Clinical Significance Aortic dissection is associated with high rates of morbidity and mortality 🡪 early diagnosis and prompt intervention greatly improve patient outcomes Mortality rate of 1-2% per hour during first 48 hours Provide real-world validation of FDA 510(k)-approved software application in expediting detection, triage, and ultimately treatment of patients with suspected aortic dissection Viz Aortic Dissection algorithm, in collaboration with Avicenna.AI (La Ciotat, France) Growing concern that algorithmic biases may perpetuate existing health inequities Objective: to assess the real-world performance of deep learning algorithm for detection of aortic dissection on computed tomography angiography (CTA) with a focus on evaluating differences in performance across age, sex, geography, and manufacturer Study Methods 1,303 chest and thoracoabdominal CTA exams from 200+ U.S. hospitals Ground-truth classification for presence or absence of aortic dissection determined through consensus evaluation by three board-certified radiologists Exams analyzed using FDA 510(k)-approved Viz Aortic Dissection algorithm Deep learning model trained on a representative, diverse cohort across age, sex, disease prevalence, race, and clinical settings Algorithmic performance stratified by Age (18-40, 40-60, 60+) Sex (male, female) Geographic region (Continental, Northeast, Pacific, Southeast) Manufacturer (GE Medical Systems, Philips, Siemens, Toshiba) Measured algorithmic fairness across subgroups using equalized odds (EO) differences across true positive rates (TPR) and false positive rates (FPR) Also report overall accuracy, sensitivity, specificity, PPV, and NPV Study Results 1,166 (89.5%) dissection-negative exams, 137 (10.5%) dissection-positive exams Overall accuracy: 97% Sensitivity: 94.2% [95% CI: 88.8% - 97.5%] Specificity: 97.3% [95% CI: 96.2% - 98.1%] PPV of 80.1%, NPV of 99.3% 8 false negatives, largely complex cases 32 false positives, largely result of imaging quality Overall mean EO differences across subgroups was 0.031, with individual EO values noted to be small and consistent for: age [18-40: 0.0584, 40-60: 0.0294, 60+: 0.0368] sex [M: 0.0227, F: 0.0359] geographic region [Continental: 0.0584, NE: 0.0487, Pacific: 0.0227, SE: 0.0314] manufacturer [GE: 0.0111, Philips: 0.013, Siemens: 0.0047, Toshiba: 0.0274] In general, small decreases in TPR or FPR often balanced by small increases in the complimentary metric for most subgroups. Clinical Takeaways Real-world validation of a deep learning AI-based detection algorithm for suspected aortic dissection Sensitivity: 94.2% Specificity: 97.3% Allows for rapid patient triage 🡪 earlier diagnoses 🡪 accelerated care coordination 🡪 timely initiation of life-saving interventions 🡪 better patient outcomes Generalizability across demographics and clinical parameters is critical in preventing algorithmic biases and promoting equitable health outcomes Deep learning tool for aortic dissection detection yields no significant biases across patient demographics and scanner manufacturers from 200+ U.S. hospitals Citations Gawinecka J, Schönrath F, von Eckardstein A. Acute aortic dissection: pathogenesis, risk factors and diagnosis. Swiss Med Wkly. 2017 Aug 25;147:w14489. doi: 10.4414/smw.2017.14489. PMID: 28871571. Gudbjartsson T, Ahlsson A, Geirsson A, Gunn J, Hjortdal V, Jeppsson A, Mennander A, Zindovic I, Olsson C. Acute type A aortic dissection - a review. Scand Cardiovasc J. 2020 Feb;54(1):1-13. doi: 10.1080/14017431.2019.1660401. Epub 2019 Sep 23. PMID: 31542960. Harris KM, Nienaber CA, Peterson MD, et al. Early Mortality in Type A Acute Aortic Dissection: Insights From the International Registry of Acute Aortic Dissection. JAMA Cardiol. 2022;7(10):1009–1015. doi:10.1001/jamacardio.2022.2718

PURPOSE: Since the prompt recognition of acute pulmonary embolism (PE) and the immediate initiation of treatment can significantly reduce the risk of death, we developed a deep learning (DL)-based application aimed to automatically detect PEs on chest computed tomography angiograms (CTAs) and alert radiologists for an urgent interpretation. Convolutional neural networks (CNNs) were used to design the application. The associated algorithm used a hybrid 3D/2D UNet topology. The training phase was performed on datasets adequately distributed in terms of vendors, patient age, slice thickness, and kVp. The objective of this study was to validate the performance of the algorithm in detecting suspected PEs on CTAs. METHODS: The validation dataset included 387 anonymized real-world chest CTAs from multiple clinical sites (228 U.S. cities). The data were acquired on 41 different scanner models from five different scanner makers. The ground truth (presence or absence of PE on CTA images) was established by three independent U.S. board-certified radiologists. RESULTS: The algorithm correctly identified 170 of 186 exams positive for PE (sensitivity 91.4% [95% CI: 86.4-95.0%]) and 184 of 201 exams negative for PE (specificity 91.5% [95% CI: 86.8-95.0%]), leading to an accuracy of 91.5%. False negative cases were either chronic PEs or PEs at the limit of subsegmental arteries and close to partial volume effect artifacts. Most of the false positive findings were due to contrast agent-related fluid artifacts, pulmonary veins, and lymph nodes. CONCLUSIONS: The DL-based algorithm has a high degree of diagnostic accuracy with balanced sensitivity and specificity for the detection of PE on CTAs.

