Quantifying iodine concentration in the normal bowel wall using dual-energy CT: influence of patient and contrast characteristics.

This study aimed to establish quantitative references of the normal bowel wall iodine concentration (BWIC) using dual energy CT (DECT). This single-center retrospective study included 248 patients with no history of gastrointestinal disease who underwent abdominal contrast-enhanced DECT between January and April 2022. The BWIC was normalized by the iodine concentration of upper abdominal organs (BWICorgan,) and the iodine concentration (IC) of the aorta (BWICaorta). BWIC decreased from the stomach to the rectum (mean 2.16 ± 0.63 vs. 2.19 ± 0.63 vs. 2.1 ± 0.58 vs. 1.67 ± 0.47 vs. 1.31 ± 0.4 vs. 1.18 ± 0.34 vs. 0.94 ± 0.26 mgI/mL for the stomach, duodenum, jejunum, ileum, right colon, left colon and rectum, respectively; P < 0.001). By multivariate analysis, BWIC was associated with a higher BMI (OR:1.01, 95% CI 1.00-1.02, P < 0.001) and with a higher injected contrast dose (OR: 1.51; 95% CI 1.36-1.66, P < 0.001 and 2.06; 95% CI 1.88-2.26, P < 0.001 for 500 mgI/kg and 600 mgI/kg doses taking 400 mgI/kg dose as reference). The BWICorgan was shown independent from patients and contrast-related variables while the BWICaorta was not. BWIC varies according to bowel segments and is dependent on the total iodine dose injected. It shall be normalized with the IC of the upper abdominal organs.

Short-term mortality prediction using a combination of clinical and CT features: Refining the prognosis of critically ill patients in shock.

PURPOSE: To assess the predictive value of combining CT and clinical findings for predicting 10-day mortality in critically ill patients in shock. MATERIALS AND METHODS: From January 1, 2018, to December 31, 2021, 289 consecutives critically ill patients in shock who underwent a contrast enhanced CT were included. Variables at the time of the CT were retrospectively extracted from medical charts. CT examinations were blindly analyzed by two independent radiologists. Multivariable analysis was performed, combining clinical and CT features. A simple survival score for 10-day mortality prediction was built and validated in a further independent external cohort of 70 patients. RESULTS: 10-day mortality rate was 135/289 (47%) in the study sample. At multivariate analysis, catecholamine infusion (OR = 2.11; 95%CI [1.21-4.18], P = 0.011), lactates level > 5 mmol/l (OR = 3.54; 95%CI [1.94-6.54], P < 0.001); total bilirubin > 50 mg/l (OR = 1.79 CI 95% [1.03-3.13], P = 0.039); small bowel dilation (OR = 1.82; 95%CI [1.01-3.32], P = 0.047); diffuse kidney infarction (OR = 2.76; 95%CI [1.26-6.37], P = 0.013) and superior mesentery artery < 5 mm (OR = 1.96; 95%CI [1.10-3.49], P = 0.021) were associated with 10-days mortality. The AUC of the combined model was 0.79; 95%CI [0.74-0.85] in the study sample and 0.87; 95%CI [0.71-0.91] in the validation cohort. CONCLUSION: The combination of CT imaging features and clinical data should emerge as a novel approach to predict short-term mortality in critically ill patients in shock.

Score to Predict the Occurrence of Pneumothorax After Computed Tomography-guided Percutaneous Transthoracic Lung Biopsy.

