Evaluating the Performance of a Commercially Available AI Algorithm for Automated Detection of Pulmonary Embolism on CECT and CTPA of COVID-19 Patients

Objective To investigate the performance of a commercially available artificial intelligence (AI) algorithm for detection of pulmonary embolism (PE) on contrast-enhanced CTs in patients hospitalized for COVID-19. Patients & Methods Retrospective analysis was performed of all contrast-enhanced chest CTs on patients admitted for COVID-19 between March 2020 and December 2021. Based on the original radiology reports, all PE-positive exams were included (n=527). Using a reversed flow single gate diagnostic accuracy case-control model, a randomly selected cohort of PE-negative exams (n=977) was included. Pulmonary parenchymal disease severity was assessed for all included studies using a semi-quantitative system, the Total Severity Score (TSS). All included CTs were sent for interpretation by the commercially available AI algorithm, Aidoc. Discrepancies between AI and original radiology reports were resolved by three blinded radiologists, who rendered a final determination of indeterminate, positive, or negative. Results A total of 78 studies were found to be discrepant, of which 13 (16.6%) were deemed indeterminate by readers and excluded. The sensitivity and specificity of AI was 93.2%;(95% confidence interval [CI] 90.6-95.2%), and 99.6%;(95% CI 98.9-99.9%), respectively. AI's accuracy for all TSS groups (mild, moderate, severe) was high (98.4%, 96.7%, and 97.2%, respectively). AI was more accurate in PE detection on CTPAs vs CECTs (P < .001), with optimal HU of 362 (P=.048). Conclusion The AI algorithm demonstrated high sensitivity, specificity, and accuracy for PE on contrast enhanced CTs in COVID-19 patients regardless of parenchymal disease. Accuracy was significantly affected by the mean attenuation of the pulmonary vasculature. How this affects the legitimacy of the binary outcomes reported by AI is not yet known.

Evaluating the Performance of a Commercially Available Artificial Intelligence Algorithm for Automated Detection of Pulmonary Embolism on Contrast-Enhanced Computed Tomography and Computed Tomography Pulmonary Angiography in Patients With Coronavirus Disease 2019.

Objective: To investigate the performance of a commercially available artificial intelligence (AI) algorithm for the detection of pulmonary embolism (PE) on contrast-enhanced computed tomography (CT) scans in patients hospitalized for coronavirus disease 2019 (COVID-19). Patients and Methods: Retrospective analysis was performed of all contrast-enhanced chest CT scans of patients admitted for COVID-19 between March 1, 2020 and December 31, 2021. Based on the original radiology reports, all PE-positive examinations were included (n=527). Using a reversed-flow single-gate diagnostic accuracy case-control model, a randomly selected cohort of PE-negative examinations (n=977) was included. Pulmonary parenchymal disease severity was assessed for all the included studies using a semiquantitative system, the total severity score. All included CT scans were sent for interpretation by the commercially available AI algorithm, Aidoc. Discrepancies between AI and original radiology reports were resolved by 3 blinded radiologists, who rendered a final determination of indeterminate, positive, or negative. Results: A total of 78 studies were found to be discrepant, of which 13 (16.6%) were deemed indeterminate by readers and were excluded. The sensitivity and specificity of AI were 93.2% (95% CI, 90.6%-95.2%) and 99.6% (95% CI, 98.9%-99.9%), respectively. The accuracy of AI for all total severity score groups (mild, moderate, and severe) was high (98.4%, 96.7%, and 97.2%, respectively). Artificial intelligence was more accurate in PE detection on CT pulmonary angiography scans than on contrast-enhanced CT scans (P<.001), with an optimal Hounsfield unit of 362 (P=.048). Conclusion: The AI algorithm demonstrated high sensitivity, specificity, and accuracy for PE on contrast-enhanced CT scans in patients with COVID-19 regardless of parenchymal disease. Accuracy was significantly affected by the mean attenuation of the pulmonary vasculature. How this affects the legitimacy of the binary outcomes reported by AI is not yet known.

Race affects adverse outcomes of deep vein thrombosis, pulmonary embolism, and acute kidney injury in coronavirus disease 2019 hospitalized patients.

