Development of an Artificial Intelligence Tool for Intraoperative Guidance During Endovascular Abdominal Aortic Aneurysm Repair.

BACKGROUND: Adverse events during surgery can occur in part due to errors in visual perception and judgment. Deep learning is a branch of artificial intelligence (AI) that has shown promise in providing real-time intraoperative guidance. This study aims to train and test the performance of a deep learning model that can identify inappropriate landing zones during endovascular aneurysm repair (EVAR). METHODS: A deep learning model was trained to identify a "No-Go" landing zone during EVAR, defined by coverage of the lowest renal artery by the stent graft. Fluoroscopic images from elective EVAR procedures performed at a single institution and from open-access sources were selected. Annotations of the "No-Go" zone were performed by trained annotators. A 10-fold cross-validation technique was used to evaluate the performance of the model against human annotations. Primary outcomes were intersection-over-union (IoU) and F1 score and secondary outcomes were pixel-wise accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). RESULTS: The AI model was trained using 369 images procured from 110 different patients/videos, including 18 patients/videos (44 images) from open-access sources. For the primary outcomes, IoU and F1 were 0.43 (standard deviation ± 0.29) and 0.53 (±0.32), respectively. For the secondary outcomes, accuracy, sensitivity, specificity, NPV, and PPV were 0.97 (±0.002), 0.51 (±0.34), 0.99 (±0.001). 0.99 (±0.002), and 0.62 (±0.34), respectively. CONCLUSIONS: AI can effectively identify suboptimal areas of stent deployment during EVAR. Further directions include validating the model on datasets from other institutions and assessing its ability to predict optimal stent graft placement and clinical outcomes.

Machine learning to predict outcomes following endovascular abdominal aortic aneurysm repair.

BACKGROUND: Endovascular aneurysm repair (EVAR) for abdominal aortic aneurysm (AAA) carries important perioperative risks; however, there are no widely used outcome prediction tools. The aim of this study was to apply machine learning (ML) to develop automated algorithms that predict 1-year mortality following EVAR. METHODS: The Vascular Quality Initiative database was used to identify patients who underwent elective EVAR for infrarenal AAA between 2003 and 2023. Input features included 47 preoperative demographic/clinical variables. The primary outcome was 1-year all-cause mortality. Data were split into training (70 per cent) and test (30 per cent) sets. Using 10-fold cross-validation, 6 ML models were trained using preoperative features with logistic regression as the baseline comparator. The primary model evaluation metric was area under the receiver operating characteristic curve (AUROC). Model robustness was evaluated with calibration plot and Brier score. RESULTS: Some 63 655 patients were included. One-year mortality occurred in 3122 (4.9 per cent) patients. The best performing prediction model for 1-year mortality was XGBoost, achieving an AUROC (95 per cent c.i.) of 0.96 (0.95-0.97). Comparatively, logistic regression had an AUROC (95 per cent c.i.) of 0.69 (0.68-0.71). The calibration plot showed good agreement between predicted and observed event probabilities with a Brier score of 0.04. The top 3 predictive features in the algorithm were 1) unfit for open AAA repair, 2) functional status, and 3) preoperative dialysis. CONCLUSIONS: In this data set, machine learning was able to predict 1-year mortality following EVAR using preoperative data and outperformed standard logistic regression models.

Using machine learning to predict outcomes following open abdominal aortic aneurysm repair.