PURPOSE: The main objective of this study was to identify risk factors for post-percutaneous transthoracic lung biopsy (PTLB) pneumothorax and to establish and validate a predictive score for pneumothorax occurrence to identify patients eligible for outpatient care. MATERIAL AND METHODS: Patients who underwent PTLB between November 1, 2012 and March 1, 2017 were retrospectively evaluated for clinical and radiologic factors potentially related to pneumothorax occurrence. Multivariate logistic regression was used to identify risk factors, and the model coefficient for each factor was used to compute a score. Then, a validation cohort was prospectively evaluated from March 2018 to October 2019. RESULTS: Among the 498 eligible patients in the study cohort, pneumothorax occurred in 124 patients (24.9%) and required drainage in 34 patients (6.8%). Pneumothorax risk factors were chronic obstructive pulmonary disease (OR 95% CI 2.28[1.18-4.43]), several passages through the pleura (OR 95% CI 7.71[1.95-30.48]), an anterior biopsy approach (OR 95% CI 6.36 3.82-10.58]), skin-to-pleura distance ≤30 mm (OR 95% CI 2.25[1.09-6.65]), and aerial effusion >10 mm (OR 95% CI 9.27 [5.16-16.65]). Among the 236 patients in the prospective validation cohort, pneumothorax occurred in 18% and 8% were drained. A negative score (<73 points) predicted a probability of pneumothorax occurrence of 7.4% and late evacuation of 2.5% (OR 95% CI respectively 0.18[0.08-0.39] and 0.15[0.04-0.55]) and suggested a reduced length of hospital stay (P=0.009). CONCLUSION: This predictive score for pneumothorax secondary to PTLB has high prognostic performance and accuracy to direct patients toward outpatient management. CLINICAL TRIALS: NCT03488043.

Insights into acute mesenteric ischaemia: an up-to-date, evidence-based review from a mesenteric stroke centre unit.

Radiologists play a central role in the diagnostic and prognostic evaluation of patients with acute mesenteric ischaemia (AMI). Unfortunately, more than half of AMI patients undergo imaging with no prior suspicion of AMI, making identifying this disease even more difficult. A confirmed diagnosis of AMI is ideally made with dynamic contrast-enhanced CT but the diagnosis may be made on portal-venous phase images in appropriate clinical settings. AMI is diagnosed on CT based on the identification of vascular impairment and bowel ischaemic injury with no other cause. Moreover, radiologists must evaluate the probability of bowel necrosis, which will influence the treatment options.AMI is usually separated into different entities: arterial, venous, non-occlusive and ischaemic colitis. Arterial AMI can be occlusive or stenotic, the dominant causes being atherothrombosis, embolism and isolated superior mesenteric artery (SMA) dissection. The main finding in the bowel is decreased wall enhancement, and necrosis can be suspected when dilatation >25 mm is identified. Venous AMI is related to superior mesenteric vein (SMV) thrombosis as a result of a thrombophilic state (acquired or inherited), local injury (cancer, inflammation or trauma) or underlying SMV insufficiency. The dominant features in the bowel are hypoattenuating wall thickening with submucosal oedema. Decreased enhancement of the involved bowel suggests necrosis. Non-occlusive mesenteric ischaemia (NOMI) is related to impaired SMA flow following global hypoperfusion associated with low-flow states. There are numerous findings in the bowel characterised by diffuse extension. An absence of bowel enhancement and a thin bowel wall suggest necrosis in NOMI. Finally, ischaemic colitis is a sub-entity of arterial AMI and reflects localised colon ischaemia-reperfusion injury. The main CT finding is a thickened colon wall with fat stranding, which seems to be unrelated to SMA or inferior mesenteric artery lesions. A precise identification and description of vascular lesions, bowel involvement and features associated with transmural necrosis is needed to determine patient treatment and outcome.

Left gastric vein embolization during TIPS placement for acute variceal bleeding has no effect on bleeding recurrence: Results of a multicenter study.

