Surgical data science for next-generation interventions

Deep learning-based surgical step recognition for laparoscopic right-sided colectomy

PURPOSE

Understanding the complex anatomy and surgical steps involved in laparoscopic right-sided colectomy (LAP-RC) is essential for standardizing the surgical procedure. Deep-learning (DL)-based computer vision can achieve this. This study aimed to develop a step recognition model for LAP-RC using a dataset of surgical videos with annotated step information and evaluate its recognition performance.

METHODS

This single-center retrospective study utilized a video dataset of laparoscopic ileocecal resection (LAP-ICR) and laparoscopic right-sided hemicolectomy (LAP-RHC) for right-sided colon cancer performed between January 2018 and March 2022. The videos were split into still images, which were divided into training, validation, and test sets using 66%, 17%, and 17% of the data, respectively. Videos were manually classified into eight main steps: 1) medial mobilization, 2) central vascular ligation, 3) dissection of the superior mesenteric vein, 4) retroperitoneal mobilization, 5) lateral mobilization, 6) cranial mobilization, 7) mesocolon resection, and 8) intracorporeal anastomosis. In a simpler version, consecutive surgical steps were combined, resulting in five steps. Precision, recall, F1 scores, and overall accuracy were assessed to evaluate the model's performance in the surgical step classification task.

RESULTS

Seventy-eight patients were included; LAP-ICR and LAP-RHC were performed in 35 (44%) and 44 (56%) patients, respectively. The overall accuracy was 72.1% and 82.9% for the eight-step and combined five-step classification tasks, respectively.

CONCLUSIONS

The automatic surgical step-recognition model for LAP-RCs, developed using a DL algorithm, exhibited a fairly high classification performance. A model that understands the complex steps of LAP-RC will aid the standardization of the surgical procedure.

Present and Future Applications of Artificial Intelligence in Kidney Transplantation

Artificial intelligence (AI) has a wide and increasing range of applications across various sectors. In medicine, AI has already made an impact in numerous fields, rapidly transforming healthcare delivery through its growing applications in diagnosis, treatment and overall patient care. Equally, AI is swiftly and essentially transforming the landscape of kidney transplantation (KT), offering innovative solutions for longstanding problems that have eluded resolution through traditional approaches outside its spectrum. The purpose of this review is to explore the present and future applications of artificial intelligence in KT, with a focus on pre-transplant evaluation, surgical assistance, outcomes and post-transplant care. We discuss its great potential and the inevitable limitations that accompany these technologies. We conclude that by fostering collaboration between AI technologies and medical practitioners, we can pave the way for a future where advanced, personalised care becomes the standard in KT and beyond.

Towards multimodal graph neural networks for surgical instrument anticipation

PURPOSE

Decision support systems and context-aware assistance in the operating room have emerged as the key clinical applications supporting surgeons in their daily work and are generally based on single modalities. The model- and knowledge-based integration of multimodal data as a basis for decision support systems that can dynamically adapt to the surgical workflow has not yet been established. Therefore, we propose a knowledge-enhanced method for fusing multimodal data for anticipation tasks.

METHODS

We developed a holistic, multimodal graph-based approach combining imaging and non-imaging information in a knowledge graph representing the intraoperative scene of a surgery. Node and edge features of the knowledge graph are extracted from suitable data sources in the operating room using machine learning. A spatiotemporal graph neural network architecture subsequently allows for interpretation of relational and temporal patterns within the knowledge graph. We apply our approach to the downstream task of instrument anticipation while presenting a suitable modeling and evaluation strategy for this task.

RESULTS

Our approach achieves an F1 score of 66.86% in terms of instrument anticipation, allowing for a seamless surgical workflow and adding a valuable impact for surgical decision support systems. A resting recall of 63.33% indicates the non-prematurity of the anticipations.

CONCLUSION

This work shows how multimodal data can be combined with the topological properties of an operating room in a graph-based approach. Our multimodal graph architecture serves as a basis for context-sensitive decision support systems in laparoscopic surgery considering a comprehensive intraoperative operating scene.

Reducing annotating load: Active learning with synthetic images in surgical instrument segmentation

Accurate instrument segmentation in the endoscopic vision of minimally invasive surgery is challenging due to complex instruments and environments. Deep learning techniques have shown competitive performance in recent years. However, deep learning usually requires a large amount of labeled data to achieve accurate prediction, which poses a significant workload. To alleviate this workload, we propose an active learning-based framework to generate synthetic images for efficient neural network training. In each active learning iteration, a small number of informative unlabeled images are first queried by active learning and manually labeled. Next, synthetic images are generated based on these selected images. The instruments and backgrounds are cropped out and randomly combined with blending and fusion near the boundary. The proposed method leverages the advantage of both active learning and synthetic images. The effectiveness of the proposed method is validated on two sinus surgery datasets and one intraabdominal surgery dataset. The results indicate a considerable performance improvement, especially when the size of the annotated dataset is small. All the code is open-sourced at: https://github.com/HaonanPeng/active_syn_generator.

Extra-abdominal trocar and instrument detection for enhanced surgical workflow understanding

PURPOSE

Video-based intra-abdominal instrument tracking for laparoscopic surgeries is a common research area. However, the tracking can only be done with instruments that are actually visible in the laparoscopic image. By using extra-abdominal cameras to detect trocars and classify their occupancy state, additional information about the instrument location, whether an instrument is still in the abdomen or not, can be obtained. This can enhance laparoscopic workflow understanding and enrich already existing intra-abdominal solutions.

METHODS

A data set of four laparoscopic surgeries recorded with two time-synchronized extra-abdominal 2D cameras was generated. The preprocessed and annotated data were used to train a deep learning-based network architecture consisting of a trocar detection, a centroid tracker and a temporal model to provide the occupancy state of all trocars during the surgery.

RESULTS

The trocar detection model achieves an F1 score of 95.06 ± 0.88 % . The prediction of the occupancy state yields an F1 score of 89.29 ± 5.29 % , providing a first step towards enhanced surgical workflow understanding.

CONCLUSION

The current method shows promising results for the extra-abdominal tracking of trocars and their occupancy state. Future advancements include the enlargement of the data set and incorporation of intra-abdominal imaging to facilitate accurate assignment of instruments to trocars.

Aliado - A design concept of AI for decision support in oncological liver surgery

BACKGROUND

The interest in artificial intelligence (AI) is increasing. Systematic reviews suggest that there are many machine learning algorithms in surgery, however, only a minority of the studies integrate AI applications in clinical workflows. Our objective was to design and evaluate a concept to use different kinds of AI for decision support in oncological liver surgery along the treatment path.

METHODS

In an exploratory co-creation between design experts, surgeons, and data scientists, pain points along the treatment path were identified. Potential designs for AI-assisted solutions were developed and iteratively refined. Finally, an evaluation of the design concept was performed with n = 20 surgeons to get feedback on the different functionalities and evaluate the usability with the System Usability Scale (SUS). Participating surgeons had a mean of 14.0 ± 5.0 years of experience after graduation.

RESULTS

The design concept was named "Aliado". Five different scenarios were identified where AI could support surgeons. Mean score of SUS was 68.2 ( ± 13.6 SD). The highest valued functionalities were "individualized prediction of survival, short-term mortality and morbidity", and "individualized recommendation of surgical strategy".

CONCLUSION

Aliado is a design prototype that shows how AI could be integrated into the clinical workflow. Even without a fleshed out user interface, the SUS already yielded borderline good results. Expert surgeons rated the functionalities favorably, and most of them expressed their willingness to work with a similar application in the future. Thus, Aliado can serve as a surgical vision of how an ideal AI-based assistance could look like.

