Anterior cervical discectomy and fusion for the treatment of pediatric Hirayama disease.

PURPOSE: Hirayama disease, a rare cervical myelopathy in children and young adults, leads to progressive upper limb weakness and muscle loss. Non-invasive external cervical orthosis has been shown to prevent further neurologic decline; however, this treatment modality has not been successful at restoring neurologic and motor function, especially in long standing cases with significant weakness. The pathophysiology remains not entirely understood, complicating standardized operative guidelines; however, some studies report favorable outcomes with internal fixation. We report a successful surgically treated case of pediatric Hirayama disease, supplemented by a systematic review and collation of reported cases in the literature. METHODS: A review of the literature was performed by searching PubMed, Embase, and Web of Science. Full-length articles were included if they reported clinical data regarding the treatment of at least one patient with Hirayama disease and the neurologic outcome of that treatment. Articles were excluded if they did not provide information on treatment outcomes, were abstract-only publications, or were published in languages other than English. RESULTS: Of the fifteen articles reviewed, 63 patients were described, with 59 undergoing surgery. This encompassed both anterior and posterior spinal procedures and 1 hand tendon transfer. Fifty-five patients, including one from our institution, showed improvement post-treatment. Eleven of these patients were under 18 years old. CONCLUSION: Hirayama disease is an infrequent yet impactful cervical myelopathy with limited high-quality evidence available for optimal treatment. The current literature supports surgical decompression and stabilization as promising interventions. However, comprehensive research is crucial for evolving diagnosis and treatment paradigms.

The role of focused ultrasound for pediatric brain tumors: current insights and future implications on treatment strategies.

INTRODUCTION: Focused ultrasound (FUS) is an innovative and emerging technology for the treatment of adult and pediatric brain tumors and illustrates the intersection of various specialized fields, including neurosurgery, neuro-oncology, radiation oncology, and biomedical engineering. OBJECTIVE: The authors provide a comprehensive overview of the application and implications of FUS in treating pediatric brain tumors, with a special focus on pediatric low-grade gliomas (pLGGs) and the evolving landscape of this technology and its clinical utility. METHODS: The fundamental principles of FUS include its ability to induce thermal ablation or enhance drug delivery through transient blood-brain barrier (BBB) disruption, emphasizing the adaptability of high-intensity focused ultrasound (HIFU) and low-intensity focused ultrasound (LIFU) applications. RESULTS: Several ongoing clinical trials explore the potential of FUS in offering alternative therapeutic strategies for pathologies where conventional treatments fall short, specifically centrally-located benign CNS tumors and diffuse intrinsic pontine glioma (DIPG). A case illustration involving the use of HIFU for pilocytic astrocytoma is presented. CONCLUSION: Discussions regarding future applications of FUS for the treatment of gliomas include improved drug delivery, immunomodulation, radiosensitization, and other technological advancements.

MR-guided focused ultrasound in pediatric neurosurgery: current insights, technical challenges, and lessons learned from 45 treatments at Children's National Hospital.

OBJECTIVE: MR-guided focused ultrasound (MRgFUS) is an evolving technology with numerous present and potential applications in pediatric neurosurgery. The aim of this study was to describe the use of MRgFUS, technical challenges, complications, and lessons learned at a single children's hospital. METHODS: A retrospective analysis was performed of a prospectively collected database of all pediatric patients undergoing investigational use of MRgFUS for treatment of various neurosurgical pathologies at Children's National Hospital. Treatment details, clinical workflow, and standard operating procedures are described. Patient demographics, procedure duration, and complications were obtained through a chart review of anesthesia and operative reports. RESULTS: In total, 45 MRgFUS procedures were performed on 14 patients for treatment of diffuse intrinsic pontine glioma (n = 12), low-grade glioma (n = 1), or secondary dystonia (n = 1) between January 2022 and April 2024. The mean age at treatment was 9 (range 5-22) years, and 64% of the patients were male. With increased experience, the total anesthesia time, sonication time, and change in core body temperature during treatment all significantly decreased. Complications affected 4.4% of patients, including 1 case of scalp edema and 1 patient with a postprocedure epidural hematoma. Device malfunction requiring abortion of the procedure occurred in 1 case (2.2%). Technical challenges related to transducer malfunction and sonication errors occurred in 6.7% and 11.1% of cases, respectively, all overcome by subsequent user modifications. CONCLUSIONS: The authors describe the largest series on MRgFUS technical aspects in pediatric neurosurgery at a single institution, comprising 45 total treatments. This study emphasizes potential technical challenges and provides valuable insights into the nuances of its application in pediatric patients.