OBJECTIVE: Prediction of outcomes following open abdominal aortic aneurysm (AAA) repair remains challenging with a lack of widely used tools to guide perioperative management. We developed machine learning (ML) algorithms that predict outcomes following open AAA repair. METHODS: The Vascular Quality Initiative (VQI) database was used to identify patients who underwent elective open AAA repair between 2003 and 2023. Input features included 52 preoperative demographic/clinical variables. All available preoperative variables from VQI were used to maximize predictive performance. The primary outcome was in-hospital major adverse cardiovascular event (MACE; composite of myocardial infarction, stroke, or death). Secondary outcomes were individual components of the primary outcome, other in-hospital complications, and 1-year mortality and any reintervention. We split our data into training (70%) and test (30%) sets. Using 10-fold cross-validation, six ML models were trained using preoperative features (Extreme Gradient Boosting [XGBoost], random forest, Naïve Bayes classifier, support vector machine, artificial neural network, and logistic regression). The primary model evaluation metric was area under the receiver operating characteristic curve (AUROC). Model robustness was evaluated with calibration plot and Brier score. The top 10 predictive features in our final model were determined based on variable importance scores. Performance was assessed on subgroups based on age, sex, race, ethnicity, rurality, median area deprivation index, proximal clamp site, prior aortic surgery, and concomitant procedures. RESULTS: Overall, 12,027 patients were included. The primary outcome of in-hospital MACE occurred in 630 patients (5.2%). Compared with patients without a primary outcome, those who developed in-hospital MACE were older with more comorbidities, demonstrated poorer functional status, had more complex aneurysms, and were more likely to require concomitant procedures. Our best performing prediction model for in-hospital MACE was XGBoost, achieving an AUROC of 0.93 (95% confidence interval, 0.92-0.94). Comparatively, logistic regression had an AUROC of 0.71 (95% confidence interval, 0.70-0.73). For secondary outcomes, XGBoost achieved AUROCs between 0.84 and 0.94. The calibration plot showed good agreement between predicted and observed event probabilities with a Brier score of 0.05. These findings highlight the excellent predictive performance of the XGBoost model. The top three predictive features in our algorithm for in-hospital MACE following open AAA repair were: (1) coronary artery disease; (2) American Society of Anesthesiologists classification; and (3) proximal clamp site. Model performance remained robust on all subgroup analyses. CONCLUSIONS: Open AAA repair outcomes can be accurately predicted using preoperative data with our ML models, which perform better than logistic regression. Our automated algorithms can help guide risk-mitigation strategies for patients being considered for open AAA repair to improve outcomes.

Prediction of bird-beak configuration in thoracic endovascular aortic repair preoperatively using patient-specific finite element simulations.

Objectives: Formation of bird-beak configuration in thoracic endovascular aortic repair (TEVAR) has been shown to be correlated with the risk of complications such as type Ia endoleaks, stent graft migration, and collapse. The aim of this study was to use patient-specific computational simulations of TEVAR to predict the formation of bird-beak configuration preoperatively. Methods: Patient-specific TEVAR computational simulations are developed using a retrospective cohort of patients treated for thoracic aortic aneurysm. The preoperative computed tomography images were segmented to develop three-dimensional geometry of the thoracic aorta. These geometries were used in finite element simulations of stent graft deployment during TEVAR. Simulated results were compared against the postoperative computed tomography images to assess the accuracy of simulations in predicting the proximal position of a deployed stent graft and presence of bird-beak. In cases with a bird-beak configuration, the length and angle of the bird-beak were measured and compared between the simulated and postoperative results. Results: Twelve TEVAR patient cases were simulated. Computational simulations were able to accurately predict whether the proximal stent graft was fully apposed, proximal bare stents were protruded, or bird-beak configuration was present. In three cases with bird-beak configuration, simulations predicted the length and angle of the bird-beak with less than 10% and 24% error, respectively. Other factors such as a small aortic arch angle, small oversizing value, and landing zones close to the arch apex may have played a role in formation of bird-beak in these patients. Conclusions: Computational simulations of TEVAR accurately predicted the proximal position of a deployed stent graft and the presence of bird-beak preoperatively. The computational models were able to predict the length and angle of bird-beak configurations with good accuracy. These simulations can provide insight into the surgical planning process with the goal of minimizing bird-beak occurrence.

Ascending aortic geometry and its relationship to the biomechanical properties of aortic tissue.

Objective: The objective of this study was to evaluate the relationship between ascending aortic geometry and biomechanical properties. Methods: Preoperative computed tomography scans from ascending aortic aneurysm patients were analyzed using a center line technique (n = 68). Aortic length was measured from annulus to innominate artery, and maximal diameter from this segment was recorded. Biaxial tensile testing of excised tissue was performed to derive biomechanical parameters energy loss (efficiency in performing the Windkessel function) and modulus of elasticity (stiffness). Delamination testing (simulation of dissection) was performed to derive delamination strength (strength between tissue layers). Results: Aortic diameter weakly correlated with energy loss (r 2 = 0.10; P < .01), but not with modulus of elasticity (P = .13) or delamination strength (P = .36). Aortic length was not associated with energy loss (P = .87), modulus of elasticity (P = .13) or delamination strength (P = .90). Using current diameter guidelines, aortas >55 mm (n = 33) demonstrated higher energy loss than those <55 mm (n = 35; P = .05), but no difference in modulus of elasticity (P = .25) or delamination strength (P = .89). A length cutoff of 110 mm was proposed as an indication for repair. Aortas >110 mm (n = 37) did not exhibit a difference in energy loss (P = .40), modulus of elasticity (P = .69), or delamination strength (P = .68) compared with aortas <110 mm (n = 31). Aortas above diameter and length thresholds (n = 21) showed no difference in energy loss (P = .35), modulus of elasticity (P = .55), or delamination strength (P = .61) compared with smaller aortas (n = 47). Conclusions: Aortic geometry poorly reflects the mechanical properties of aortic tissue. Weak association between energy loss and diameter supports intervention at larger diameters. Further research into markers that better capture aortic biomechanics is needed.