Real-time segmentation of biliary structure in pure laparoscopic donor hepatectomy

Pure laparoscopic donor hepatectomy (PLDH) has become a standard practice for living donor liver transplantation in expert centers. Accurate understanding of biliary structures is crucial during PLDH to minimize the risk of complications. This study aims to develop a deep learning-based segmentation model for real-time identification of biliary structures, assisting surgeons in determining the optimal transection site during PLDH. A single-institution retrospective feasibility analysis was conducted on 30 intraoperative videos of PLDH. All videos were selected for their use of the indocyanine green near-infrared fluorescence technique to identify biliary structure. From the analysis, 10 representative frames were extracted from each video specifically during the bile duct division phase, resulting in 300 frames. These frames underwent pixel-wise annotation to identify biliary structures and the transection site. A segmentation task was then performed using a DeepLabV3+ algorithm, equipped with a ResNet50 encoder, focusing on the bile duct (BD) and anterior wall (AW) for transection. The model's performance was evaluated using the dice similarity coefficient (DSC). The model predicted biliary structures with a mean DSC of 0.728 ± 0.01 for BD and 0.429 ± 0.06 for AW. Inference was performed at a speed of 15.3 frames per second, demonstrating the feasibility of real-time recognition of anatomical structures during surgery. The deep learning-based semantic segmentation model exhibited promising performance in identifying biliary structures during PLDH. Future studies should focus on validating the clinical utility and generalizability of the model and comparing its efficacy with current gold standard practices to better evaluate its potential clinical applications.

Artificial intelligence automated surgical phases recognition in intraoperative videos of laparoscopic pancreatoduodenectomy

BACKGROUND

Laparoscopic pancreatoduodenectomy (LPD) is one of the most challenging operations and has a long learning curve. Artificial intelligence (AI) automated surgical phase recognition in intraoperative videos has many potential applications in surgical education, helping shorten the learning curve, but no study has made this breakthrough in LPD. Herein, we aimed to build AI models to recognize the surgical phase in LPD and explore the performance characteristics of AI models.

METHODS

Among 69 LPD videos from a single surgical team, we used 42 in the building group to establish the models and used the remaining 27 videos in the analysis group to assess the models' performance characteristics. We annotated 13 surgical phases of LPD, including 4 key phases and 9 necessary phases. Two minimal invasive pancreatic surgeons annotated all the videos. We built two AI models for the key phase and necessary phase recognition, based on convolutional neural networks. The overall performance of the AI models was determined mainly by mean average precision (mAP).

RESULTS

Overall mAPs of the AI models in the test set of the building group were 89.7% and 84.7% for key phases and necessary phases, respectively. In the 27-video analysis group, overall mAPs were 86.8% and 71.2%, with maximum mAPs of 98.1% and 93.9%. We found commonalities between the error of model recognition and the differences of surgeon annotation, and the AI model exhibited bad performance in cases with anatomic variation or lesion involvement with adjacent organs.

CONCLUSIONS

AI automated surgical phase recognition can be achieved in LPD, with outstanding performance in selective cases. This breakthrough may be the first step toward AI- and video-based surgical education in more complex surgeries.

Introducing surgical intelligence in gynecology: Automated identification of key steps in hysterectomy

OBJECTIVE

The analysis of surgical videos using artificial intelligence holds great promise for the future of surgery by facilitating the development of surgical best practices, identifying key pitfalls, enhancing situational awareness, and disseminating that information via real-time, intraoperative decision-making. The objective of the present study was to examine the feasibility and accuracy of a novel computer vision algorithm for hysterectomy surgical step identification.

METHODS

This was a retrospective study conducted on surgical videos of laparoscopic hysterectomies performed in 277 patients in five medical centers. We used a surgical intelligence platform (Theator Inc.) that employs advanced computer vision and AI technology to automatically capture video data during surgery, deidentify, and upload procedures to a secure cloud infrastructure. Videos were manually annotated with sequential steps of surgery by a team of annotation specialists. Subsequently, a computer vision system was trained to perform automated step detection in hysterectomy. Analyzing automated video annotations in comparison to manual human annotations was used to determine accuracy.

RESULTS

The mean duration of the videos was 103 ± 43 min. Accuracy between AI-based predictions and manual human annotations was 93.1% on average. Accuracy was highest for the dissection and mobilization step (96.9%) and lowest for the adhesiolysis step (70.3%).

CONCLUSION

The results of the present study demonstrate that a novel AI-based model achieves high accuracy for automated steps identification in hysterectomy. This lays the foundations for the next phase of AI, focused on real-time clinical decision support and prediction of outcome measures, to optimize surgeon workflow and elevate patient care.

Smart Operating Room in Digestive Surgery: A Narrative Review

The introduction of new technologies in current digestive surgical practice is progressively reshaping the operating room, defining the fourth surgical revolution. The implementation of black boxes and control towers aims at streamlining workflow and reducing surgical error by early identification and analysis, while augmented reality and artificial intelligence augment surgeons' perceptual and technical skills by superimposing three-dimensional models to real-time surgical images. Moreover, the operating room architecture is transitioning toward an integrated digital environment to improve efficiency and, ultimately, patients' outcomes. This narrative review describes the most recent evidence regarding the role of these technologies in transforming the current digestive surgical practice, underlining their potential benefits and drawbacks in terms of efficiency and patients' outcomes, as an attempt to foresee the digestive surgical practice of tomorrow.

Instrument-Tissue Interaction Detection Framework for Surgical Video Understanding

Instrument-tissue interaction detection task, which helps understand surgical activities, is vital for constructing computer-assisted surgery systems but with many challenges. Firstly, most models represent instrument-tissue interaction in a coarse-grained way which only focuses on classification and lacks the ability to automatically detect instruments and tissues. Secondly, existing works do not fully consider relations between intra- and inter-frame of instruments and tissues. In the paper, we propose to represent instrument-tissue interaction as 〈 instrument class, instrument bounding box, tissue class, tissue bounding box, action class 〉 quintuple and present an Instrument-Tissue Interaction Detection Network (ITIDNet) to detect the quintuple for surgery videos understanding. Specifically, we propose a Snippet Consecutive Feature (SCF) Layer to enhance features by modeling relationships of proposals in the current frame using global context information in the video snippet. We also propose a Spatial Corresponding Attention (SCA) Layer to incorporate features of proposals between adjacent frames through spatial encoding. To reason relationships between instruments and tissues, a Temporal Graph (TG) Layer is proposed with intra-frame connections to exploit relationships between instruments and tissues in the same frame and inter-frame connections to model the temporal information for the same instance. For evaluation, we build a cataract surgery video (PhacoQ) dataset and a cholecystectomy surgery video (CholecQ) dataset. Experimental results demonstrate the promising performance of our model, which outperforms other state-of-the-art models on both datasets.

A surgical activity model of laparoscopic cholecystectomy for co-operation with collaborative robots

BACKGROUND

Laparoscopic cholecystectomy is a very frequent surgical procedure. However, in an ageing society, less surgical staff will need to perform surgery on patients. Collaborative surgical robots (cobots) could address surgical staff shortages and workload. To achieve context-awareness for surgeon-robot collaboration, the intraoperative action workflow recognition is a key challenge.

METHODS

A surgical process model was developed for intraoperative surgical activities including actor, instrument, action and target in laparoscopic cholecystectomy (excluding camera guidance). These activities, as well as instrument presence and surgical phases were annotated in videos of laparoscopic cholecystectomy performed on human patients (n = 10) and on explanted porcine livers (n = 10). The machine learning algorithm Distilled-Swin was trained on our own annotated dataset and the CholecT45 dataset. The validation of the model was conducted using a fivefold cross-validation approach.