Code-free machine learning for object detection in surgical video: a benchmarking, feasibility, and cost study.

OBJECTIVE: While the utilization of machine learning (ML) for data analysis typically requires significant technical expertise, novel platforms can deploy ML methods without requiring the user to have any coding experience (termed AutoML). The potential for these methods to be applied to neurosurgical video and surgical data science is unknown. METHODS: AutoML, a code-free ML (CFML) system, was used to identify surgical instruments contained within each frame of endoscopic, endonasal intraoperative video obtained from a previously validated internal carotid injury training exercise performed on a high-fidelity cadaver model. Instrument-detection performances using CFML were compared with two state-of-the-art ML models built using the Python coding language on the same intraoperative video data set. RESULTS: The CFML system successfully ingested surgical video without the use of any code. A total of 31,443 images were used to develop this model; 27,223 images were uploaded for training, 2292 images for validation, and 1928 images for testing. The mean average precision on the test set across all instruments was 0.708. The CFML model outperformed two standard object detection networks, RetinaNet and YOLOv3, which had mean average precisions of 0.669 and 0.527, respectively, in analyzing the same data set. Significant advantages to the CFML system included ease of use, relatively low cost, displays of true/false positives and negatives in a user-friendly interface, and the ability to deploy models for further analysis with ease. Significant drawbacks of the CFML model included an inability to view the structure of the trained model, an inability to update the ML model once trained with new examples, and the inability for robust downstream analysis of model performance and error modes. CONCLUSIONS: This first report describes the baseline performance of CFML in an object detection task using a publicly available surgical video data set as a test bed. Compared with standard, code-based object detection networks, CFML exceeded performance standards. This finding is encouraging for surgeon-scientists seeking to perform object detection tasks to answer clinical questions, perform quality improvement, and develop novel research ideas. The limited interpretability and customization of CFML models remain ongoing challenges. With the further development of code-free platforms, CFML will become increasingly important across biomedical research. Using CFML, surgeons without significant coding experience can perform exploratory ML analyses rapidly and efficiently.

Robotic and robot-assisted skull base neurosurgery: systematic review of current applications and future directions.

OBJECTIVE: The utility of robotic instrumentation is expanding in neurosurgery. Despite this, successful examples of robotic implementation for endoscopic endonasal or skull base neurosurgery remain limited. Therefore, the authors performed a systematic review of the literature to identify all articles that used robotic systems to access the sella or anterior, middle, or posterior cranial fossae. METHODS: A systematic review of MEDLINE and PubMed in accordance with PRISMA guidelines performed for articles published between January 1, 1990, and August 1, 2021, was conducted to identify all robotic systems (autonomous, semiautonomous, or surgeon-controlled) used for skull base neurosurgical procedures. Cadaveric and human clinical studies were included. Studies with exclusively otorhinolaryngological applications or using robotic microscopes were excluded. RESULTS: A total of 561 studies were identified from the initial search, of which 22 were included following full-text review. Transoral robotic surgery (TORS) using the da Vinci Surgical System was the most widely reported system (4 studies) utilized for skull base and pituitary fossa procedures; additionally, it has been reported for resection of sellar masses in 4 patients. Seven cadaveric studies used the da Vinci Surgical System to access the skull base using alternative, non-TORS approaches (e.g., transnasal, transmaxillary, and supraorbital). Five cadaveric studies investigated alternative systems to access the skull base. Six studies investigated the use of robotic endoscope holders. Advantages to robotic applications in skull base neurosurgery included improved lighting and 3D visualization, replication of more traditional gesture-based movements, and the ability for dexterous movements ordinarily constrained by small operative corridors. Limitations included the size and angulation capacity of the robot, lack of drilling components preventing fully robotic procedures, and cost. Robotic endoscope holders may have been particularly advantageous when the use of a surgical assistant or second surgeon was limited. CONCLUSIONS: Robotic skull base neurosurgery has been growing in popularity and feasibility, but significant limitations remain. While robotic systems seem to have allowed for greater maneuverability and 3D visualization, their size and lack of neurosurgery-specific tools have continued to prevent widespread adoption into current practice. The next generation of robotic technologies should prioritize overcoming these limitations.