PURPOSE: The purpose of this study was to evaluate whether concomitant left gastric vein embolization (LGVE) during transjugular intrahepatic portosystemic shunt (TIPS) for acute variceal hemorrhage could reduce the risk of bleeding recurrence. MATERIAL AND METHOD: A national multicenter observational study was conducted in 14 centers between January 2019 and December 2020. All cirrhotic patients who underwent TIPS placement for acute variceal bleeding were included. During TIPS procedure, size of left gastric vein (LGV), performance of LGVE, material used for LGVE and portosystemic pressure gradient (PPG) before and after TIPS placement were collected. A propensity score for the occurrence of LGVE was calculated to assess effect of LGVE on rebleeding recurrence at six weeks and one year. RESULTS: A total of 356 patients were included (mean age 57.3 ± 10.8 [standard deviation] years; 283/356 [79%] men). Median follow-up was 11.2 months [interquartile range: 1.2, 13.3]. The main indication for TIPS was pre-emptive TIPS (162/356; 46%), rebleeding despite secondary prophylaxis (105/356; 29%), and salvage TIPS (89/356; 25%). Overall, 128/356 (36%) patients underwent LGVE during TIPS procedure. At six weeks and one year, rebleeding-free survival did not differ significantly between patients who underwent LGVE and those who did not (6/128 [5%] vs. 15/228 [7%] at six weeks, and 11/128 [5%] vs. 22/228 [7%] at one year, P = 0.622 and P = 0.889 respectively). A total of 55 pairs of patients were retained after propensity score matching. In patients without LGVE, the rebleeding rate was not different from those with LGVE (3/55 [5%] vs. 4/55 [7%], P > 0.99, and 5/55 [9%] vs. 6/55[11%], P > 0.99, at six weeks and one year respectively). Multivariable analysis identified PPG after TIPS placement as the only predictor of bleeding recurrence (hazard ratio = 1.09; 95% confidence interval: 1.02-1.18; P = 0.012). CONCLUSION: In this multicenter national real-life study, we did not observe any benefit of concomitant LGVE during TIPS placement for acute variceal bleeding on bleeding recurrence rate.

Increased blood lactate during enteral nutrition in the critically ill: associations and treatment options.

PURPOSE OF REVIEW: To evaluate the significance of blood lactate increase during enteral nutrition in the critically ill, and to propose diagnostic and therapeutic strategies. RECENT FINDINGS: Acute mesenteric ischemia occurs in approximately 1% of critically ill patients treated with catecholamine. Recent literature suggests that enteral nutrition is a risk factor of acute mesenteric ischemia, in particular in case of low cardiac output, by a mechanism of nonocclusive mesenteric ischemia. The association of clinical, biological, and computed tomography imaging might help to evaluate the reversibility of acute mesenteric ischemia. SUMMARY: As enteral nutrition induces an increased metabolic work of the gut, the inadequation between oxygen delivery and demand exposes the gut to a phenomenon of nonocclusive mesenteric ischemia. Before initiation of enteral nutrition,, and before each increase of the enteral nutrition dose, the risk factors of nonocclusive mesenteric ischemia should be searched in order to prevent it. While under enteral nutrition, increased lactate concentration while receiving enteral nutrition requires the urgent search for nonocclusive mesenteric ischemia, and the adaptation of enteral nutrition (reduction, stop, and/or switch to parenteral nutrition or tolerate early nutrient restriction). Early signs of nonocclusive mesenteric ischemia should be searched in order to allow for a rapid diagnosis, before development of irreversible transmural necrosis. After the diagnosis of acute mesenteric ischemia, improving the balance between oxygen demand and delivery to the gut, evaluating the reversibility of the gut ischemia, and performing urgent resection in case of irreversible transmural necrosis should be the main objectives. After the resolution of acute mesenteric ischemia, the benefit risk analysis of enteral nutrition reintroduction should be evaluated.

The Angers CT Score is a Risk Factor for the Failure of the Conservative Management of Adhesive Small Bowel Obstruction: A Prospective Observational Multicentric Study.