A Comparison of Vessel Patch Materials in Tetralogy of Fallot Patients Using Virtual Surgery Techniques.

Tetralogy of Fallot (ToF) is characterized by stenosis causing partial obstruction of the right ventricular outflow tract, typically alleviated through the surgical application of a vessel patch made from a biocompatible material. In this study, we use computational simulations to compare the mechanical performance of four patch materials for various stenosis locations. Nine idealized pre-operative ToF geometries were created by imposing symmetrical stenoses on each of three anatomical sub-regions of the pulmonary arteries of three patients with previously repaired ToF. A virtual surgery methodology was implemented to replicate the steps of vessel de-pressurization, surgical patching, and subsequent vessel expansion after reperfusion. Significant differences in patch average stress (p < 0.001) were found between patch materials. Biological patch materials (porcine xenopericardium, human pericardium) exhibited higher patch stresses in comparison to synthetic patch materials (Dacron and PTFE). Observed differences were consistent across the various stenosis locations and were insensitive to patient anatomy.

Editor's Choice - Meta-Analysis of Compliance with Endovascular Aneurysm Repair Surveillance: The EVAR Surveillance Paradox.

OBJECTIVE: To compare the survival of patients who attended surveillance after endovascular aneurysm repair (EVAR) with those who were non-compliant. DATA SOURCES: MEDLINE and Embase were searched using the Ovid interface. REVIEW METHODS: A systematic review was conducted complying with the PRISMA guidelines. Eligible studies compared survival in EVAR surveillance compliant patients with non-compliant patients. Non-compliance was defined as failure to attend at least one post-EVAR follow up. The risk of bias was assessed with the Newcastle-Ottawa scale, and the certainty of evidence using the GRADE framework. Primary outcomes were survival and aneurysm related death. Effect measures were the hazard ratio (HR) or odds ratio (OR) and 95% confidence interval (CI) calculated using the inverse variance or Mantel-Haenszel statistical method and random effects models. RESULTS: Thirteen cohort studies with a total of 22 762 patients were included. Eight studies were deemed high risk of bias. The pooled proportion of patients who were non-compliant with EVAR surveillance was 43% (95% CI 36 - 51). No statistically significant difference was found in the hazard of all cause mortality (HR 1.04, 95% CI 0.61 - 1.77), aneurysm related mortality (HR 1.80, 95% CI 0.85-3.80), or secondary intervention (HR 0.66, 95% CI 0.31 - 1.41) between patients who had incomplete and complete follow up after EVAR. The odds of aneurysm rupture were lower in non-compliant patients (OR 0.63, 95% CI 0.39 - 1.01). The certainty of evidence was very low for all outcomes. Subgroup analysis for patients who had no surveillance vs. those with complete surveillance showed no significant difference in all cause mortality (HR 1.10, 95% CI 0.43 - 2.80). CONCLUSION: Patients who were non-compliant with EVAR surveillance had similar survival to those who were compliant. These findings question the value of intense surveillance in all patients post-EVAR and highlight the need for further research on individualised or risk adjusted surveillance.

Assessing the robustness of negative vascular surgery randomized controlled trials using their reverse fragility index.