Late Presenting Multi-Suture Craniosynostosis.

ABSTRACT: The authors provide the case of a 6-year-old male who presented late with multi-suture craniosynostosis and chronically elevated intracranial pressures (ICPs). He was surgically managed with frontal orbital advancement. This particular case illustrates the significant bleeding and unique bony pathology that can occur in patients with high ICP with concomitant venous collateralization. At 1-month follow-up, he demonstrated significant improvement with maintained expansion and no signs of elevated ICP despite delayed intervention. Frontal orbital advancement serves as an effective method for cranial vault expansion and correction of frontal deformities caused by craniosynostosis.

A systematic review of virtual reality for the assessment of technical skills in neurosurgery.

OBJECTIVE: Virtual reality (VR) and augmented reality (AR) systems are increasingly available to neurosurgeons. These systems may provide opportunities for technical rehearsal and assessments of surgeon performance. The assessment of neurosurgeon skill in VR and AR environments and the validity of VR and AR feedback has not been systematically reviewed. METHODS: A systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines was conducted through MEDLINE and PubMed. Studies published in English between January 1990 and February 2021 describing the use of VR or AR to quantify surgical technical performance of neurosurgeons without the use of human raters were included. The types and categories of automated performance metrics (APMs) from each of these studies were recorded. RESULTS: Thirty-three VR studies were included in the review; no AR studies met inclusion criteria. VR APMs were categorized as either distance to target, force, kinematics, time, blood loss, or volume of resection. Distance and time were the most well-studied APM domains, although all domains were effective at differentiating surgeon experience levels. Distance was successfully used to track improvements with practice. Examining volume of resection demonstrated that attending surgeons removed less simulated tumor but preserved more normal tissue than trainees. More recently, APMs have been used in machine learning algorithms to predict level of training with a high degree of accuracy. Key limitations to enhanced-reality systems include limited AR usage for automated surgical assessment and lack of external and longitudinal validation of VR systems. CONCLUSIONS: VR has been used to assess surgeon performance across a wide spectrum of domains. The VR environment can be used to quantify surgeon performance, assess surgeon proficiency, and track training progression. AR systems have not yet been used to provide metrics for surgeon performance assessment despite potential for intraoperative integration. VR-based APMs may be especially useful for metrics that are difficult to assess intraoperatively, including blood loss and extent of resection.

Use of optical fluorescence agents during surgery for pituitary adenomas: current state of the field.

INTRODUCTION: Differentiation of normal pituitary from abnormal tumor tissue remains a surgical challenge despite improvements in optical visualization technology for pituitary adenoma (PA) surgery. During neurosurgical procedures for other tumor types, 5-aminolevulinic acid (5-ALA) has become a focus of investigation based on its high specificity in differentiating tumor tissue. However, the role of 5-ALA and other optical fluorescent agents in PA surgery remains less clear. OBJECTIVE: To perform a systematic review on the use of various optical fluorescent agents in PA surgery. METHOD: Using PRISMA guidelines, a systematic literature review to identify reports describing 5-ALA and other optical agents for fluorescence-guided surgery for PA was performed. Eleven research studies met inclusion criteria and were reviewed. RESULTS: In two studies, 5-ALA was not shown to be effective in aiding PA resection using standard neurosurgical endoscopic/microscopic approaches. 5-ALA photodynamic therapy was evaluated in two in-vitro models with inconsistent results. Intraoperative use of indocyanine green (ICG) concluded with varying results, but showed a tendency towards improved differentiation of functional PA. OTL38 showed potential for intraoperative identification of nonfunctioning PA, particularly in tumors with high folate receptor expression. One study reported clinically useful fluorescence following sodium fluorescein administration. CONCLUSION: We conclude that selected optical fluorescent agents, including ICG and folate receptors, are most likely to hold promise for clinical use in differentiating PA from normal tissue.

Predictors of 30- and 90-day readmission following craniotomy for malignant brain tumors: analysis of nationwide data.

