" \",\"pages\":\"1071e-1081e\"},\"PeriodicalIF\":3.2000,\"publicationDate\":\"2025-06-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Plastic and reconstructive surgery\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.1097/PRS.0000000000011924\",\"RegionNum\":2,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"2024/12/17 0:00:00\",\"PubModel\":\"Epub\",\"JCR\":\"Q1\",\"JCRName\":\"SURGERY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Plastic and reconstructive surgery","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1097/PRS.0000000000011924","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2024/12/17 0:00:00","PubModel":"Epub","JCR":"Q1","JCRName":"SURGERY","Score":null,"Total":0}
Fine-Tuning on AI-Driven Video Analysis through Machine Learning: Development of an Automated Evaluation Tool of Facial Palsy.
Background: Establishment of a quantitative, objective evaluation tool for facial palsy has been a challenging issue for clinicians and researchers, and artificial intelligence-driven video analysis can be considered a reasonable solution. The authors introduced facial keypoint detection, which detects facial landmarks with 68 points, but existing models had been organized almost solely with images of healthy individuals, and low accuracy was presumed in the prediction of asymmetric faces of patients with facial palsy. The accuracy of the existing model was assessed by applying it to videos of 30 patients with facial palsy. Qualitative review clearly showed its insufficiency. The model was prone to detect patients' faces as symmetric, and was unable to detect eye closure. Thus, the authors enhanced the model through the machine-learning process of annotation (ie, fine-tuning).
Methods: A total of 1181 images extracted from the videos of 196 patients were enrolled in the training, and these images underwent manual correction of 68 keypoints. The annotated data were integrated into the previous model with a stack of 2 hourglass networks combined with channel aggregation block.
Results: The postannotation model showed improvement in normalized mean error from 0.026 to 0.018, and qualitative keypoint detection on each facial unit revealed improvements.
Conclusions: Strict control of inter- and intra-annotator variability successfully fine-tuned the presented model. The new model is a promising solution for objective assessment of facial palsy.
期刊介绍:
For more than 70 years Plastic and Reconstructive Surgery® has been the one consistently excellent reference for every specialist who uses plastic surgery techniques or works in conjunction with a plastic surgeon. Plastic and Reconstructive Surgery® , the official journal of the American Society of Plastic Surgeons, is a benefit of Society membership, and is also available on a subscription basis.
Plastic and Reconstructive Surgery® brings subscribers up-to-the-minute reports on the latest techniques and follow-up for all areas of plastic and reconstructive surgery, including breast reconstruction, experimental studies, maxillofacial reconstruction, hand and microsurgery, burn repair, cosmetic surgery, as well as news on medicolegal issues. The cosmetic section provides expanded coverage on new procedures and techniques and offers more cosmetic-specific content than any other journal. All subscribers enjoy full access to the Journal''s website, which features broadcast quality videos of reconstructive and cosmetic procedures, podcasts, comprehensive article archives dating to 1946, and additional benefits offered by the newly-redesigned website.