OBJECTIVE: The reverse fragility index (RFI) describes the number of event conversions needed to convert a statistically nonsignificant dichotomous outcome to a significant one. The objective of the present study was to assess the RFI of vascular surgery randomized controlled trials (RCTs) comparing endovascular vs open surgery for the treatment of abdominal aortic aneurysms (AAAs), carotid artery stenosis (CAS), and peripheral artery disease (PAD). METHODS: MEDLINE and Embase were searched for RCTs that had investigated AAAs, CAS, or PAD with statistically nonsignificant binary primary outcomes. The primary outcome for the present study was the median RFI. Calculation of the RFI was performed by creating two-by-two contingency tables and subtracting events from the group with fewer events and adding nonevents to the same group until a two-tailed Fisher exact test had produced a statistically significant result (P ≤ .05). RESULTS: Of 4187 reports, 49 studies reporting 103 different primary end points were included. The overall median RFI was 7 (interquartile range [IQR], 5-13). The specific RFIs for AAA, CAS, and PAD were 10 (IQR, 6-15.5), 6 (IQR, 5-9.5), and 7 (IQR, 5.5-10), respectively. Of the 103 end points, 42 (47%) had had a loss to follow-up greater than the RFI, of which 10 were AAA trials (24%), 23 were CAS trials (55%), and 9 were PAD trials (21%). The Pearson correlation demonstrated a significant positive relationship between a study's RFI and the impact factor of its publishing journal (r = 0.38; 95% confidence interval [CI], 0.20-0.54; P < .01), length of follow-up (r = 0.43; 95% CI, 0.26-0.58; P < .01), and sample size (r = 0.28; 95% CI, 0.09-0.45; P < .01). CONCLUSIONS: A small number of events (median, 7) was required to change the outcome of negative RCTs from statistically nonsignificant to significant, with 47% of the studies having missing data that could have reversed the finding of its primary outcome. Reporting of the RFI relative to the loss to follow-up could be of benefit in future trials and provide confidence regarding the robustness of the P value.

A modeling framework for computational simulations of thoracic endovascular aortic repair.

Thoracic endovascular aortic repair (TEVAR) is a minimally invasive treatment for thoracic aortic conditions including aneurysms and is associated with a number of postoperative stent graft related complications. Computational simulations of TEVAR have the potential to predict surgical outcomes and complications preoperatively. When using simulations for stent graft design and prediction of complications in a population, it is difficult to generalize patient-specific TEVAR computational models due to patient variability. This study proposes a novel modeling framework for creating realistic population-based computational models of TEVAR focused on aneurysms that allow for developing various clinically relevant geometric configurations and scenarios that are not easily attainable with limited patient data. The framework includes a methodology for developing population-based thoracic aortic geometries and defining age-dependent aortic tissue material models, as well as detailed steps and boundary conditions for finite element modeling of stent graft deployment during TEVAR. The simulation framework is illustrated for predicting the formation of a bird-beak configuration, a wedge-shaped gap at the proximal end of the deployed stent graft in TEVAR that leads to incomplete seal. A baseline TEVAR simulation model was developed along with three simulations in which the value of aortic curvature, aortic arch angle, or aortic tissue properties varied from the baseline model. Analyzing the length and angle of the bird-beak configuration in each case shows that the bird-beak size is sensitive to different values of the aortic geometry highlighting the importance of using realistic parameter values.

The Impact of Sex on Outcomes following Carotid Endarterectomy.

BACKGROUND: Previous studies have demonstrated significant sex differences in vascular surgery outcomes. We assessed stroke or death rates following carotid endarterectomy (CEA) in women versus men. METHODS: The Vascular Quality Initiative was used to identify all patients who underwent CEA between 2010 and 2019. Demographic, clinical, and procedural characteristics were recorded and differences between women and men were assessed using independent t-test and chi-squared test. The primary outcomes were 30-day and 1-year stroke or death. Associations between sex and outcomes were assessed using univariate/multivariate logistic regression and Cox proportional hazards analysis. RESULTS: Overall, 52,137 women and 79,974 men underwent CEA in Vascular Quality Initiative sites during the study period. Women were younger (70.3 vs. 70.5 years, P < 0.001) and more likely to have hypertension (89.2% vs. 88.9%, P < 0.05) and diabetes (36.2% vs. 35.8%, P < 0.001) but less likely to be diagnosed with coronary artery disease (23.2% vs. 31.0%, P < 0.001). A greater proportion of men were receiving cardiovascular risk reduction medications and had symptomatic carotid stenosis (28.5% vs. 26.7%, P < 0.001). Women had shorter procedure times (113 vs. 122 min, P < 0.001) and were less likely to receive electroencephalography neuromonitoring (27.9% vs. 28.8%, P < 0.001), drain (35.9% vs. 37.3%, P < 0.001), and protamine (67.4% vs. 68.0%, P < 0.01). Stroke or death at 30 days (1.9% vs. 1.8%, P = 0.60) and 1 year (hazard ratio [HR] 0.98, 95% confidence interval [CI] 0.94-1.01, P = 0.20) were similar between groups, which persisted in asymptomatic patients (HR 0.97, 95% CI 0.93-1.01, P = 0.17) and symptomatic patients (HR 0.99, 95% CI 0.93-1.05, P = 0.71). The similarities in 1-year stroke or death rates existed in both the United States (HR 0.96, 95% CI 0.92-1.01, P = 0.09) and Canada (HR 1.21, 95% CI 0.47-3.11, P = 0.70). CONCLUSIONS: Despite sex differences in clinical and procedural characteristics, women and men have similar 30-day and 1-year outcomes following CEA.