BACKGROUND: Identifying the 30% of adhesive small bowel obstructions (aSBO) for which conservative management will require surgery is essential. The association between the previously described radiological score and failure of the conservative management of aSBO remains to be confirmed in a large prospective multicentric cohort. Our aim was to assess the risk factors of failure of the conservative management of aSBO considering the radiological score. MATERIAL AND METHODS: This prospective observational study took place in 15 French centers over 3 months. Consecutive patients experiencing aSBO with no early surgery were included. The six radiological features from the Angers radiological computed tomography (CT) score were noted (beak sign, closed loop, focal or diffuse intraperitoneal liquid, focal or diffuse mesenteric haziness, focal or diffuse mesenteric liquid, and diameter of the most dilated small bowel loop > 40 mm). RESULTS: Two hundred and seventy nine patients with aSBO were screened. Sixty patients (21.5%) underwent early surgery, and 219 (78.5%) had primary conservative management. In the end, 218 patients were included in the analysis of the risk factors for conservative treatment failure. Among them, 162 (74.3%) had had successful management while for 56 (25.7%) management had failed. In multivariate analysis, a history of surgery was not a significant risk factor for the failure of conservative treatment (OR = 0.11; 95%CI = 0-1.23). A previous episode of aSBO was protective against the failure of conservative treatment (OR = 0.36; 95%CI = 0.15-0.85) and an Angers CT score ≥ 5 as the only individual risk factor (OR = 2.39; 95%CI = 1.01-5.69). CONCLUSION: The radiological score of aSBO is a promising tool in improving the management of aSBO patients. A first episode of aSBO and/or a radiological score ≥5 should lead physicians to consider early surgical management.

Abdominal atherosclerosis is not a risk factor of nonocclusive mesenteric ischemia among critically ill patients: a propensity matching study.

BACKGROUND: Although risk factors of occlusive acute mesenteric ischemia are well known, triggering factors of nonocclusive mesenteric ischemia (NOMI) remain unclear. Alongside to the known risk factors for NOMI, the role of atherosclerosis is not fully elucidated. The purpose of our study was to evaluate whether abdominal atherosclerosis is a risk factor for NOMI. METHODS: From January 2018 to December 2021, all consecutive patients admitted to the intensive care unit who underwent contrast-enhanced CT for suspicion of NOMI were evaluated for inclusion. Clinical and biological data at the time of the CT scan were retrospectively extracted from medical charts and reviewed by a single radiologist. The cohorts were matched by a 1:1 propensity score based on the patient clinical, biological data, and abdominal CT features associated with NOMI. Noncontrast CT acquisitions were used to calculate calcium scores of the abdominal aorta, celiac trunk, superior mesenteric artery (SMA), and common iliac artery according to the Agatston method. Analyses were performed before and after propensity score matching. RESULTS: Among the 165 critically ill patients included, 59 (36%) had NOMI. Before matching analysis, the SMA and total abdominal Agatston calcium scores were not different between patients without and with NOMI (52.00 [IQR = 0, 473] vs. 137.00 [IQR = 0, 259], P = 0.857, respectively, and 7253 [IQR = 1220, 21738] versus 5802 [IQR = 2075, 15,084]; P = 0.723). The results were similar after matching 38 patients with NOMI and 38 without: 153 [IQR = 0, 665] versus 85 [IQR = 0, 240] (P = 0.312) for the SMA calcium score, and 7915 [IQR = 1812, 21561] versus 4139 [IQR = 1440, 9858] (P = 0.170) for the total abdominal Agatston calcium score. CONCLUSION: Our results suggest that atherosclerosis is not a risk factor for NOMI in critically ill patients.

Added value of an artificial intelligence solution for fracture detection in the radiologist's daily trauma emergencies workflow.