RESULTS

In total, 22,351 activities were annotated with a cumulative duration of 24.9 h of video segments. The machine learning algorithm trained and validated on our own dataset scored a mean average precision (mAP) of 25.7% and a top K = 5 accuracy of 85.3%. With training and validation on our dataset and CholecT45, the algorithm scored a mAP of 37.9%.

CONCLUSIONS

An activity model was developed and applied for the fine-granular annotation of laparoscopic cholecystectomies in two surgical settings. A machine recognition algorithm trained on our own annotated dataset and CholecT45 achieved a higher performance than training only on CholecT45 and can recognize frequently occurring activities well, but not infrequent activities. The analysis of an annotated dataset allowed for the quantification of the potential of collaborative surgical robots to address the workload of surgical staff. If collaborative surgical robots could grasp and hold tissue, up to 83.5% of the assistant's tissue interacting tasks (i.e. excluding camera guidance) could be performed by robots.

SimCol3D - 3D reconstruction during colonoscopy challenge

Colorectal cancer is one of the most common cancers in the world. While colonoscopy is an effective screening technique, navigating an endoscope through the colon to detect polyps is challenging. A 3D map of the observed surfaces could enhance the identification of unscreened colon tissue and serve as a training platform. However, reconstructing the colon from video footage remains difficult. Learning-based approaches hold promise as robust alternatives, but necessitate extensive datasets. Establishing a benchmark dataset, the 2022 EndoVis sub-challenge SimCol3D aimed to facilitate data-driven depth and pose prediction during colonoscopy. The challenge was hosted as part of MICCAI 2022 in Singapore. Six teams from around the world and representatives from academia and industry participated in the three sub-challenges: synthetic depth prediction, synthetic pose prediction, and real pose prediction. This paper describes the challenge, the submitted methods, and their results. We show that depth prediction from synthetic colonoscopy images is robustly solvable, while pose estimation remains an open research question.

A Concept for Preoperative and Intraoperative Molecular Imaging and Detection for Assessing Extent of Disease of Solid Tumors

The authors propose a concept of "systems engineering," the approach to assessing the extent of diseased tissue (EODT) in solid tumors. We modeled the proof of this concept based on our clinical experience with colorectal carcinoma (CRC) and gastrinoma that included short and long-term survival data of CRC patients. This concept, applicable to various solid tumors, combines resources from surgery, nuclear medicine, radiology, pathology, and oncology needed for preoperative and intraoperative assessments of a patient's EODT. The concept begins with a patient presenting with biopsy-proven cancer. An appropriate preferential locator (PL) is a molecule that preferentially binds to a cancer-related molecular target (i.e., tumor marker) lacking in non-malignant tissue and is the essential element. Detecting the PL after an intravenous injection requires the PL labeling with an appropriate tracer radionuclide, a fluoroprobe, or both. Preoperative imaging of the tracer's signal requires molecular imaging modalities alone or in combination with computerized tomography (CT). These include positron emission tomography (PET), PET/CT, single-photon emission computed tomography (SPECT), SPECT/CT for preoperative imaging, gamma cameras for intraoperative imaging, and gamma-detecting probes for precise localization. Similarly, fluorescent-labeled PLs require appropriate cameras and probes. This approach provides the surgeon with real-time information needed for R0 resection.

Applications of artificial intelligence in surgery: clinical, technical, and governance considerations

Artificial intelligence (AI) will power many of the tools in the armamentarium of digital surgeons. AI methods and surgical proof-of-concept flourish, but we have yet to witness clinical translation and value. Here we exemplify the potential of AI in the care pathway of colorectal cancer patients and discuss clinical, technical, and governance considerations of major importance for the safe translation of surgical AI for the benefit of our patients and practices.

Model-based individual life-spanning documentation in visceral surgery: a proof of concept

INTRODUCTION

Surgical documentation has many implications. However, its primary function is to transfer information about surgical procedures to other medical professionals. Thereby, written reports describing procedures in detail are the current standard, impeding comprehensive understanding of patient-individual life-spanning surgical course, especially if surgeries are performed at a timely distance and in diverse facilities. Therefore, we developed a novel model-based approach for documentation of visceral surgeries, denoted as 'Surgical Documentation Markup-Modeling' (SDM-M).

MATERIAL AND METHODS

For scientific evaluation, we developed a web-based prototype software allowing for creating hierarchical anatomical models that can be modified by individual surgery-related markup information. Thus, a patient's cumulated 'surgical load' can be displayed on a timeline deploying interactive anatomical 3D models. To evaluate the possible impact on daily clinical routine, we performed an evaluation study with 24 surgeons and advanced medical students, elaborating on simulated complex surgical cases, once with classic written reports and once with our prototypical SDM-M software.

RESULTS

Leveraging SDM-M in an experimental environment reduced the time needed for elaborating simulated complex surgical cases from 354 ± 85 s with the classic approach to 277 ± 128 s. (p = 0.00109) The perceived task load measured by the Raw NASA-TLX was reduced significantly (p = 0.00003) with decreased mental (p = 0.00004) and physical (p = 0.01403) demand. Also, time demand (p = 0.00041), performance (p = 0.00161), effort (p = 0.00024), and frustration (p = 0.00031) were improved significantly.

DISCUSSION

Model-based approaches for life-spanning surgical documentation could improve the daily clinical elaboration and understanding of complex cases in visceral surgery. Besides reduced workload and time sparing, even a more structured assessment of individual surgical cases could foster improved planning of further surgeries, information transfer, and even scientific evaluation, considering the cumulative 'surgical load.'

CONCLUSION

Life-spanning model-based documentation of visceral surgical cases could significantly improve surgery and workload.

Audio-based event detection in the operating room

PURPOSE

Even though workflow analysis in the operating room has come a long way, current systems are still limited to research. In the quest for a robust, universal setup, hardly any attention has been given to the dimension of audio despite its numerous advantages, such as low costs, location, and sight independence, or little required processing power.

METHODOLOGY

We present an approach for audio-based event detection that solely relies on two microphones capturing the sound in the operating room. Therefore, a new data set was created with over 63 h of audio recorded and annotated at the University Hospital rechts der Isar. Sound files were labeled, preprocessed, augmented, and subsequently converted to log-mel-spectrograms that served as a visual input for an event classification using pretrained convolutional neural networks.

RESULTS

Comparing multiple architectures, we were able to show that even lightweight models, such as MobileNet, can already provide promising results. Data augmentation additionally improved the classification of 11 defined classes, including inter alia different types of coagulation, operating table movements as well as an idle class. With the newly created audio data set, an overall accuracy of 90%, a precision of 91% and a F1-score of 91% were achieved, demonstrating the feasibility of an audio-based event recognition in the operating room.

CONCLUSION

With this first proof of concept, we demonstrated that audio events can serve as a meaningful source of information that goes beyond spoken language and can easily be integrated into future workflow recognition pipelines using computational inexpensive architectures.

EndoViT: pretraining vision transformers on a large collection of endoscopic images

PURPOSE

Automated endoscopy video analysis is essential for assisting surgeons during medical procedures, but it faces challenges due to complex surgical scenes and limited annotated data. Large-scale pretraining has shown great success in natural language processing and computer vision communities in recent years. These approaches reduce the need for annotated data, which is of great interest in the medical domain. In this work, we investigate endoscopy domain-specific self-supervised pretraining on large collections of data.