Identification of geometric and mechanical factors predictive of bird-beak configuration in thoracic endovascular aortic repair using computational models of stent graft deployment.

Objective: Formation of a bird-beak configuration in thoracic endovascular aortic repair (TEVAR) has been shown to be influenced by various factors. However, the main cause of bird-beak formation remains poorly understood. The hypothesis has been that the geometric and mechanical properties of both the aorta and the stent graft contribute to the formation and extent of a bird-beak configuration. The goal of the present study was to use parameter-based computational simulations of TEVAR to predict for bird-beak formation and identify its most significant contributing factors. Methods: In the present study, we considered five parameters for the computational simulations of TEVAR, including aortic curvature, aortic arch angle, age as a surrogate for thoracic aortic tissue properties, TEVAR landing zone, and stent graft oversizing. Using an experimental design approach, computational models for 160 TEVAR scenarios were developed by varying the values of the simulation parameters within clinically relevant ranges. The bird-beak length and angle were used as metrics to evaluate the simulation results. Statistical analysis of the simulation data using a random forest model was conducted to identify significant parameters and interactions. Results: The mean ± standard deviation of the bird-beak length and angle across 160 simulations were 4.32 ± 4.87 mm and 9.16° ± 12.21°, respectively. The largest mean bird-beak length and angle were found in the most distal location in zone 0 (10.04 mm) and zone 2 (21.48°), respectively. An inverse correlation was found between the aortic arch angle and the bird-beak length and angle. In ∼75% of the scenarios, increased stent graft oversizing either fully resolved the presence of the bird-beak configuration or had reduced its size. In the remaining 25%, oversizing minimally changed the bird-beak length and enlarged the bird-beak angle, which mainly occurred in cases with a smaller aortic arch angle and landing zones near the arch apex. This was justified by the mechanism of stent graft bending in the arch angulation. The aortic curvature and tissue properties were shown to be statistically insignificant in relation to bird-beak formation. Conclusions: Significant parameters predictive of a bird-beak configuration in TEVAR were identified, and the trends in which each parameter influenced the bird-beak size were determined. The findings from the present study can inform the surgical planning and device selection process with the goal of minimizing bird-beak formation.

Sex Differences in Outcomes Following Ruptured Abdominal Aortic Aneurysm Repair.

Importance: Sex differences in aortic surgery outcomes are commonly reported. However, data on ruptured abdominal aortic aneurysm (rAAA) repair outcomes in women vs men are limited. Objective: To assess differences in perioperative and long-term mortality following rAAA repair in women vs men. Design, Setting, and Participants: A multicenter, retrospective cohort study was conducted using the Vascular Quality Initiative database, which prospectively captures information on patients who undergo vascular surgery across 796 academic and community hospitals in North America. All patients who underwent endovascular or open rAAA repair between January 1, 2003, and December 31, 2019, were included. Outcomes were assessed up to January 1, 2020. Exposures: Patient sex. Main Outcomes and Measures: Demographic, clinical, and procedural characteristics were recorded, and differences between women vs men were assessed using independent t test and χ2 test. The primary outcomes were in-hospital and 8-year mortality. Associations between sex and outcomes were analyzed using univariable and multivariable logistic regression and Cox proportional hazards regression analysis. Results: A total of 1160 (21.9%) women and 4148 (78.1%) men underwent rAAA repair during the study period. There was a similar proportion of endovascular repairs in women and men (654 [56.4%] vs 2386 [57.5%]). Women were older (mean [SD] age, 75.8 [9.3] vs 71.7 [9.6] years), more likely to have chronic kidney disease (718 [61.9%] vs 2184 [52.7%]), and presented with ruptured aneurysms of smaller diameters (mean [SD] 68 [18.2] vs 78 [30.2] mm). In-hospital mortality was higher in women (34.4% vs 26.6%; odds ratio, 1.44; 95% CI, 1.25-1.66), which persisted after adjusting for demographic, clinical, and procedural characteristics (adjusted odds ratio, 1.36; 95% CI, 1.12-1.66; P = .002). Eight-year survival was lower in women (36.7% vs 49.5%; hazard ratio, 1.25; 95% CI, 1.04-1.50; P = .02), which persisted when stratified by endovascular and open repair. This survival difference existed in both the US and Canada. Variables associated with long-term mortality in women included older age and chronic kidney disease. Conclusions and Relevance: Women who underwent rAAA repair had higher perioperative and 8-year mortality rates following both endovascular and open repair compared with men. Older age and higher rates of chronic kidney disease in women were associated with higher mortality rates. These findings suggest that future studies should assess the reasons for these disparities and whether opportunities exist to improve AAA care for women.