Hospital readmissions are a major contributor to increased health care costs and are associated with worse patient outcomes after neurosurgery. We used the newly released Nationwide Readmissions Database (NRD) to describe the association between patient, hospital and payer factors with 30- and 90-day readmission following craniotomy for malignant brain tumor. All adult inpatients undergoing craniotomy for primary and secondary malignant brain tumors in the NRD from 2013 to 2014 were included. We identified all cause readmissions within 30- and 90-days following craniotomy for tumor, excluding scheduled chemotherapeutic procedures. We used univariate and multivariate models to identify patient, hospital and administrative factors associated with readmission. We identified 27,717 admissions for brain tumor craniotomy in 2013-2014, with 3343 (13.2%) 30-day and 5271 (25.7%) 90-day readmissions. In multivariate analysis, patients with Medicaid and Medicare were more likely to be readmitted at 30- and 90-days compared to privately insured patients. Patients with two or more comorbidities were more likely to be readmitted at 30- and 90-days, and patients discharged to skilled nursing facilities or home health care were associated with increased 90-day readmission rates. Finally, hospital procedural volume above the 75th percentile was associated with decreased 90-day readmission rates. Patients treated at high volume hospitals are less likely to be readmitted at 90-days. Insurance type, non-routine discharge and patient comorbidities are predictors of postoperative non-scheduled readmission. Further studies may elucidate potentially modifiable risk factors when attempting to improve outcomes and reduce cost associated with brain tumor surgery.

Management of aggressive growth hormone secreting pituitary adenomas.

Aggressive GH-secreting pituitary adenomas (GHPAs) represent an important clinical problem in patients with acromegaly. Surgical therapy, although often the mainstay of treatment for GHPAs, is less effective in aggressive GHPAs due to their invasive and destructive growth patterns, and their proclivity for infrasellar invasion. Medical therapies for GHPAs, including somatostatin analogues and GH receptor antagonists, are becoming increasingly important adjuncts to surgical intervention. Stereotactic radiosurgery serves as an important fallback therapy for tumors that cannot be cured with surgery and medications. Data suggests that patients with aggressive and refractory GHPAs are best treated at dedicated tertiary pituitary centers with multidisciplinary teams of neuroendocrinologists, neurosurgeons, radiation oncologists and other specialists who routinely provide advanced care to GHPA patients. Future research will help clarify the defining features of "aggressive" and "atypical" PAs, likely based on tumor behavior, preoperative imaging characteristics, histopathological characteristics, and molecular markers.

Socioeconomic Disparities Affecting the Presentation and Outcomes in Pediatric Subdural Empyema Patients.

BACKGROUND AND OBJECTIVES: Pediatric subdural empyemas (SDE) carry significant morbidity and mortality, and prompt diagnosis and treatment are essential to ensure optimal outcomes. Nonclinical factors affect presentation, time to diagnosis, and outcomes in several neurosurgical conditions and are potential causes of delay in presentation and treatment for patients with SDE. To evaluate whether socioeconomic status, race, and insurance status affect presentation, time to diagnosis, and outcomes for children with subdural empyema. METHODS: We conducted a retrospective cohort study with patients diagnosed with SDE between 2005 and 2020 at our institution. Information regarding demographics (age, sex, zip code, insurance status, race/ethnicity) and presentation (symptoms, number of prior visits, duration of symptoms) was collected. Outcome measures included mortality, postoperative complications, length of stay, and discharge disposition. RESULTS: 42 patients were diagnosed with SDE with a mean age of 9.5 years. Most (85.7%) (n = 36) were male ( P = .0004), and a majority, 28/42 (66.7%), were African American ( P < .0001). There was no significant difference in socioeconomic status based on zip codes, although a significantly higher number of patients were on public insurance ( P = .015). African American patients had a significantly longer duration of symptoms than their Caucasian counterparts (8.4 days vs 1.8 days P = .0316). In total, 41/42 underwent surgery for the SDE, most within 24 hours of initial neurosurgical evaluation. There were no significant differences in the average length of stay. The average length of antibiotic duration was 57.2 days and was similar for all patients. There were no significant differences in discharge disposition based on any of the factors identified with most of the patients (52.4%) being discharged to home. There was 1 mortality (2.4%). CONCLUSION: Although there were no differences in outcomes based on nonclinical factors, African American men on public insurance bear a disproportionately high burden of SDE. Further investigation into the causes of this is warranted.

A microdiscectomy surgical video annotation framework for supervised machine learning applications.