METHODS

To this end, we first collect Endo700k, the largest publicly available corpus of endoscopic images, extracted from nine public Minimally Invasive Surgery (MIS) datasets. Endo700k comprises more than 700,000 images. Next, we introduce EndoViT, an endoscopy-pretrained Vision Transformer (ViT), and evaluate it on a diverse set of surgical downstream tasks.

RESULTS

Our findings indicate that domain-specific pretraining with EndoViT yields notable advantages in complex downstream tasks. In the case of action triplet recognition, our approach outperforms ImageNet pretraining. In semantic segmentation, we surpass the state-of-the-art (SOTA) performance. These results demonstrate the effectiveness of our domain-specific pretraining approach in addressing the challenges of automated endoscopy video analysis.

CONCLUSION

Our study contributes to the field of medical computer vision by showcasing the benefits of domain-specific large-scale self-supervised pretraining for vision transformers. We release both our code and pretrained models to facilitate further research in this direction: https://github.com/DominikBatic/EndoViT .

Kirchhoff's law-based velocity-controlled motion models to predict real-time cutting forces in minimally invasive surgeries

A theoretical framework, united by a "system effect" is formulated to model the cutting/haptic force evolution at the cutting edge of a surgical cutting instrument during its penetration into soft biological tissue in minimally invasive surgery. Other cutting process responses, including tissue fracture force, friction force, and damping, are predicted by the model as well. The model is based on a velocity-controlled formulation of the corresponding equations of motion, derived for a surgical cutting instrument and tissue based on Kirchhoff's fundamental energy conservation law. It provides nearly zero residues (absolute errors) in the equations of motion balances. In addition, concurrent closing relationships for the fracture force, friction coefficient, friction force, process damping, strain rate function (a constitutive tissue model), and their implementation within the proposed theoretical framework are established. The advantage of the method is its ability to make precise real-time predictions of the aperiodic fluctuating evolutions of the cutting forces and the other process responses. It allows for the robust modeling of the interactions between a medical instrument and a nonlinear viscoelastic tissue under any physically feasible working conditions. The cutting process model was partially qualitatively verified through numerical simulations and by comparing the computed cutting forces with experimentally measured values during robotic uniaxial biopsy needle constant velocity insertion into artificial gel tissue, obtained from previous experimental research. The comparison has shown a qualitatively similar adequate trend in the evolution of the experimentally measured and numerically predicted cutting forces during insertion of the needle.

Multimodal semi-supervised learning for online recognition of multi-granularity surgical workflows

Purpose Surgical workflow recognition is a challenging task that requires understanding multiple aspects of surgery, such as gestures, phases, and steps. However, most existing methods focus on single-task or single-modal models and rely on costly annotations for training. To address these limitations, we propose a novel semi-supervised learning approach that leverages multimodal data and self-supervision to create meaningful representations for various surgical tasks. Methods Our representation learning approach conducts two processes. In the first stage, time contrastive learning is used to learn spatiotemporal visual features from video data, without any labels. In the second stage, multimodal VAE fuses the visual features with kinematic data to obtain a shared representation, which is fed into recurrent neural networks for online recognition. Results Our method is evaluated on two datasets: JIGSAWS and MISAW. We confirmed that it achieved comparable or better performance in multi-granularity workflow recognition compared to fully supervised models specialized for each task. On the JIGSAWS Suturing dataset, we achieve a gesture recognition accuracy of 83.3%. In addition, our model is more efficient in annotation usage, as it can maintain high performance with only half of the labels. On the MISAW dataset, we achieve 84.0% AD-Accuracy in phase recognition and 56.8% AD-Accuracy in step recognition. Conclusion Our multimodal representation exhibits versatility across various surgical tasks and enhances annotation efficiency. This work has significant implications for real-time decision-making systems within the operating room.

Exploring semantic consistency in unpaired image translation to generate data for surgical applications

PURPOSE

In surgical computer vision applications, data privacy and expert annotation challenges impede the acquisition of labeled training data. Unpaired image-to-image translation techniques have been explored to automatically generate annotated datasets by translating synthetic images into a realistic domain. The preservation of structure and semantic consistency, i.e., per-class distribution during translation, poses a significant challenge, particularly in cases of semantic distributional mismatch.

METHOD

This study empirically investigates various translation methods for generating data in surgical applications, explicitly focusing on semantic consistency. Through our analysis, we introduce a novel and simple combination of effective approaches, which we call ConStructS. The defined losses within this approach operate on multiple image patches and spatial resolutions during translation.

RESULTS

Various state-of-the-art models were extensively evaluated on two challenging surgical datasets. With two different evaluation schemes, the semantic consistency and the usefulness of the translated images on downstream semantic segmentation tasks were evaluated. The results demonstrate the effectiveness of the ConStructS method in minimizing semantic distortion, with images generated by this model showing superior utility for downstream training.

CONCLUSION

In this study, we tackle semantic inconsistency in unpaired image translation for surgical applications with minimal labeled data. The simple model (ConStructS) enhances consistency during translation and serves as a practical way of generating fully labeled and semantically consistent datasets at minimal cost. Our code is available at https://gitlab.com/nct_tso_public/constructs .

[The digital operating room : Chances and risks of artificial intelligence]

At the central workplace of the surgeon the digitalization of the operating room has particular consequences for the surgical work. Starting with intraoperative cross-sectional imaging and sonography, through functional imaging, minimally invasive and robot-assisted surgery up to digital surgical and anesthesiological documentation, the vast majority of operating rooms are now at least partially digitalized. The increasing digitalization of the whole process chain enables not only for the collection but also the analysis of big data. Current research focuses on artificial intelligence for the analysis of intraoperative data as the prerequisite for assistance systems that support surgical decision making or warn of risks; however, these technologies raise new ethical questions for the surgical community that affect the core of surgical work.

On the pitfalls of Batch Normalization for end-to-end video learning: A study on surgical workflow analysis

Batch Normalization's (BN) unique property of depending on other samples in a batch is known to cause problems in several tasks, including sequence modeling. Yet, BN-related issues are hardly studied for long video understanding, despite the ubiquitous use of BN in CNNs (Convolutional Neural Networks) for feature extraction. Especially in surgical workflow analysis, where the lack of pretrained feature extractors has led to complex, multi-stage training pipelines, limited awareness of BN issues may have hidden the benefits of training CNNs and temporal models end to end. In this paper, we analyze pitfalls of BN in video learning, including issues specific to online tasks such as a 'cheating' effect in anticipation. We observe that BN's properties create major obstacles for end-to-end learning. However, using BN-free backbones, even simple CNN-LSTMs beat the state of the art on three surgical workflow benchmarks by utilizing adequate end-to-end training strategies which maximize temporal context. We conclude that awareness of BN's pitfalls is crucial for effective end-to-end learning in surgical tasks. By reproducing results on natural-video datasets, we hope our insights will benefit other areas of video learning as well. Code is available at: https://gitlab.com/nct_tso_public/pitfalls_bn.

Holistic OR domain modeling: a semantic scene graph approach

PURPOSE

Surgical procedures take place in highly complex operating rooms (OR), involving medical staff, patients, devices and their interactions. Until now, only medical professionals are capable of comprehending these intricate links and interactions. This work advances the field toward automated, comprehensive and semantic understanding and modeling of the OR domain by introducing semantic scene graphs (SSG) as a novel approach to describing and summarizing surgical environments in a structured and semantically rich manner.