OBJECTIVE: The purpose of the present study was to explore the racial disparities in the incidence of deep vein thrombosis (DVT), pulmonary embolism (PE), and acute kidney injury (AKI) in hospitalized patients with coronavirus disease 2019 (COVID-19). METHODS: A retrospective analysis was performed of prospectively collected data of consecutive COVID-19 patients hospitalized from March 11, 2020 to May 27, 2021. The primary outcome measures were the incidence of DVT/PE and mortality. The secondary outcome measures included differences in the length of hospitalization, need for intensive care unit care, readmission, and AKI. Multivariable regression models were used to assess for independent predictors of the primary and secondary outcome measures. RESULTS: The present study included 876 hospitalized patients with COVID-19. The mean age was 64.4 ± 16.2 years, and 355 were women (40.5%). Of the 876 patients, 694 (79.2%) had identified as White, 111 (12.7%) as Black/African American, 48 (5.5%) as Asian, and 23 (2.6%) as other. The overall incidence of DVT/PE was 8.7%. The DVT/PE incidence rates differed across the race groups and was highest for Black/African American patients (n = 18; 16.2%), followed by Asian patients (n = 5; 10.4%), White patients (n = 52; 7.5%), and other (n = 1; 4.4%; P = .03). All but one of the hospitalization outcomes examined demonstrated no differences according to race, including the hospitalization stay (P = .33), need for intensive care unit care (P = .20), readmission rates (P = .52), and hospital all-cause mortality (P = .29). The AKI incidence differed among races, affecting a higher proportion of Black/African American patients (P=.003). On multivariable regression analysis, Black/African American race (odds ratio [OR], 2.0; 95% confidence interval [CI], 1.0-4.0; P = .04) and higher D-dimer levels (OR, 1.1; 95% CI, 1.1-1.2; P < .0001) were predictors of DVT/PE. In addition, Black/African American race (OR, 2.3; 95% CI, 1.4-3.7; P = .001), lower hemoglobin levels (OR, 0.84; 95% CI, 0.8-0.9; P ≤ .0001), male sex (OR, 1.7; 95% CI, 1.2-2.4; P = .005), hypertension (OR, 2.1; 95% CI, 1.4-3.1; P = .0005), and older age (OR, 1.02; 95% CI, 1.006-1.03; P = .003) were predictors of AKI. CONCLUSIONS: In our single-center case series, we found a higher incidence of DVT/PE and AKI among Black/African American patients with COVID-19. Black/African American race and D-dimer levels were independent predictors of DVT/PE, and Black/African American race, hemoglobin, and D-dimer levels were independent predictors of AKI.

Iatrogenic left common carotid artery to right internal jugular vein arteriovenous fistula closure.

We report a minimally invasive approach to the repair of a 2.8-cm × 6.0-cm left proximal common carotid to right internal jugular vein arteriovenous fistula. A 47-year-old woman with coronavirus disease 2019 pneumonia had received extracorporeal membranous oxygenation and developed a rare, presumably cannulation-related, vascular injury. We used a plug designed and typically used for the endovascular management of a patent ductus arteriosus.

Resolution of acute pulmonary embolism using anticoagulation therapy alone in coronavirus disease 2019.

OBJECTIVE: To investigate the radiographic resolution of acute pulmonary embolism (PE) using contrast-enhanced computed tomography (CECT) examinations in patients diagnosed with acute PE while hospitalized with coronavirus disease 2019 (COVID-19) and to understand the mid-term and long-term implications of anticoagulation therapy. METHODS: We identified patients with acute PE per CECT and at least one follow-up CECT from March 11, 2020, to May 27, 2021, using a prospective registry of all hospitalized patients with COVID-19 infection receiving care within a multicenter Health System. Initial and follow-up CECT examinations were reviewed independently by two radiologists to evaluate for PE resolution. The Modified Miller Score was used to assess for thrombus burden at diagnosis and on follow-up. RESULTS: Of the 6070 hospitalized patients with COVID-19 infection, 5.7% (348/6070) were diagnosed with acute PE and 13.5% (47/348) had a follow-up CECT examination. The mean ± standard deviation time to follow-up imaging was 44 ± 48 days (range, 3-161 days). Of 47 patients, 47 (72.3%) had radiographic resolution of PE, with a mean time to follow-up of 48 ± 43 days (range, 6-239 days). All patients received anticoagulation monotherapy for a mean of 149 ± 95 days and this included apixaban (63.8%), warfarin (12.8%), and rivaroxaban (8.5%), among others. The mean Modified Miller Score at PE diagnosis and follow-up was 4.8 ± 4.2 (range, 1-14) and 1.4 ± 3.3 (range, 0-16; P < .0001), respectively. Nine patients (19%) died at a mean of 13 ± 8 days after follow-up CECT (range, 1-27 days) and at a mean of 28 ± 16 days after admission (range, 11-68 days). Seen of the nine deaths (78%) deaths were associated with progression of COVID-19 pneumonia. CONCLUSIONS: Hospitalized patients with COVID-19 have a clinically apparent 5.7% rate of developing PE. In patients with follow-up imaging, 72.3% had radiographic thrombus resolution at a mean of 44 days while on anticoagulation. Prospective studies of the natural history of PEs with COVID-19 that include systematic follow-up imaging are warranted to help guide anticoagulation recommendations.

