Search

Purpose: Surgical workflow and skill analysis are key technologies for the next generation of cognitive surgical assistance systems. These systems could increase the safety of the operation through context-sensitive warnings and semi-autonomous robotic assistance or improve training of surgeons via data-driven feedback. In surgical workflow analysis up to 91% average precision has been reported for phase recognition on an open data single-center video dataset. In this work we investigated the generalizability of phase recognition algorithms in a multicenter setting including more difficult recognition tasks such as surgical action and surgical skill. Methods: To achieve this goal, a dataset with 33 laparoscopic cholecystectomy videos from three surgical centers with a total operation time of 22 h was created. Labels included framewise annotation of seven surgical phases with 250 phase transitions, 5514 occurences of four surgical actions, 6980 occurences of 21 surgical instruments from seven instrument categories and 495 skill classifications in five skill dimensions. The dataset was used in the 2019 international Endoscopic Vision challenge, sub-challenge for surgical workflow and skill analysis. Here, 12 research teams trained and submitted their machine learning algorithms for recognition of phase, action, instrument and/or skill assessment. Results: F1-scores were achieved for phase recognition between 23.9% and 67.7% (n = 9 teams), for instrument presence detection between 38.5% and 63.8% (n = 8 teams), but for action recognition only between 21.8% and 23.3% (n = 5 teams). The average absolute error for skill assessment was 0.78 (n = 1 team). Conclusion: Surgical workflow and skill analysis are promising technologies to support the surgical team, but there is still room for improvement, as shown by our comparison of machine learning algorithms. This novel HeiChole benchmark can be used for comparable evaluation and validation of future work. In future studies, it is of utmost importance to create more open, high-quality datasets in order to allow the development of artificial intelligence and cognitive robotics in surgery.

Surgical skills are associated with clinical outcomes. To improve surgical skills and thereby reduce adverse outcomes, continuous surgical training and feedback is required. Currently, assessment of surgical skills is a manual and time-consuming process which is prone to subjective interpretation. This study aims to automate surgical skill assessment in laparoscopic cholecystectomy videos using machine learning algorithms. To address this, a three-stage machine learning method is proposed: first, a Convolutional Neural Network was trained to identify and localize surgical instruments. Second, motion features were extracted from the detected instrument localizations throughout time. Third, a linear regression model was trained based on the extracted motion features to predict surgical skills. This three-stage modeling approach achieved an accuracy of 87 ± 0.2% in distinguishing good versus poor surgical skill. While the technique cannot reliably quantify the degree of surgical skill yet it represents an important advance towards automation of surgical skill assessment.

Purpose Cholecystectomy is one of the most common laparoscopic procedures. A critical phase of laparoscopic cholecystectomy consists in clipping the cystic duct and artery before cutting them. Surgeons can improve the clipping safety by ensuring full visibility of the clipper, while enclosing the artery or the duct with the clip applier jaws. This can prevent unintentional interaction with neighboring tissues or clip misplacement. In this article, we present a novel real-time feedback to ensure safe visibility of the instrument during this critical phase. This feedback incites surgeons to keep the tip of their clip applier visible while operating. Methods We present a new dataset of 300 laparoscopic cholecystectomy videos with frame-wise annotation of clipper tip visibility. We further present ClipAssistNet, a neural network-based image classifier which detects the clipper tip visibility in single frames. ClipAssistNet ensembles predictions from 5 neural networks trained on different subsets of the dataset. Results Our model learns to classify the clipper tip visibility by detecting its presence in the image. Measured on a separate test set, ClipAssistNet classifies the clipper tip visibility with an AUROC of 0.9107, and 66.15% specificity at 95% sensitivity. Additionally, it can perform real-time inference (16 FPS) on an embedded computing board; this enables its deployment in operating room settings. Conclusion This work presents a new application of computer-assisted surgery for laparoscopic cholecystectomy, namely real-time feedback on adequate visibility of the clip applier. We believe this feedback can increase surgeons’ attentiveness when departing from safe visibility during the critical clipping of the cystic duct and artery.