METHODS

We create the first open-source 4D SSG dataset. 4D-OR includes simulated total knee replacement surgeries captured by RGB-D sensors in a realistic OR simulation center. It includes annotations for SSGs, human and object pose, clinical roles and surgical phase labels. We introduce a neural network-based SSG generation pipeline for semantic reasoning in the OR and apply our approach to two downstream tasks: clinical role prediction and surgical phase recognition.

RESULTS

We show that our pipeline can successfully reason within the OR domain. The capabilities of our scene graphs are further highlighted by their successful application to clinical role prediction and surgical phase recognition tasks.

CONCLUSION

This work paves the way for multimodal holistic operating room modeling, with the potential to significantly enhance the state of the art in surgical data analysis, such as enabling more efficient and precise decision-making during surgical procedures, and ultimately improving patient safety and surgical outcomes. We release our code and dataset at github.com/egeozsoy/4D-OR.

Artificial intelligence in surgery

Artificial intelligence (AI) is rapidly emerging in healthcare, yet applications in surgery remain relatively nascent. Here we review the integration of AI in the field of surgery, centering our discussion on multifaceted improvements in surgical care in the preoperative, intraoperative and postoperative space. The emergence of foundation model architectures, wearable technologies and improving surgical data infrastructures is enabling rapid advances in AI interventions and utility. We discuss how maturing AI methods hold the potential to improve patient outcomes, facilitate surgical education and optimize surgical care. We review the current applications of deep learning approaches and outline a vision for future advances through multimodal foundation models.

Adoption of routine surgical video recording: a nationwide freedom of information act request across England and Wales

Background

Surgical video contains data with significant potential to improve surgical outcome assessment, quality assurance, education, and research. Current utilisation of surgical video recording is unknown and related policies/governance structures are unclear.

Methods

A nationwide Freedom of Information (FOI) request concerning surgical video recording, technology, consent, access, and governance was sent to all acute National Health Service (NHS) trusts/boards in England/Wales between 20th February and 20th March 2023.

Findings

140/144 (97.2%) trusts/boards in England/Wales responded to the FOI request. Surgical procedures were routinely recorded in 22 trusts/boards. The median estimate of consultant surgeons routinely recording their procedures was 20%. Surgical video was stored on internal systems (n = 27), third-party products (n = 29), and both (n = 9). 32/140 (22.9%) trusts/boards ask for consent to record procedures as part of routine care. Consent for recording included non-clinical purposes in 55/140 (39.3%) trusts/boards. Policies for surgeon/patient access to surgical video were available in 48/140 (34.3%) and 32/140 (22.9%) trusts/boards, respectively. Surgical video was used for non-clinical purposes in 64/140 (45.7%) trusts/boards. Governance policies covering surgical video recording, use, and/or storage were available from 59/140 (42.1%) trusts/boards.

Interpretation

There is significant heterogeneity in surgical video recording practices in England and Wales. A minority of trusts/boards routinely record surgical procedures, with large variation in recording/storage practices indicating scope for NHS-wide coordination. Revision of surgical video consent, accessibility, and governance policies should be prioritised by trusts/boards to protect key stakeholders. Increased availability of surgical video is essential for patients and surgeons to maximally benefit from the ongoing digital transformation of surgery.

Funding

KL is supported by an NIHR Academic Clinical Fellowship and acknowledges infrastructure support for this research from the National Institute for Health Research (NIHR) Imperial Biomedical Research Centre (BRC).

MT4MTL-KD: A Multi-Teacher Knowledge Distillation Framework for Triplet Recognition

The recognition of surgical triplets plays a critical role in the practical application of surgical videos. It involves the sub-tasks of recognizing instruments, verbs, and targets, while establishing precise associations between them. Existing methods face two significant challenges in triplet recognition: 1) the imbalanced class distribution of surgical triplets may lead to spurious task association learning, and 2) the feature extractors cannot reconcile local and global context modeling. To overcome these challenges, this paper presents a novel multi-teacher knowledge distillation framework for multi-task triplet learning, known as MT4MTL-KD. MT4MTL-KD leverages teacher models trained on less imbalanced sub-tasks to assist multi-task student learning for triplet recognition. Moreover, we adopt different categories of backbones for the teacher and student models, facilitating the integration of local and global context modeling. To further align the semantic knowledge between the triplet task and its sub-tasks, we propose a novel feature attention module (FAM). This module utilizes attention mechanisms to assign multi-task features to specific sub-tasks. We evaluate the performance of MT4MTL-KD on both the 5-fold cross-validation and the CholecTriplet challenge splits of the CholecT45 dataset. The experimental results consistently demonstrate the superiority of our framework over state-of-the-art methods, achieving significant improvements of up to 6.4% on the cross-validation split.

YOLOv7-RepFPN: Improving real-time performance of laparoscopic tool detection on embedded systems

This study focuses on enhancing the inference speed of laparoscopic tool detection on embedded devices. Laparoscopy, a minimally invasive surgery technique, markedly reduces patient recovery times and postoperative complications. Real-time laparoscopic tool detection helps assisting laparoscopy by providing information for surgical navigation, and its implementation on embedded devices is gaining interest due to the portability, network independence and scalability of the devices. However, embedded devices often face computation resource limitations, potentially hindering inference speed. To mitigate this concern, the work introduces a two-fold modification to the YOLOv7 model: the feature channels and integrate RepBlock is halved, yielding the YOLOv7-RepFPN model. This configuration leads to a significant reduction in computational complexity. Additionally, the focal EIoU (efficient intersection of union) loss function is employed for bounding box regression. Experimental results on an embedded device demonstrate that for frame-by-frame laparoscopic tool detection, the proposed YOLOv7-RepFPN achieved an mAP of 88.2% (with IoU set to 0.5) on a custom dataset based on EndoVis17, and an inference speed of 62.9 FPS. Contrasting with the original YOLOv7, which garnered an 89.3% mAP and 41.8 FPS under identical conditions, the methodology enhances the speed by 21.1 FPS while maintaining detection accuracy. This emphasizes the effectiveness of the work.

Revisiting instrument segmentation: Learning from decentralized surgical sequences with various imperfect annotations

This paper focuses on a new and challenging problem related to instrument segmentation. This paper aims to learn a generalizable model from distributed datasets with various imperfect annotations. Collecting a large-scale dataset for centralized learning is usually impeded due to data silos and privacy issues. Besides, local clients, such as hospitals or medical institutes, may hold datasets with diverse and imperfect annotations. These datasets can include scarce annotations (many samples are unlabelled), noisy labels prone to errors, and scribble annotations with less precision. Federated learning (FL) has emerged as an attractive paradigm for developing global models with these locally distributed datasets. However, its potential in instrument segmentation has yet to be fully investigated. Moreover, the problem of learning from various imperfect annotations in an FL setup is rarely studied, even though it presents a more practical and beneficial scenario. This work rethinks instrument segmentation in such a setting and propose a practical FL framework for this issue. Notably, this approach surpassed centralized learning under various imperfect annotation settings. This method established a foundational benchmark, and future work can build upon it by considering each client owning various annotations and aligning closer with real-world complexities.

Design criteria for AI-based IT systems

PURPOSE

This editorial relates to a panel discussion during the CARS 2023 congress that addressed the question on how AI-based IT systems should be designed that record and (transparently) display a reproducible path on clinical decision making. Even though the software engineering approach suggested for this endeavor is of a generic nature, it is assumed that the listed design criteria are applicable to IT system development also for the domain of radiology and surgery.

METHODS

An example of a possible design approach is outlined by illustrating on how to move from data, information, knowledge and models to wisdom-based decision making in the context of a conceptual GPT system design. In all these design steps, the essential requirements for system quality, information quality, and service quality may be realized by following the design cycle as suggested by A.R. Hevner, appropriately applied to AI-based IT systems design.