PURPOSE: Lumbar discectomy is among the most common spine procedures in the US, with 300,000 procedures performed each year. Like other surgical procedures, this procedure is not excluded from potential complications. This paper presents a video annotation methodology for microdiscectomy including the development of a surgical workflow. In future work, this methodology could be combined with computer vision and machine learning models to predict potential adverse events. These systems would monitor the intraoperative activities and possibly anticipate the outcomes. METHODS: A necessary step in supervised machine learning methods is video annotation, which involves labeling objects frame-by-frame to make them recognizable for machine learning applications. Microdiscectomy video recordings of spine surgeries were collected from a multi-center research collaborative. These videos were anonymized and stored in a cloud-based platform. Videos were uploaded to an online annotation platform. An annotation framework was developed based on literature review and surgical observations to ensure proper understanding of the instruments, anatomy, and steps. RESULTS: An annotated video of microdiscectomy was produced by a single surgeon. Multiple iterations allowed for the creation of an annotated video complete with labeled surgical tools, anatomy, and phases. In addition, a workflow was developed for the training of novice annotators, which provides information about the annotation software to assist in the production of standardized annotations. CONCLUSIONS: A standardized workflow for managing surgical video data is essential for surgical video annotation and machine learning applications. We developed a standard workflow for annotating surgical videos for microdiscectomy that may facilitate the quantitative analysis of videos using supervised machine learning applications. Future work will demonstrate the clinical relevance and impact of this workflow by developing process modeling and outcome predictors.

Simulated outcomes for durotomy repair in minimally invasive spine surgery.

Minimally invasive spine surgery (MISS) is increasingly performed using endoscopic and microscopic visualization, and the captured video can be used for surgical education and development of predictive artificial intelligence (AI) models. Video datasets depicting adverse event management are also valuable, as predictive models not exposed to adverse events may exhibit poor performance when these occur. Given that no dedicated spine surgery video datasets for AI model development are publicly available, we introduce Simulated Outcomes for Durotomy Repair in Minimally Invasive Spine Surgery (SOSpine). A validated MISS cadaveric dural repair simulator was used to educate neurosurgery residents, and surgical microscope video recordings were paired with outcome data. Objects including durotomy, needle, grasper, needle driver, and nerve hook were then annotated. Altogether, SOSpine contains 15,698 frames with 53,238 annotations and associated durotomy repair outcomes. For validation, an AI model was fine-tuned on SOSpine video and detected surgical instruments with a mean average precision of 0.77. In summary, SOSpine depicts spine surgeons managing a common complication, providing opportunities to develop surgical AI models.

Potential Applications of Artificial Intelligence and Machine Learning in Spine Surgery Across the Continuum of Care.

The worlds of spinal surgery and computational science are intersecting at the nexus of the operating room and across the continuum of patient care. As medicine moves toward digitizing all aspects of a patient's care, immense amounts of patient data generated and aggregated across surgeons, procedures, and institutions will enable previously inaccessible computationally driven insights. These early insights from artificial intelligence (AI) and machine learning (ML)-enabled technologies are beginning to transform medicine and surgery. The complex pathologies facing spine surgeons and their patients require integrative, multimodal, data-driven management strategies. As these data and the technological tools to computationally process them become increasingly available to spine surgeons, AI and ML methods will inform patient selection, preoperatively risk-stratify patients based on myriad factors, and inform interoperative surgical decisions. Once these tools enter early clinical practice, their use creates a virtual flywheel whereby the use of these tools generates additional data that further accelerate the evolution of computational "knowledge" systems. At this digital crossroads, interested and motivated surgeons have an opportunity to understand these technologies, guide their application toward optimal care, and advocate for opportunities where these powerful new tools can deliver step changes in efficiency, accuracy, and intelligence. In the present article, we review the nomenclature and basics of AI and ML and highlight the current and future applications of these technologies across the care continuum of spinal surgery.

Pilot Analysis of Surgeon Instrument Utilization Signatures Based on Shannon Entropy and Deep Learning for Surgeon Performance Assessment in a Cadaveric Carotid Artery Injury Control Simulation.