Purpose: Recently developed machine-learning algorithms have demonstrated strong performance in the detection of intracranial hemorrhage (ICH) and large vessel occlusion (LVO). However, their generalizability is often limited by geographic bias of studies. The aim of this study was to validate a commercially available deep learning-based tool in the detection of both ICH and LVO across multiple hospital sites and vendors throughout the U.S. Materials and Methods: This was a retrospective and multicenter study using anonymized data from two institutions. Eight hundred fourteen non-contrast CT cases and 378 CT angiography cases were analyzed to evaluate ICH and LVO, respectively. The tool's ability to detect and quantify ICH, LVO, and their various subtypes was assessed among multiple CT vendors and hospitals across the United States. Ground truth was based off imaging interpretations from two board-certified neuroradiologists. Results: There were 255 positive and 559 negative ICH cases. Accuracy was 95.6%, sensitivity was 91.4%, and specificity was 97.5% for the ICH tool. ICH was further stratified into the following subtypes: intraparenchymal, intraventricular, epidural/subdural, and subarachnoid with true positive rates of 92.9, 100, 94.3, and 89.9%, respectively. ICH true positive rates by volume [small (<5 mL), medium (5-25 mL), and large (>25 mL)] were 71.8, 100, and 100%, respectively. There were 156 positive and 222 negative LVO cases. The LVO tool demonstrated an accuracy of 98.1%, sensitivity of 98.1%, and specificity of 98.2%. A subset of 55 randomly selected cases were also assessed for LVO detection at various sites, including the distal internal carotid artery, middle cerebral artery M1 segment, proximal middle cerebral artery M2 segment, and distal middle cerebral artery M2 segment with an accuracy of 97.0%, sensitivity of 94.3%, and specificity of 97.4%. Conclusion: Deep learning tools can be effective in the detection of both ICH and LVO across a wide variety of hospital systems. While some limitations were identified, specifically in the detection of small ICH and distal M2 occlusion, this study highlights a deep learning tool that can assist radiologists in the detection of emergent findings in a variety of practice settings.

Clinical Significance • Aortic dissection is associated with high rates of morbidity and mortality • Mortality rate of 1-2% per hour during first 48 hours • Early diagnosis and prompt intervention greatly improve patient outcomes • FDA 510k-approved software application expedites detection, triage, and ultimately treatment of patients with suspected aortic dissection • Viz Aortic Dissection algorithm, in collaboration with Avicenna.ai • Objective to evaluate performance of AI algorithm across diverse clinical settings Study Methods • Large-scale, blinded algorithm validation study • 1303 retrospectively collected chest and thoraco-abdominal CT angiography images • Diverse representation of hospitals in 200+ U.S. cities • Ground-truth consensus diagnosis determined by three board-certified radiologists Study Results • 1166 (89.5%) dissection-negative exams, 137 (10.5%) dissection-positive exams • Sensitivity: 94.2% • [95% CI: 88.8% - 97.5%] • Specificity: 97.3% • [95% CI: 96.2% - 98.1%] • PPV of 80.1%, NPV of 99.3% • 8 false negatives, largely complex cases • 32 false positives, largely result of imaging quality Clinical Takeaways • Real-world validation of a deep learning AI-based detection algorithm for suspected aortic dissection • Allows for rapid patient triage à earlier diagnoses à accelerated care coordination à timely initiation of life-saving interventions à better patient outcomes Citations • Gawinecka J, Schönrath F, von Eckardstein A. Acute aortic dissection: pathogenesis, risk factors and diagnosis. Swiss Med Wkly. 2017 Aug 25;147:w14489. doi: 10.4414/smw.2017.14489. PMID: 28871571. • Gudbjartsson T, Ahlsson A, Geirsson A, Gunn J, Hjortdal V, Jeppsson A, Mennander A, Zindovic I, Olsson C. Acute type A aortic dissection – a review. Scand Cardiovasc J. 2020 Feb;54(1):1-13. doi: 10.1080/14017431.2019.1660401. Epub 2019 Sep 23. PMID: 31542960. • Harris KM, Nienaber CA, Peterson MD, et al. Early Mortality in Type A Acute Aortic Dissection: Insights From the International Registry of Acute Aortic Dissection. JAMA Cardiol. 2022;7(10):1009–1015.doi:10.1001/jamacardio.2022.2718

This multicenter retrospective study evaluated the diagnostic performance of a deep learning (DL)-based application for detecting, classifying, and highlighting suspected aortic dissections (ADs) on chest and thoraco-abdominal CT angiography (CTA) scans. CTA scans from over 200 U.S. and European cities acquired on 52 scanner models from six manufacturers were retrospectively collected and processed by CINA-CHEST (AD) (Avicenna.AI, La Ciotat, France) device. The diagnostic performance of the device was compared with the ground truth established by the majority agreement of three U.S. board-certified radiologists. Furthermore, the DL algorithms time to notification was evaluated to demonstrate clinical effectiveness. The study included 1303 CTAs (mean age 58.8 ± 16.4 years old, 46.7% male, 10.5% positive). The device demonstrated a sensitivity of 94.2% [95% CI: 88.8-97.5%] and a specificity of 97.3% [95% CI: 96.2-98.1%]. The application classified positive cases by the AD type with an accuracy of 99.5% [95% CI: 98.9-99.8%] for type A and 97.5 [95% CI: 96.4-98.3%] for type B. The application did not miss any type A cases. The device flagged 32 cases incorrectly, primarily due to acquisition artefacts and aortic pathologies mimicking AD. The mean time to process and notify of potential AD cases was 27.9 ± 8.7 s. This deep learning-based application demonstrated a strong performance in detecting and classifying aortic dissection cases, potentially enabling faster triage of these urgent cases in clinical settings.