RESULTS

It can be observed that certain state-of-the-art AI algorithms and systems, such as large language models or generative pre-trained transformers (GPTs), are becoming increasingly complex and, therefore, need to be rigorously examined to render them transparent and comprehensible in their usage for all stakeholders involved in health care. Further critical questions that need to be addressed are outlined and complemented with some suggestions, that a possible design framework for a stakeholder specific AI system could be a (modest) GPT based on a small language model.

DISCUSSION

A fundamental question for the future remains whether society wants a quasi-wisdom-oriented healthcare system, based on data-driven intelligence with AI, or a human curated wisdom based on model-driven intelligence (with and without AI). Special CARS workshops and think tanks are planned to address this challenging question and possible new direction for assisting selected medical disciplines, e.g., radiology and surgery.

Surgical phase and instrument recognition: how to identify appropriate dataset splits

PURPOSE

Machine learning approaches can only be reliably evaluated if training, validation, and test data splits are representative and not affected by the absence of classes. Surgical workflow and instrument recognition are two tasks that are complicated in this manner, because of heavy data imbalances resulting from different length of phases and their potential erratic occurrences. Furthermore, sub-properties like instrument (co-)occurrence are usually not particularly considered when defining the split.

METHODS

We present a publicly available data visualization tool that enables interactive exploration of dataset partitions for surgical phase and instrument recognition. The application focuses on the visualization of the occurrence of phases, phase transitions, instruments, and instrument combinations across sets. Particularly, it facilitates assessment of dataset splits, especially regarding identification of sub-optimal dataset splits.

RESULTS

We performed analysis of the datasets Cholec80, CATARACTS, CaDIS, M2CAI-workflow, and M2CAI-tool using the proposed application. We were able to uncover phase transitions, individual instruments, and combinations of surgical instruments that were not represented in one of the sets. Addressing these issues, we identify possible improvements in the splits using our tool. A user study with ten participants demonstrated that the participants were able to successfully solve a selection of data exploration tasks.

CONCLUSION

In highly unbalanced class distributions, special care should be taken with respect to the selection of an appropriate dataset split because it can greatly influence the assessments of machine learning approaches. Our interactive tool allows for determination of better splits to improve current practices in the field. The live application is available at https://cardio-ai.github.io/endovis-ml/ .

Human robotic surgery with intraoperative tissue identification using rapid evaporation ionisation mass spectrometry

Instantaneous, continuous, and reliable information on the molecular biology of surgical target tissue could significantly contribute to the precision, safety, and speed of the intervention. In this work, we introduced a methodology for chemical tissue identification in robotic surgery using rapid evaporative ionisation mass spectrometry. We developed a surgical aerosol evacuation system that is compatible with a robotic platform enabling consistent intraoperative sample collection and assessed the feasibility of this platform during head and neck surgical cases, using two different surgical energy devices. Our data showed specific, characteristic lipid profiles associated with the tissue type including various ceramides, glycerophospholipids, and glycerolipids, as well as different ion formation mechanisms based on the energy device used. This platform allows continuous and accurate intraoperative mass spectrometry-based identification of ablated/resected tissue and in combination with robotic registration of images, time, and anatomical positions can improve the current robot-assisted surgical platforms and guide surgical strategy.

Research on workflow recognition for liver rupture repair surgery

Liver rupture repair surgery serves as one tool to treat liver rupture, especially beneficial for cases of mild liver rupture hemorrhage. Liver rupture can catalyze critical conditions such as hemorrhage and shock. Surgical workflow recognition in liver rupture repair surgery videos presents a significant task aimed at reducing surgical mistakes and enhancing the quality of surgeries conducted by surgeons. A liver rupture repair simulation surgical dataset is proposed in this paper which consists of 45 videos collaboratively completed by nine surgeons. Furthermore, an end-to-end SA-RLNet, a self attention-based recurrent convolutional neural network, is introduced in this paper. The self-attention mechanism is used to automatically identify the importance of input features in various instances and associate the relationships between input features. The accuracy of the surgical phase classification of the SA-RLNet approach is 90.6%. The present study demonstrates that the SA-RLNet approach shows strong generalization capabilities on the dataset. SA-RLNet has proved to be advantageous in capturing subtle variations between surgical phases. The application of surgical workflow recognition has promising feasibility in liver rupture repair surgery.

Early-stage clinical evaluation of real-time artificial intelligence assistance for laparoscopic cholecystectomy

The growing availability of surgical digital data and developments in analytics such as artificial intelligence (AI) are being harnessed to improve surgical care. However, technical and cultural barriers to real-time intraoperative AI assistance exist. This early-stage clinical evaluation shows the technical feasibility of concurrently deploying several AIs in operating rooms for real-time assistance during procedures. In addition, potentially relevant clinical applications of these AI models are explored with a multidisciplinary cohort of key stakeholders.

Defining the Standard for Surgical Video Deidentification

This article reviews the implementation of standards for surgical video deidentification.

Self-knowledge distillation for surgical phase recognition

PURPOSE

Advances in surgical phase recognition are generally led by training deeper networks. Rather than going further with a more complex solution, we believe that current models can be exploited better. We propose a self-knowledge distillation framework that can be integrated into current state-of-the-art (SOTA) models without requiring any extra complexity to the models or annotations.

METHODS

Knowledge distillation is a framework for network regularization where knowledge is distilled from a teacher network to a student network. In self-knowledge distillation, the student model becomes the teacher such that the network learns from itself. Most phase recognition models follow an encoder-decoder framework. Our framework utilizes self-knowledge distillation in both stages. The teacher model guides the training process of the student model to extract enhanced feature representations from the encoder and build a more robust temporal decoder to tackle the over-segmentation problem.

RESULTS

We validate our proposed framework on the public dataset Cholec80. Our framework is embedded on top of four popular SOTA approaches and consistently improves their performance. Specifically, our best GRU model boosts performance by [Formula: see text] accuracy and [Formula: see text] F1-score over the same baseline model.

CONCLUSION

We embed a self-knowledge distillation framework for the first time in the surgical phase recognition training pipeline. Experimental results demonstrate that our simple yet powerful framework can improve performance of existing phase recognition models. Moreover, our extensive experiments show that even with 75% of the training set we still achieve performance on par with the same baseline model trained on the full set.

Surgical Phase Duration in Robot-Assisted Partial Nephrectomy: A Surgical Data Science Exploration for Clinical Relevance

(1) Background: Surgical phases form the basic building blocks for surgical skill assessment, feedback, and teaching. The phase duration itself and its correlation with clinical parameters at diagnosis have not yet been investigated. Novel commercial platforms provide phase indications but have not been assessed for accuracy yet. (2) Methods: We assessed 100 robot-assisted partial nephrectomy videos for phase durations based on previously defined proficiency metrics. We developed an annotation framework and subsequently compared our annotations to an existing commercial solution (Touch Surgery, Medtronic™). We subsequently explored clinical correlations between phase durations and parameters derived from diagnosis and treatment. (3) Results: An objective and uniform phase assessment requires precise definitions derived from an iterative revision process. A comparison to a commercial solution shows large differences in definitions across phases. BMI and the duration of renal tumor identification are positively correlated, as are tumor complexity and both tumor excision and renorrhaphy duration. (4) Conclusions: The surgical phase duration can be correlated with certain clinical outcomes. Further research should investigate whether the retrieved correlations are also clinically meaningful. This requires an increase in dataset sizes and facilitation through intelligent computer vision algorithms. Commercial platforms can facilitate this dataset expansion and help unlock the full potential, provided that the phase annotation details are disclosed.