BACKGROUND AND OBJECTIVES: Assessment and feedback are critical to surgical education, but direct observational feedback by experts is rarely provided because of time constraints and is typically only qualitative. Automated, video-based, quantitative feedback on surgical performance could address this gap, improving surgical training. The authors aim to demonstrate the ability of Shannon entropy (ShEn), an information theory metric that quantifies series diversity, to predict surgical performance using instrument detections generated through deep learning. METHODS: Annotated images from a publicly available video data set of surgeons managing endoscopic endonasal carotid artery lacerations in a perfused cadaveric simulator were collected. A deep learning model was implemented to detect surgical instruments across video frames. ShEn score for the instrument sequence was calculated from each surgical trial. Logistic regression using ShEn was used to predict hemorrhage control success. RESULTS: ShEn scores and instrument usage patterns differed between successful and unsuccessful trials (ShEn: 0.452 vs 0.370, P < .001). Unsuccessful hemorrhage control trials displayed lower entropy and less varied instrument use patterns. By contrast, successful trials demonstrated higher entropy with more diverse instrument usage and consistent progression in instrument utilization. A logistic regression model using ShEn scores (78% accuracy and 97% average precision) was at least as accurate as surgeons' attending/resident status and years of experience for predicting trial success and had similar accuracy as expert human observers. CONCLUSION: ShEn score offers a summative signal about surgeon performance and predicted success at controlling carotid hemorrhage in a simulated cadaveric setting. Future efforts to generalize ShEn to additional surgical scenarios can further validate this metric.

Pediatric Brain Tissue Segmentation Using a Snapshot Hyperspectral Imaging (sHSI) Camera and Machine Learning Classifier.

Pediatric brain tumors are the second most common type of cancer, accounting for one in four childhood cancer types. Brain tumor resection surgery remains the most common treatment option for brain cancer. While assessing tumor margins intraoperatively, surgeons must send tissue samples for biopsy, which can be time-consuming and not always accurate or helpful. Snapshot hyperspectral imaging (sHSI) cameras can capture scenes beyond the human visual spectrum and provide real-time guidance where we aim to segment healthy brain tissues from lesions on pediatric patients undergoing brain tumor resection. With the institutional research board approval, Pro00011028, 139 red-green-blue (RGB), 279 visible, and 85 infrared sHSI data were collected from four subjects with the system integrated into an operating microscope. A random forest classifier was used for data analysis. The RGB, infrared sHSI, and visible sHSI models achieved average intersection of unions (IoUs) of 0.76, 0.59, and 0.57, respectively, while the tumor segmentation achieved a specificity of 0.996, followed by the infrared HSI and visible HSI models at 0.93 and 0.91, respectively. Despite the small dataset considering pediatric cases, our research leveraged sHSI technology and successfully segmented healthy brain tissues from lesions with a high specificity during pediatric brain tumor resection procedures.

Vertebral Aneurysmal Bone Cyst Mimicking Osteosarcoma: Case Report and Review of the Literature.

Aneurysmal bone cysts are benign osseous lesions containing blood-filled cavities separated by walls of connective tissue. They can be difficult to identify clinically due to similarities in presentation, imaging, and histology with other pathologies. Specifically, it is important to distinguish these benign lesions from malignant processes, as both surgical and medical management differ. We present the case of a 21-year-old patient who presented with impaired motor and sensory function in his lower extremities. Radiologic findings were concerning for an invasive neoplasm, and the intraoperative frozen section supported this conclusion. However, an additional histological investigation was confirmatory for a diagnosis of an aneurysmal bone cyst. The patient underwent corpectomy, laminectomy, and a posterior spinal fusion, and regained motor and sensory function shortly thereafter. This report details the importance of considering aneurysmal bone cysts in the differential of infiltrative bone lesions, despite their benign nature, as medical and surgical management can vary greatly.

A multi-institutional study using artificial intelligence to provide reliable and fair feedback to surgeons.

BACKGROUND: Surgeons who receive reliable feedback on their performance quickly master the skills necessary for surgery. Such performance-based feedback can be provided by a recently-developed artificial intelligence (AI) system that assesses a surgeon's skills based on a surgical video while simultaneously highlighting aspects of the video most pertinent to the assessment. However, it remains an open question whether these highlights, or explanations, are equally reliable for all surgeons. METHODS: Here, we systematically quantify the reliability of AI-based explanations on surgical videos from three hospitals across two continents by comparing them to explanations generated by humans experts. To improve the reliability of AI-based explanations, we propose the strategy of training with explanations -TWIX -which uses human explanations as supervision to explicitly teach an AI system to highlight important video frames. RESULTS: We show that while AI-based explanations often align with human explanations, they are not equally reliable for different sub-cohorts of surgeons (e.g., novices vs. experts), a phenomenon we refer to as an explanation bias. We also show that TWIX enhances the reliability of AI-based explanations, mitigates the explanation bias, and improves the performance of AI systems across hospitals. These findings extend to a training environment where medical students can be provided with feedback today. CONCLUSIONS: Our study informs the impending implementation of AI-augmented surgical training and surgeon credentialing programs, and contributes to the safe and fair democratization of surgery.