Deep vein thrombosis and pulmonary embolism among hospitalized coronavirus disease 2019-positive patients predicted for higher mortality and prolonged intensive care unit and hospital stays in a multisite healthcare system.

OBJECTIVE: We assessed the incidence of deep vein thrombosis (DVT) and pulmonary embolism (PE) in hospitalized patients with coronavirus disease 2019 (COVID-19) compared with that in a matched cohort with similar cardiovascular risk factors and the effects of DVT and PE on the hospital course. METHODS: We performed a retrospective review of prospectively collected data from COVID-19 patients who had been hospitalized from March 11, 2020 to September 4, 2020. The patients were randomly matched in a 1:1 ratio by age, sex, hospital of admission, smoking history, diabetes mellitus, and coronary artery disease with a cohort of patients without COVID-19. The primary end point was the incidence of DVT/PE and the odds of developing DVT/PE using a conditional logistic regression model. The secondary end point was the hospitalization outcomes for COVID-19 patients with and without DVT/PE, including mortality, intensive care unit (ICU) admission, ICU stay, and length of hospitalization (LOH). Multivariable regression analysis was performed to identify the variables associated with mortality, ICU admission, discharge disposition, ICU duration, and LOH. RESULTS: A total of 13,310 patients had tested positive for COVID-19, 915 of whom (6.9%) had been hospitalized across our multisite health care system. The mean age of the hospitalized patients was 60.8 ± 17.0 years, and 396 (43.3%) were women. Of the 915 patients, 82 (9.0%) had had a diagnosis of DVT/PE confirmed by ultrasound examination of the extremities and/or computed tomography angiography of the chest. The odds of presenting with DVT/PE in the setting of COVID-19 infection was greater than that without COVID-19 infection (0.6% [5 of 915] vs 9.0% [82 of 915]; odds ratio [OR], 18; 95% confidence interval [CI], 8.0-51.2; P < .001). The vascular risk factors were not different between the COVID-19 patients with and without DVT/PE. Mortality (P = .02), the need for ICU stay (P < .001), duration of ICU stay (P < .001), and LOH (P < .001) were greater in the DVT/PE cohort than in the cohort without DVT/PE. On multivariable logistic regression analysis, the hemoglobin (OR, 0.71; 95% CI, 0.46-0.95; P = .04) and D-dimer (OR, 1.0; 95% CI, 0.33-1.56; P = .03) levels were associated with higher mortality. Higher activated partial thromboplastin times (OR, 1.1; 95% CI, 1.00-1.12; P = .03) and higher interleukin-6 (IL-6) levels (OR, 1.0; 95% CI, 1.01-1.07; P = .05) were associated with a greater risk of ICU admission. IL-6 (OR, 1.0; 95% CI, 1.00-1.02; P = .05) was associated with a greater risk of rehabilitation placement after discharge. On multivariable gamma regression analysis, hemoglobin (coefficient, -3.0; 95% CI, 0.03-0.08; P = .005) was associated with a prolonged ICU stay, and the activated partial thromboplastin time (coefficient, 2.0; 95% CI, 0.003-0.006; P = .05), international normalized ratio (coefficient, -3.2; 95% CI, 0.06-0.19; P = .002) and IL-6 (coefficient, 2.4; 95% CI, 0.0011-0.0027; P = .02) were associated with a prolonged LOH. CONCLUSIONS: A significantly greater incidence of DVT/PE occurred in hospitalized COVID-19-positive patients compared with a non-COVID-19 cohort matched for cardiovascular risk factors. Patients affected by DVT/PE were more likely to experience greater mortality, to require ICU admission, and experience prolonged ICU stays and LOH compared with COVID-19-positive patients without DVT/PE. Advancements in DVT/PE prevention are needed for patients hospitalized for COVID-19 infection.