From Nanobots to Neural Networks: Multifaceted Revolution of Artificial Intelligence in Surgical Medicine and Therapeutics

This comprehensive exploration unveils the transformative potential of Artificial Intelligence (AI) within medicine and surgery. Through a meticulous journey, we examine AI's current applications in healthcare, including medical diagnostics, surgical procedures, and advanced therapeutics. Delving into the theoretical foundations of AI, encompassing machine learning, deep learning, and Natural Language Processing (NLP), we illuminate the critical underpinnings supporting AI's integration into healthcare. Highlighting the symbiotic relationship between humans and machines, we emphasize how AI augments clinical capabilities without supplanting the irreplaceable human touch in healthcare delivery. Also, we'd like to briefly mention critical findings and takeaways they can expect to encounter in the article. A thoughtful analysis of the economic, societal, and ethical implications of AI's integration into healthcare underscores our commitment to addressing critical issues, such as data privacy, algorithmic transparency, and equitable access to AI-driven healthcare services. As we contemplate the future landscape, we project an exciting vista where more sophisticated AI algorithms and real-time surgical visualizations redefine the boundaries of medical achievement. While acknowledging the limitations of the present research, we shed light on AI's pivotal role in enhancing patient engagement, education, and data security within the burgeoning realm of AI-driven healthcare.

Visual modalities-based multimodal fusion for surgical phase recognition

Surgical workflow analysis is essential to help optimize surgery by encouraging efficient communication and the use of resources. However, the performance of phase recognition is limited by the use of information related to the presence of surgical instruments. To address the problem, we propose visual modality-based multimodal fusion for surgical phase recognition to overcome the limited diversity of information such as the presence of instruments. Using the proposed methods, we extracted a visual kinematics-based index related to using instruments, such as movement and their interrelations during surgery. In addition, we improved recognition performance using an effective convolutional neural network (CNN)-based fusion method for visual features and a visual kinematics-based index (VKI). The visual kinematics-based index improves the understanding of a surgical procedure since information is related to instrument interaction. Furthermore, these indices can be extracted in any environment, such as laparoscopic surgery, and help obtain complementary information for system kinematics log errors. The proposed methodology was applied to two multimodal datasets, a virtual reality (VR) simulator-based dataset (PETRAW) and a private distal gastrectomy surgery dataset, to verify that it can help improve recognition performance in clinical environments. We also explored the influence of a visual kinematics-based index to recognize each surgical workflow by the instrument's existence and the instrument's trajectory. Through the experimental results of a distal gastrectomy video dataset, we validated the effectiveness of our proposed fusion approach in surgical phase recognition. The relatively simple yet index-incorporated fusion we propose can yield significant performance improvements over only CNN-based training and exhibits effective training results compared to fusion based on Transformers, which require a large amount of pre-trained data.

Deep Learning in Surgical Workflow Analysis: A Review of Phase and Step Recognition

OBJECTIVE

In the last two decades, there has been a growing interest in exploring surgical procedures with statistical models to analyze operations at different semantic levels. This information is necessary for developing context-aware intelligent systems, which can assist the physicians during operations, evaluate procedures afterward or help the management team to effectively utilize the operating room. The objective is to extract reliable patterns from surgical data for the robust estimation of surgical activities performed during operations. The purpose of this article is to review the state-of-the-art deep learning methods that have been published after 2018 for analyzing surgical workflows, with a focus on phase and step recognition.

METHODS

Three databases, IEEE Xplore, Scopus, and PubMed were searched, and additional studies are added through a manual search. After the database search, 343 studies were screened and a total of 44 studies are selected for this review.

CONCLUSION

The use of temporal information is essential for identifying the next surgical action. Contemporary methods used mainly RNNs, hierarchical CNNs, and Transformers to preserve long-distance temporal relations. The lack of large publicly available datasets for various procedures is a great challenge for the development of new and robust models. As supervised learning strategies are used to show proof-of-concept, self-supervised, semi-supervised, or active learning methods are used to mitigate dependency on annotated data.

SIGNIFICANCE

The present study provides a comprehensive review of recent methods in surgical workflow analysis, summarizes commonly used architectures, datasets, and discusses challenges.

Active learning for extracting surgomic features in robot-assisted minimally invasive esophagectomy: a prospective annotation study

BACKGROUND

With Surgomics, we aim for personalized prediction of the patient's surgical outcome using machine-learning (ML) on multimodal intraoperative data to extract surgomic features as surgical process characteristics. As high-quality annotations by medical experts are crucial, but still a bottleneck, we prospectively investigate active learning (AL) to reduce annotation effort and present automatic recognition of surgomic features.

METHODS

To establish a process for development of surgomic features, ten video-based features related to bleeding, as highly relevant intraoperative complication, were chosen. They comprise the amount of blood and smoke in the surgical field, six instruments, and two anatomic structures. Annotation of selected frames from robot-assisted minimally invasive esophagectomies was performed by at least three independent medical experts. To test whether AL reduces annotation effort, we performed a prospective annotation study comparing AL with equidistant sampling (EQS) for frame selection. Multiple Bayesian ResNet18 architectures were trained on a multicentric dataset, consisting of 22 videos from two centers.

RESULTS

In total, 14,004 frames were tag annotated. A mean F1-score of 0.75 ± 0.16 was achieved for all features. The highest F1-score was achieved for the instruments (mean 0.80 ± 0.17). This result is also reflected in the inter-rater-agreement (1-rater-kappa > 0.82). Compared to EQS, AL showed better recognition results for the instruments with a significant difference in the McNemar test comparing correctness of predictions. Moreover, in contrast to EQS, AL selected more frames of the four less common instruments (1512 vs. 607 frames) and achieved higher F1-scores for common instruments while requiring less training frames.

CONCLUSION

We presented ten surgomic features relevant for bleeding events in esophageal surgery automatically extracted from surgical video using ML. AL showed the potential to reduce annotation effort while keeping ML performance high for selected features. The source code and the trained models are published open source.

Machine learning models to predict surgical case duration compared to current industry standards: scoping review

BACKGROUND

Surgical waiting lists have risen dramatically across the UK as a result of the COVID-19 pandemic. The effective use of operating theatres by optimal scheduling could help mitigate this, but this requires accurate case duration predictions. Current standards for predicting the duration of surgery are inaccurate. Artificial intelligence (AI) offers the potential for greater accuracy in predicting surgical case duration. This study aimed to investigate whether there is evidence to support that AI is more accurate than current industry standards at predicting surgical case duration, with a secondary aim of analysing whether the implementation of the models used produced efficiency savings.

METHOD

PubMed, Embase, and MEDLINE libraries were searched through to July 2023 to identify appropriate articles. PRISMA extension for scoping reviews and the Arksey and O'Malley framework were followed. Study quality was assessed using a modified version of the reporting guidelines for surgical AI papers by Farrow et al. Algorithm performance was reported using evaluation metrics.

RESULTS

The search identified 2593 articles: 14 were suitable for inclusion and 13 reported on the accuracy of AI algorithms against industry standards, with seven demonstrating a statistically significant improvement in prediction accuracy (P < 0.05). The larger studies demonstrated the superiority of neural networks over other machine learning techniques. Efficiency savings were identified in a RCT. Significant methodological limitations were identified across most studies.

CONCLUSION

The studies suggest that machine learning and deep learning models are more accurate at predicting the duration of surgery; however, further research is required to determine the best way to implement this technology.