Surgeons aim to master skills necessary for surgery. One such skill is suturing which involves connecting objects together through a series of stitches. Mastering these surgical skills can be improved by providing surgeons with feedback on the quality of their performance. However, such feedback is often absent from surgical practice. Although performance-based feedback can be provided, in theory, by recently-developed artificial intelligence (AI) systems that use a computational model to assess a surgeon's skill, the reliability of this feedback remains unknown. Here, we compare AI-based feedback to that provided by human experts and demonstrate that they often overlap with one another. We also show that explicitly teaching an AI system to align with human feedback further improves the reliability of AI-based feedback on new videos of surgery. Our findings outline the potential of AI systems to support the training of surgeons by providing feedback that is reliable and focused on a particular skill, and guide programs that give surgeons qualifications by complementing skill assessments with explanations that increase the trustworthiness of such assessments.

A vision transformer for decoding surgeon activity from surgical videos.

The intraoperative activity of a surgeon has substantial impact on postoperative outcomes. However, for most surgical procedures, the details of intraoperative surgical actions, which can vary widely, are not well understood. Here we report a machine learning system leveraging a vision transformer and supervised contrastive learning for the decoding of elements of intraoperative surgical activity from videos commonly collected during robotic surgeries. The system accurately identified surgical steps, actions performed by the surgeon, the quality of these actions and the relative contribution of individual video frames to the decoding of the actions. Through extensive testing on data from three different hospitals located in two different continents, we show that the system generalizes across videos, surgeons, hospitals and surgical procedures, and that it can provide information on surgical gestures and skills from unannotated videos. Decoding intraoperative activity via accurate machine learning systems could be used to provide surgeons with feedback on their operating skills, and may allow for the identification of optimal surgical behaviour and for the study of relationships between intraoperative factors and postoperative outcomes.

A methodology for the annotation of surgical videos for supervised machine learning applications.

PURPOSE: Surgical data science is an emerging field focused on quantitative analysis of pre-, intra-, and postoperative patient data (Maier-Hein et al. in Med Image Anal 76: 102306, 2022). Data science approaches can decompose complex procedures, train surgical novices, assess outcomes of actions, and create predictive models of surgical outcomes (Marcus et al. in Pituitary 24: 839-853, 2021; Røadsch et al. in Nat Mach Intell, 2022). Surgical videos contain powerful signals of events that may impact patient outcomes. A necessary step before the deployment of supervised machine learning methods is the development of labels for objects and anatomy. We describe a complete method for annotating videos of transsphenoidal surgery. METHODS: Endoscopic video recordings of transsphenoidal pituitary tumor removal surgeries were collected from a multicenter research collaborative. These videos were anonymized and stored in a cloud-based platform. Videos were uploaded to an online annotation platform. Annotation framework was developed based on a literature review and surgical observations to ensure proper understanding of the tools, anatomy, and steps present. A user guide was developed to trained annotators to ensure standardization. RESULTS: A fully annotated video of a transsphenoidal pituitary tumor removal surgery was produced. This annotated video included over 129,826 frames. To prevent any missing annotations, all frames were later reviewed by highly experienced annotators and a surgeon reviewer. Iterations to annotated videos allowed for the creation of an annotated video complete with labeled surgical tools, anatomy, and phases. In addition, a user guide was developed for the training of novice annotators, which provides information about the annotation software to ensure the production of standardized annotations. CONCLUSIONS: A standardized and reproducible workflow for managing surgical video data is a necessary prerequisite to surgical data science applications. We developed a standard methodology for annotating surgical videos that may facilitate the quantitative analysis of videos using machine learning applications. Future work will demonstrate the clinical relevance and impact of this workflow by developing process modeling and outcome predictors.