Risk factors, risk stratification and risk-specific surveillance strategies after endovascular aneurysm repair: study protocol for a Delphi study by the International RIsk Stratification in EVAR (IRIS-EVAR) working group.

INTRODUCTION: Several risk factors for adverse events after endovascular aneurysm repair (EVAR) have been described, but there is no consensus on their comparative prognostic significance, use in risk stratification and application in determining postoperative surveillance. METHODS AND ANALYSIS: A scoping review of the literature was conducted to identify risk factors for adverse events after EVAR. Main adverse events were considered post-EVAR abdominal aortic aneurysm rupture and reintervention. Risk factors were grouped into four domains: (1) preoperative anatomy, (2) aortic device, (3) procedure performance and (4) postoperative surveillance. The Delphi methodology will be used to steer a group of experts in the field towards consensus organised into three tiers. In tier 1, participants will be asked to independently rate risk factors for adverse events after EVAR. In tier 2, the panel will be asked to independently rate a range of combinations of risk factors across the four domains derived from tier 1. A risk-stratification tool will then be built, which will include algorithms that map responses to signalling questions onto a proposed risk judgement for each domain. Domain-level judgements will in turn provide the basis for an overall risk judgement for the individual patient. In tier 3, risk factor-informed surveillance strategies will be developed. Each tier will typically include three rounds and rating will be conducted using a 4-point Likert scale, with an option for free-text responses. ETHICS AND DISSEMINATION: Research Ethics Committee and Health Research Authority approval has been waived, since this is a professional staff study and no duty of care lies with the National Health Service to any of the participants. The results will be presented at regional, national and international meetings and will be submitted for publication in peer-reviewed journals. The risk stratification tool and surveillance algorithms will be made publicly available for clinical use and validation.

A Systematic Review and Bibliometric Analysis of Applications of Artificial Intelligence and Machine Learning in Vascular Surgery.

BACKGROUND: Artificial intelligence (AI) and machine learning (ML) have seen increasingly intimate integration with medicine and healthcare in the last 2 decades. The objective of this study was to summarize all current applications of AI and ML in the vascular surgery literature and to conduct a bibliometric analysis of published studies. METHODS: A comprehensive literature search was conducted through Embase, MEDLINE, and Ovid HealthStar from inception until February 19, 2021. Reporting of this study was conducted using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Title and abstract screening, full-text screening, and data extraction were conducted in duplicate. Data extracted included study metadata, the clinical area of study within vascular surgery, type of AI/ML method used, dataset, and the application of AI/ML. Publishing journals were classified as having either a clinical scope or technical scope. The author academic background was classified as clinical, nonclinical (e.g., engineering), or both, depending on author affiliation. RESULTS: The initial search identified 7,434 studies, of which 249 were included for a final analysis. The rate of publications is exponentially increasing, with 158 (63%) studies being published in the last 5 years alone. Studies were most commonly related to carotid artery disease (118, 47%), abdominal aortic aneurysms (51, 20%), and peripheral arterial disease (26, 10%). Study authors employed an average of 1.50 (range: 1-6) distinct AI methods in their studies. The application of AI/ML methods broadly related to predictive models (54, 22%), image segmentation (49, 19.4%), diagnostic methods (46, 18%), or multiple combined applications (91, 37%). The most commonly used AI/ML methods were artificial neural networks (155/378 use cases, 41%), support vector machines (64, 17%), k-nearest neighbors algorithm (26, 7%), and random forests (23, 6%). Datasets to which these AI/ML methods were applied frequently involved ultrasound images (87, 35%), computed tomography (CT) images (42, 17%), clinical data (34, 14%), or multiple datasets (36, 14%). Overall, 22 (9%) studies were published in journals specific to vascular surgery, with the majority (147/249, 59%) being published in journals with a scope related to computer science or engineering. Among 1,576 publishing authors, 46% had exclusively a clinical background, 48% a nonclinical background, and 5% had both a clinical and nonclinical background. CONCLUSIONS: There is an exponentially growing body of literature describing the use of AI and ML in vascular surgery. There is a focus on carotid artery disease and abdominal aortic disease, with many other areas of vascular surgery under-represented. Neural networks and support vector machines composed most AI methods in the literature. As AI/ML continue to see expanded applications in the field, it is important that vascular surgeons appreciate its potential and limitations. In addition, as it sees increasing use, there is a need for clinicians with expertise in AI/ML methods who can optimize its transition into daily practice.