Intelligent surgical workflow recognition for endoscopic submucosal dissection with real-time animal study

Recent advancements in artificial intelligence have witnessed human-level performance; however, AI-enabled cognitive assistance for therapeutic procedures has not been fully explored nor pre-clinically validated. Here we propose AI-Endo, an intelligent surgical workflow recognition suit, for endoscopic submucosal dissection (ESD). Our AI-Endo is trained on high-quality ESD cases from an expert endoscopist, covering a decade time expansion and consisting of 201,026 labeled frames. The learned model demonstrates outstanding performance on validation data, including cases from relatively junior endoscopists with various skill levels, procedures conducted with different endoscopy systems and therapeutic skills, and cohorts from international multi-centers. Furthermore, we integrate our AI-Endo with the Olympus endoscopic system and validate the AI-enabled cognitive assistance system with animal studies in live ESD training sessions. Dedicated data analysis from surgical phase recognition results is summarized in an automatically generated report for skill assessment.

Weakly Supervised Temporal Convolutional Networks for Fine-Grained Surgical Activity Recognition

Automatic recognition of fine-grained surgical activities, called steps, is a challenging but crucial task for intelligent intra-operative computer assistance. The development of current vision-based activity recognition methods relies heavily on a high volume of manually annotated data. This data is difficult and time-consuming to generate and requires domain-specific knowledge. In this work, we propose to use coarser and easier-to-annotate activity labels, namely phases, as weak supervision to learn step recognition with fewer step annotated videos. We introduce a step-phase dependency loss to exploit the weak supervision signal. We then employ a Single-Stage Temporal Convolutional Network (SS-TCN) with a ResNet-50 backbone, trained in an end-to-end fashion from weakly annotated videos, for temporal activity segmentation and recognition. We extensively evaluate and show the effectiveness of the proposed method on a large video dataset consisting of 40 laparoscopic gastric bypass procedures and the public benchmark CATARACTS containing 50 cataract surgeries.

Cannulation Skill Assessment Using Functional Data Analysis

OBJECTIVE

A clinician's operative skill-the ability to safely and effectively perform a procedure-directly impacts patient outcomes and well-being. Therefore, it is necessary to accurately assess skill progression during medical training as well as develop methods to most efficiently train healthcare professionals.

METHODS

In this study, we explore whether time-series needle angle data recorded during cannulation on a simulator can be analyzed using functional data analysis methods to (1) identify skilled versus unskilled performance and (2) relate angle profiles to degree of success of the procedure.

RESULTS

Our methods successfully differentiated between types of needle angle profiles. In addition, the identified profile types were associated with degrees of skilled and unskilled behavior of subjects. Furthermore, the types of variability in the dataset were analyzed, providing particular insight into the overall range of needle angles used as well as the rate of change of angle as cannulation progressed in time. Finally, cannulation angle profiles also demonstrated an observable correlation with degree of cannulation success, a metric that is closely related to clinical outcome.

CONCLUSION

In summary, the methods presented here enable rich assessment of clinical skill since the functional (i.e., dynamic) nature of the data is duly considered.

TRandAugment: temporal random augmentation strategy for surgical activity recognition from videos

PURPOSE

Automatic recognition of surgical activities from intraoperative surgical videos is crucial for developing intelligent support systems for computer-assisted interventions. Current state-of-the-art recognition methods are based on deep learning where data augmentation has shown the potential to improve the generalization of these methods. This has spurred work on automated and simplified augmentation strategies for image classification and object detection on datasets of still images. Extending such augmentation methods to videos is not straightforward, as the temporal dimension needs to be considered. Furthermore, surgical videos pose additional challenges as they are composed of multiple, interconnected, and long-duration activities.

METHODS

This work proposes a new simplified augmentation method, called TRandAugment, specifically designed for long surgical videos, that treats each video as an assemble of temporal segments and applies consistent but random transformations to each segment. The proposed augmentation method is used to train an end-to-end spatiotemporal model consisting of a CNN (ResNet50) followed by a TCN.

RESULTS

The effectiveness of the proposed method is demonstrated on two surgical video datasets, namely Bypass40 and CATARACTS, and two tasks, surgical phase and step recognition. TRandAugment adds a performance boost of 1-6% over previous state-of-the-art methods, that uses manually designed augmentations.

CONCLUSION

This work presents a simplified and automated augmentation method for long surgical videos. The proposed method has been validated on different datasets and tasks indicating the importance of devising temporal augmentation methods for long surgical videos.

P-CSEM: An Attention Module for Improved Laparoscopic Surgical Tool Detection

Minimal invasive surgery, more specifically laparoscopic surgery, is an active topic in the field of research. The collaboration between surgeons and new technologies aims to improve operation procedures as well as to ensure the safety of patients. An integral part of operating rooms modernization is the real-time communication between the surgeon and the data gathered using the numerous devices during surgery. A fundamental tool that can aid surgeons during laparoscopic surgery is the recognition of the different phases during an operation. Current research has shown a correlation between the surgical tools utilized and the present phase of surgery. To this end, a robust surgical tool classifier is desired for optimal performance. In this paper, a deep learning framework embedded with a custom attention module, the P-CSEM, has been proposed to refine the spatial features for surgical tool classification in laparoscopic surgery videos. This approach utilizes convolutional neural networks (CNNs) integrated with P-CSEM attention modules at different levels of the architecture for improved feature refinement. The model was trained and tested on the popular, publicly available Cholec80 database. Results showed that the attention integrated model achieved a mean average precision of 93.14%, and visualizations revealed the ability of the model to adhere more towards features of tool relevance. The proposed approach displays the benefits of integrating attention modules into surgical tool classification models for a more robust and precise detection.

Machine Learning for Decision-Support in Acute Abdominal Pain - Proof of Concept and Central Considerations

Acute abdominal pain is a common presenting symptom in the emergency department and represents heterogeneous causes and diagnoses. There is often a decision to be made regarding emergency surgical care. Machine learning (ML) could be used here as a decision-support and relieve the time and personnel resource shortage.Patients with acute abdominal pain presenting to the Department of Surgery at Bonn University Hospital in 2020 and 2021 were retrospectively analyzed. Clinical parameters as well as laboratory values were used as predictors. After randomly splitting into a training and test data set (ratio 80 to 20), three ML algorithms were comparatively trained and validated. The entire procedure was repeated 20 times.A total of 1357 patients were identified and included in the analysis, with one in five (n = 276, 20.3%) requiring emergency abdominal surgery within 24 hours. Patients operated on were more likely to be male (p = 0.026), older (p = 0.006), had more gastrointestinal symptoms (nausea: p < 0.001, vomiting p < 0.001) as well as a more recent onset of pain (p < 0.001). Tenderness (p < 0.001) and guarding (p < 0.001) were more common in surgically treated patients and blood analyses showed increased inflammation levels (white blood cell count: p < 0.001, CRP: p < 0.001) and onset of organ dysfunction (creatinine: p < 0.014, quick p < 0.001). Of the three trained algorithms, the tree-based methods (h2o random forest and cforest) showed the best performance. The algorithms classified patients, i.e., predicted surgery, with a median AUC ROC of 0.81 and 0.79 and AUC PRC of 0.56 in test sets.A proof-of-concept was achieved with the development of an ML model for predicting timely surgical therapy for acute abdomen. The ML algorithm can be a valuable tool in decision-making. Especially in the context of heavily used medical resources, the algorithm can help to use these scarce resources more effectively. Technological progress, especially regarding artificial intelligence, increasingly enables evidence-based approaches in surgery but requires a strictly interdisciplinary approach. In the future, the use and handling of ML should be integrated into surgical training.