Objective Surgical skill is correlated with clinical outcomes. Therefore, the assessment of surgical skill is of major importance to improve clinical outcomes and increase patient safety. However, surgical skill assessment often lacks objectivity and reproducibility. Furthermore, it is time-consuming and expensive. Therefore, we developed an automated surgical skill assessment using machine learning algorithms. Methods Surgical skills were assessed in videos of laparoscopic cholecystectomy using a three-step machine learning algorithm. First, a three-dimensional convolutional neural network was trained to localize and classify the instruments within the videos. Second, movement patterns of the instruments were recorded over time and extracted. Third, the movement patterns were correlated with human surgical skill ratings using a linear regression model to predict surgical skill ratings automatically. Machine ratings were compared against human ratings of four board certified surgeons using a score ranging from 1 (poor skills) to 5 (excellent skills). Results Human raters and machine learning algorithms assessed surgical skills in 242 videos. Inter-rater reliability for human raters was excellent (79%, 95%CI 72-85%). Instrument detection showed an average precision of 78% and average recall of 82%. Machine learning algorithms showed an 87% accuracy in predicting good or poor surgical skills, when compared to human raters. Conclusion Machine learning algorithms can be trained to distinguish good and poor surgical skills with high accuracy. This work was published in Sci Rep 11, 5197 (2021). https://doi.org/10.1038/s41598-021-84295-6

Objective Laparoscopic cholecystectomy is one of the most common laparoscopic procedures. The critical phase of this intervention consists in dissecting the hepatocystic triangle and clipping the cystic artery and duct. Poor visibility of the clipper tips can result in unintentional clipping of neighboring tissues (“past-pointing”) or improper enclosing of the artery or duct, leading to hemorrhage or bile leaks. To improve patient safety during this clipping phase, we propose real-time intraoperative feedback to alert a surgeon when departing from safe behavior, i.e., losing visibility of the clipper tip. This is achieved using a deep learning model which classifies the clipper tip visibility in each frame. Methods We tailored a dataset for our application by selecting frames containing a clipper that were selected from 300 laparoscopic cholecystectomy videos. These 122k frames were annotated with binary labels: clipper tip visible/invisible. A frame was labelled as tip visible when the tips of both clipper jaws were visible. Frames in which the clipper tip was occluded (e.g. by tissue) or frames with poor image quality (e.g., bad contrast, blurriness/smoke) were labelled as tip invisible. Frames from 29 videos were set aside for a test set; the remaining frames were used for training/validation (80%, 20% resp.). Using a 5-fold cross-validation scheme, convolutional neural networks (Resnet50 architecture) were trained to classify the clipper tip visibility in each frame. Finally, 5 neural networks trained in the cross-validation were ensembled into a single model by averaging their predictions. Results On the test set, the ensembled model achieved an AUROC of 0.906 and a specificity of 64.5% at 95% sensitivity. Looking at per video performance, the median specificity across videos raised to 76.6% (at 95% sensitivity). That is, the model would correctly detect 95% of the clipper tip not visible cases; in the majority of the interventions, 7 out of 10 warnings would be justified. Conclusion We propose a novel safety feedback which warns on poor visibility of the clipper while clipping the cystic duct or artery. While being accurate, our technical solution runs in real-time, a requirement for intraoperative use. We believe this feedback can raise surgeons’ attentiveness when departing from safe visibility during this critical phase of laparoscopic cholecystectomy.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Intraoperative tracking of laparoscopic instruments is often a prerequisite for computer and robotic-assisted interventions. While numerous methods for detecting, segmenting and tracking of medical instruments based on endoscopic video images have been proposed in the literature, key limitations remain to be addressed: Firstly, robustness, that is, the reliable performance of state-of-the-art methods when run on challenging images (e.g. in the presence of blood, smoke or motion artifacts). Secondly, generalization; algorithms trained for a specific intervention in a specific hospital should generalize to other interventions or institutions. In an effort to promote solutions for these limitations, we organized the Robust Medical Instrument Segmentation (ROBUST-MIS) challenge as an international benchmarking competition with a specific focus on the robustness and generalization capabilities of algorithms. For the first time in the field of endoscopic image processing, our challenge included a task on binary segmentation and also addressed multi-instance detection and segmentation. The challenge was based on a surgical data set comprising 10,040 annotated images acquired from a total of 30 surgical procedures from three different types of surgery. The validation of the competing methods for the three tasks (binary segmentation, multi-instance detection and multi-instance segmentation) was performed in three different stages with an increasing domain gap between the training and the test data. The results confirm the initial hypothesis, namely that algorithm performance degrades with an increasing domain gap. While the average detection and segmentation quality of the best-performing algorithms is high, future research should concentrate on detection and segmentation of small, crossing, moving and transparent instrument(s) (parts).