PURPOSE: The main objective of this study was to compare radiologists' performance without and with artificial intelligence (AI) assistance for the detection of bone fractures from trauma emergencies. MATERIALS AND METHODS: Five hundred consecutive patients (232 women, 268 men) with a mean age of 37 ± 28 (SD) years (age range: 0.25-99 years) were retrospectively included. Three radiologists independently interpreted radiographs without then with AI assistance after a 1-month minimum washout period. The ground truth was determined by consensus reading between musculoskeletal radiologists and AI results. Patient-wise sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for fracture detection and reading time were compared between unassisted and AI-assisted readings of radiologists. Their performances were also assessed by receiver operating characteristic (ROC) curves. RESULTS: AI improved the patient-wise sensitivity of radiologists for fracture detection by 20% (95% confidence interval [CI]: 14-26), P< 0.001) and their specificity by 0.6% (95% CI: -0.9-1.5; P = 0.47). It increased the PPV by 2.9% (95% CI: 0.4-5.4; P = 0.08) and the NPV by 10% (95% CI: 6.8-13.3; P < 0.001). Thanks to AI, the area under the ROC curve for fracture detection of readers increased respectively by 10.6%, 10.2% and 9.9%. Their mean reading time per patient decreased by respectively 10, 16 and 12 s (P < 0.001). CONCLUSIONS: AI-assisted radiologists work better and faster compared to unassisted radiologists. AI is of great aid to radiologists in daily trauma emergencies, and could reduce the cost of missed fractures.

Post-Transplantation Cytomegalovirus Infection Interplays With the Development of Anastomotic Biliary Strictures After Liver Transplantation.

Background: Anastomotic biliary stricture (ABS) remains the most frequent complication after liver transplantation (LT). This study aimed to identify new anastomotic biliary stricture risk factors, with a specific focus on postoperative events. Additionally, ABS management and impact on patient and graft survival were assessed. Methods: All consecutive patients who underwent LT with duct-to-duct anastomosis between 2010 and 2019 were included. All patients who died within 90 days after LT due to non-ABS-related causes were excluded. Results: Among 240 patients, 65 (27.1%) developed ABS after a median time of 142 days (range, 13-1265). Median follow-up was 49 months (7-126). Upon multivariable analysis, donor BMI (OR=0.509, p = 0.037), post-LT CMV primoinfection (OR = 5.244, p < 0.001) or reactivation (OR = 2.421, p = 0.015) and the occurrence of post-LT anastomotic biliary fistula (OR = 2.691, p = 0.021) were associated with ABS. Anastomotic technical difficulty did not independently impact the risk of ABS (OR = 1.923, p = 0.051). First-line ABS treatment was systematically endoscopic (100%), and required a median of 2 (range, 1-11) procedures per patient. Repeat LT was not required in patients developing ABS. The occurrence of ABS was not associated with overall patient survival (p = 0.912) nor graft survival (p = 0.521). Conclusion: The risk of developing ABS after LT seems driven by the occurrence of postoperative events such as CMV infection and anastomotic fistula. In this regard, the role of CMV prophylaxis warrants further investigations.

Preoperative prediction of inadvertent enterotomy during adhesive small bowel obstruction surgery using combination of CT features.

OBJECTIVES: The purpose of this study was to identify the preoperative CT features that are associated with inadvertent enterotomy (IE) during adhesive small bowel obstruction (ASBO) surgery. METHODS: From January 2015 to December 2019, all patients with ASBO who underwent an abdominal CT were reviewed. Abdominal CT were retrospectively reviewed by two radiologists with a consensus read in case of disagreement. IE during ASBO surgery was retrospectively recorded. Univariate and multivariate analyses of CT features associated with IE were performed and a simple CT score was built to stratify the risk of IE. This score was validated in an independent retrospective cohort. Abdominal CT of the validation cohort was reviewed by a third independent reader. RESULTS: Among the 368 patients with ASBO during the study period, 169 were surgically treated, including 129 ASBO for single adhesive band and 40 for matted adhesions. Among these, there were 47 IE. By multivariate analysis, angulation of the transitional zone (OR = 4.19, 95% CI [1.10-18.09]), diffuse intestinal adhesions (OR = 4.87, 95% CI [1.37-19.76]), a fat notch sign (OR = 0.32, 95% CI [0.12-0.85]), and mesenteric haziness (OR = 0.13, 95% CI [0.03-0.48]) were independently associated with inadvertent enterotomy occurrence. The simple CT score built to stratify risk of IE displayed an AUC of 0.85 (95% CI [0.80-0.90]) in the study sample and 0.88 (95% CI [0.80-0.96]) in the validation cohort. CONCLUSION: A simple preoperative CT score is able to inform the surgeon about a high risk of IE and therefore influence the surgical procedure. KEY POINTS: • In this retrospective study of 169 patients undergoing abdominal surgery for adhesive small bowel obstruction, 47 (28%) inadvertent enterotomy occurred. • A simple preoperative CT score enables accurate stratification of inadvertent enterotomy risk (area under the curve 0.85). • By multivariable analysis, diffuse intestinal adhesions and angulation of the transitional zone were predictive of inadvertent enterotomy occurrence.

