수술 동영상에서의 인공지능을 사용한 출혈 검출 연구
A Study on the Bleeding Detection Using Artificial Intelligence in Surgery Video
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Recently, many studies have introduced artificial intelligence systems in the surgical process to reduce the incidence and mortality of complications in patients. Bleeding is a major cause of operative mortality and com- plications. However, there have been few studies conducted on detecting bleeding in surgical videos. To advance the development of deep learning models for detecting intraoperative hemorrhage, three models have been trained and compared; such as, YOLOv5, RetinaNet50, and RetinaNet101. We collected 1,016 bleeding images extracted from five surgical videos. The ground truths were labeled based on agreement from two specialists. To train and evaluate models, we divided the datasets into training data, validation data, and test data. For training, 812 images (80%) were selected from the dataset. Another 102 images (10%) were used for evaluation and the remaining 102 images (10%) were used as the evaluation data. The three main metrics used to evaluate performance are precision, recall, and false positive per image (FPPI). Based on the evaluation metrics, RetinaNet101 achieved the best detection results out of the three models (Precision rate of 0.99±0.01, Recall rate of 0.93±0.02, and FPPI of 0.01±0.01). The infor- mation on the bleeding detected in surgical videos can be quickly transmitted to the operating room, improving patient outcomes.
Abstract
Recently, many studies have introduced artificial intelligence systems in the surgical process to reduce the incidence and mortality of complications in patients. Bleeding is a major cause of operative mortality and com- plications. However, there have been few studies conducted on detecting bleeding in surgical videos. To advance the development of deep learning models for detecting intraoperative hemorrhage, three models have been trained and compared; such as, YOLOv5, RetinaNet50, and RetinaNet101. We collected 1,016 bleeding images extracted from five surgical videos. The ground truths were labeled based on agreement from two specialists. To train and evaluate models, we divided the datasets into training data, validation data, and test data. For training, 812 images (80%) were selected from the dataset. Another 102 images (10%) were used for evaluation and the remaining 102 images (10%) were used as the evaluation data. The three main metrics used to evaluate performance are precision, recall, and false positive per image (FPPI). Based on the evaluation metrics, RetinaNet101 achieved the best detection results out of the three models (Precision rate of 0.99±0.01, Recall rate of 0.93±0.02, and FPPI of 0.01±0.01). The infor- mation on the bleeding detected in surgical videos can be quickly transmitted to the operating room, improving patient outcomes.
참고문헌 (21)
- Shah A, "Strategies to minimize intraoperative blood loss during major surgery", Br J Surg, 107(2), pp.26~38 (2020)
- Wu WC, "Operative blood loss, blood transfusion, and 30-day mortality in older patients after major noncardiac surgery", Ann Surg, 252(1), pp.11~17 (2010)
- Smilowitz NR, "Association between anemia, bleeding, and transfusion with long-term mortality following noncardiac surgery", Am J Med, 129(3), pp.315~323 (2016)
- Christensen MC, "Increased chest tube drainage is independently associated with adverse outcome after cardiac surgery", J Cardiothorac Vasc Anesth, 26(1), pp.46~51 (2012)
- Stokes ME, "Impact of bleeding-related complications and", or blood product transfusions on hospital costs in inpatient surgical patients, 2011, pp.11, 135~
- Marietta M, "Pathophysiology of bleeding in surgery", Transplant Proc, 38(3), pp.812~814 (2006)
- Asfour V, "Vascular injury at laparoscopy : a guide to management", J Obstet Gynaecol, 38(5), pp.598~606 (2018)
- Rady MY, "Perioperative determinants of morbidity and mortality in elderly patients undergoing cardiac surgery", Crit Care Med, 26(225) (1998)
- Vincent JL, "Anemia and blood transfusion in critically ill patients", JAMA, 288(1499) (2002)
- Tran A, "Early identification of patients requiring massive transfusion, embolization or hemostatic surgery for traumatic hemorrhage : A systematic review and meta-analysis", J Trauma Acute Care Surg, 84(3), pp.505~516 (2018)
- Pesapane F, "Artificial intelligence in medical imaging : Threat or opportunity? Radiologists again at the forefront of innovation in medicine", Eur Radiol Exp, 2(1), pp.35~ (2018)
- Kawaguchi F, "Generalization in Deep Learning", Cambridge prisms, 1, pp.112~148 (2017)
- Andrew RB, "Universal approximation bounds for superpositions of a sigmoidal function", IEEE Trans Inf Theory, 39(3), pp.930~945 (1993)
- Leshno M, "Multilayer feedforward networks with a nonpolynomial activation function can approximate any function", Neural Netw, 6(6), pp.861~867 (1993)
- Garcia-Martinez A, "Automatic detection of surgical hemorrhage using computer vision", Artif Intell Med, 78, pp.55~60 (2017)
- Okamoto T, "Real-time identification of blood regions for hemostasis support in laparoscopic surgery", SIViP, 13, pp.405~412 (2019)
- Zhou XY, "Application of artificial intelligence in surgery", Front Med, 14(4), pp.417~430 (2020)
- Sunitha S, "An Improved Bleeding Detection Method for Wireless Capsule Endoscopy (WCE) Images Based on AlexNet", ICPSC, pp.11~15 (2021)
- Hawkins DM, "The Problem of Overfitting", J Chem Inf Comput Sci, 44(1), pp.1~12 (2004)
- 곽대원, "인공지능 의료 영상인식 기술을 활용한 유방암 영상 진단 기법 연구", 한국통신학회논문지, 48(2), pp.216~226 (2023), 한국통신학회[DOI]
- Lin TY, "Focal Loss for Dense Object Detection", ICCV, pp.2980~2988 (2017)