Management and In-hospital Mortality of 2203 Patients With a Traumatic Intimal Tear of the Thoracic Aorta.

OBJECTIVE: Our goal was to describe contemporary management and inhospital mortality associated with blunt thoracic aortic intimal tears (IT) within the American College of Surgeons Trauma Quality Improvement Program. SUMMARY BACKGROUND DATA: The evidence basis for nonoperative expectant management of traumatic iT of the thoracic aorta remains weak. METHODS: All adult patients with a thoracic aortic IT following blunt trauma were captured from Level I and II North American Centers enrolled in Trauma Quality Improvement Program from 2010 to 2017. For each patient, we extracted demographics, injury characteristics, the timing and approach of thoracic aortic repair and in-hospital mortality. Mortality attributable to IT was calculated by comparing IT patients to a propensity-score matched control cohort of severely injured blunt trauma patients without aortic injury. RESULTS: There were 2203 IT patients across 315 facilities. Injury most often resulted from motor vehicle collision (75%). A total of 758 patients (34%) underwent operative management, with 93% (N = 708) of repairs performed via an endovascular approach. Median time to surgery was 11 hours (IQR 4- 40). The frequency of operative management was higher in patients without traumatic brain injury (TBI) (35%, N = 674) compared to those with TBI (29%, N = 84) (P = 0.024). Compared to severely injured blunt trauma patients without aortic injury, ITwas not associated with additional in-hospital mortality (10.7% for IT vs 11.7% for no IT, absolute risk difference: -1.0%, 95% CI: -2.9% to 0.8%). CONCLUSIONS: The majority of blunt thoracic IT are managed nonoperatively and IT does not confer additional in-hospital mortality risk. Future studies should focus on the risk of injury progression.

Prevention and management of spinal cord ischemia following aortic surgery: A survey of contemporary practice.

OBJECTIVE: Spinal cord ischemia (SCI) is a devastating complication of thoracoabdominal aortic aneurysm repair. We aim to characterize current practices pertaining to SCI prevention and treatment across Canada. METHODS: Two questionnaires were developed by the Canadian Thoracic Aortic Collaborative and the Canadian Cardiovascular Critical Care Society targeting aortic surgeons and intensivists. A list of experts in the management of patients at risk of SCI was developed, with representation from each of the Canadian centers that perform complex aortic surgery. RESULTS: The response rate was 91% for both intensivists (21/23), and from cardiac and vascular surgeons (39/43). Most surgeons agreed that staging is important during endovascular repair of extent II thoracoabdominal aortic aneurysm (60%) but not for open repair (34%). All of the surgeons felt prophylactic lumbar drains were effective in reducing SCI, whereas only 66.7% of intensivists felt that lumbar drains were effective (P < .001). There was consensus among surgeons over when to employ lumbar drains. A majority of surgeons preferred to keep the hemoglobin over 100 g/L if the patient demonstrated loss of lower-extremity function, whereas most intensivists felt a target of 80 g/L was adequate (P < .001). Management of perioperative antihypertensives, use of intraoperative adjuncts, and management of venous thromboembolism prophylaxis in the presence of a lumbar drain, were highly variable. CONCLUSIONS: We observed some consensus but considerable variability in the approach to SCI prevention and management across Canada. Future studies focused on the areas of variability may lead to more consistent and improved care for this high-risk population.