Liver spontaneous hypoattenuation on CT is an imaging biomarker of the severity of acute pancreatitis.

PURPOSE: The purpose of this study was to evaluate the relationship between liver spontaneous attenuation (LSA) on computed tomography (CT) reflecting the degree of steatosis, and the severity of acute pancreatitis (AP). MATERIALS AND METHODS: All consecutive patients admitted from December 2014 to September 2020 for an episode of AP were retrospectively reviewed. LSA was evaluated on early CT examination and all abdominal CT examinations were reviewed by two abdominal radiologists. Severity of AP was categorized using Atlanta classification and CT severity index. Univariable and multivariable statistical analyses were performed. RESULTS: A total of 467 patients were included. There were 297 men and 170 women, with a mean age of 57 ± 19 (SD) years (range: 18-98 years). Among them, 236 patients (51%) had acute biliary pancreatitis, 134 (29%) had acute alcoholic pancreatitis and 97 (20%) had AP due to other etiologies. A total of 44 (9%) patients had severe AP and 423 (91%) had non severe AP. Median LSA was significantly lower in patients with severe AP (36 Hounsfield units [HU]; interquartile range [IQR]:18; 54) than in patients with non-severe AP (45 HU; IQR: 35; 51) (P < 0.001). In patients with alcoholic AP, median LSA was significantly lower in patients with severe AP (29 HU; IQR: 3; 43) than in those with non-severe AP (42 HU; IQR: 27; 50) (P = 0.022). At multivariable analysis, the third and fourth quartiles of liver spontaneous attenuation values (i.e., < 45 HU/>30 HU and < 30 HU) were independently associated with severe AP (OR, 3.23; 95% CI: 1.33-51.2; P = 0.038 and OR, 8.82; 95% CI: 1.91-69.7; P = 0.014; respectively). CONCLUSION: LSA on CT is associated with clinical severity of AP and may be used as an additional marker of disease severity.

Systematic screening for advanced liver fibrosis in patients with coronary artery disease: The CORONASH study.

Although coronary artery disease (CAD) and advanced liver fibrosis (AdLF) are commonly associated in patients with non-alcoholic fatty liver disease (NAFLD), the prevalence of AdLF and the diagnostic performance of non-invasive fibrosis tests (NITs) in CAD patients remains unknown. We aimed to prospectively screen for AdLF in patients with documented CAD using NITs and Fibroscan. High and intermediate zones of NITs were combined to define AdLF. AdLF was suspected whenever APRI ≥ 0.5, Forns index ≥ 4.2, NAFLD fibrosis score (NFS) ≥ -1.455/0.12 for age

Prognostic Factors in Non-occlusive mesenteric ischemia: A pragmatic pre-operative score for the prediction of 28-day mortality.

BACKGROUND: The prognosis of critical ill patients with non-occlusive mesenteric ischemia (NOMI) is poor and not fully understood. We aimed to determine preoperative factors associated with 28-day mortality in NOMI. METHODS: Variables associated with 28-day mortality were entered into a multivariate cox regression model and were used to compute a NOMI mortality score. RESULTS: 154 patients were included. The 28-day mortality rate was 56%. Multivariable analyses including variables at the time of the CT identified three variables (i.e. lactates > 7 mmoL/l, prothrombin rate <60% and kidney infarction), included in a simple score. Among the study population, the probability of 28-day mortality was 26% (11/42), 54% (26/48), 77% (23/30) and 100% (21/21) for a survival score of 0, 1, 2 and 3, respectively. CONCLUSION: A simple score combining these three variables, calculated preoperatively, was able to accurately predict 28-day mortality and might help to avoid futile laparotomies.

Role of the radiologist in the diagnosis and management of the two forms of hepatic echinococcosis.

Echinococcosis is a parasitic disease caused by two zoonotic tapeworms (cestodes) of the Echinocococcus genus. It can be classified as either alveolar or cystic echinococcosis. Although the two forms differ significantly in terms of imaging findings, they share similarities in terms of management and treatment. In parallel to medical treatment with albendazole (ABZ), and surgery, historically used in these diseases, various imaging-guided interventional procedures have recently emerged (drainage, stenting, or Puncture, aspiration, injection, and reaspiration (PAIR)). These options open up a new range of therapeutic options. As in oncology, multidisciplinary consultation meetings now play a major role in adapted management and patient care in hepatic echinococcosis. Consequently, diagnostic imaging and interventional expertise have brought radiologists to the fore as important members of these multidisciplinary team. The radiologist will need to evaluate parasite activity in both forms of the disease, to guide the choice of the appropriate therapy from among medical treatment, interventional radiology procedures and/or surgical treatment. Knowledge of the specific complications of the two forms of echinococcosis will also help radiologists to discuss the appropriate treatment and management. The aim of this review is to describe the core knowledge that what a radiologist should possess to actively participate in multidisciplinary meetings about hepatic echinococcosis. We discuss the role of imaging, from diagnosis to treatment, in alveolar (AE) and cystic echinococcosis (CE), respectively.

How artificial intelligence improves radiological interpretation in suspected pulmonary embolism.

OBJECTIVES: To evaluate and compare the diagnostic performances of a commercialized artificial intelligence (AI) algorithm for diagnosing pulmonary embolism (PE) on CT pulmonary angiogram (CTPA) with those of emergency radiologists in routine clinical practice. METHODS: This was an IRB-approved retrospective multicentric study including patients with suspected PE from September to December 2019 (i.e., during a preliminary evaluation period of an approved AI algorithm). CTPA quality and conclusions by emergency radiologists were retrieved from radiological reports. The gold standard was a retrospective review of CTPA, radiological and clinical reports, AI outputs, and patient outcomes. Diagnostic performance metrics for AI and radiologists were assessed in the entire cohort and depending on CTPA quality. RESULTS: Overall, 1202 patients were included (median age: 66.2 years). PE prevalence was 15.8% (190/1202). The AI algorithm detected 219 suspicious PEs, of which 176 were true PEs, including 19 true PEs missed by radiologists. In the cohort, the highest sensitivity and negative predictive values (NPVs) were obtained with AI (92.6% versus 90% and 98.6% versus 98.1%, respectively), while the highest specificity and positive predictive value (PPV) were found with radiologists (99.1% versus 95.8% and 95% versus 80.4%, respectively). Accuracy, specificity, and PPV were significantly higher for radiologists except in subcohorts with poor-to-average injection quality. Radiologists positively evaluated the AI algorithm to improve their diagnostic comfort (55/79 [69.6%]). CONCLUSION: Instead of replacing radiologists, AI for PE detection appears to be a safety net in emergency radiology practice due to high sensitivity and NPV, thereby increasing the self-confidence of radiologists. KEY POINTS: • Both the AI algorithm and emergency radiologists showed excellent performance in diagnosing PE on CTPA (sensitivity and specificity ≥ 90%; accuracy ≥ 95%). • The AI algorithm for PE detection can help increase the sensitivity and NPV of emergency radiologists in clinical practice, especially in cases of poor-to-moderate injection quality. • Emergency radiologists recommended the use of AI for PE detection in satisfaction surveys to increase their confidence and comfort in their final diagnosis.