The RSNA Pulmonary Embolism CT Dataset.
Ontology highlight
ABSTRACT:
SUBMITTER: Colak E
PROVIDER: S-EPMC8043364 | biostudies-literature | 2021 Mar
REPOSITORIES: biostudies-literature
Publications
The RSNA Pulmonary Embolism CT Dataset.
Colak Errol E Kitamura Felipe C FC Hobbs Stephen B SB Wu Carol C CC Lungren Matthew P MP Prevedello Luciano M LM Kalpathy-Cramer Jayashree J Ball Robyn L RL Shih George G Stein Anouk A Halabi Safwan S SS Altinmakas Emre E Law Meng M Kumar Parveen P Manzalawi Karam A KA Nelson Rubio Dennis Charles DC Sechrist Jacob W JW Germaine Pauline P Lopez Eva Castro EC Amerio Tomas T Gupta Pushpender P Jain Manoj M Kay Fernando U FU Lin Cheng Ting CT Sen Saugata S Revels Jonathan Wesley JW Brussaard Carola C CC Mongan John J
Radiology. Artificial intelligence 20210120 2
<i>Supplemental material is available for this article.</i> ...[more]
Similar Datasets
Project description:ObjectiveTo appraise the evidence on the diagnostic accuracy of CT pulmonary angiography and the prognostic value of a negative CT pulmonary angiogram in the diagnosis of pulmonary embolism.MethodsMedline, EMBASE, and grey literature were systematically searched by two researchers. Any study which compared CT pulmonary angiography to an acceptable reference standard or prospectively followed up a cohort of patients with a normal CT pulmonary angiogram was included. Study methods were appraised independently by two researchers, and data were extracted independently by three researchers.ResultsThirteen diagnostic and 11 follow up studies were identified. Studies varied in prevalence of pulmonary embolism (19-79%), patient groups, and method quality. Few studies recruited unselected emergency department patients. There was heterogeneity in the analysis of sensitivity (53 to 100%), specificity (79 to 100%), and false negative rate (1.0 to 10.7%). The pooled false negative rate of combined negative CT pulmonary angiography and negative deep vein thrombosis testing was 1.5% (95% CI 1.0 to 1.9%).ConclusionDiagnostic studies give conflicting results for the diagnostic accuracy of CT pulmonary angiography. Follow up studies show that CT pulmonary angiography can be used in combination with investigation for deep vein thrombosis to exclude pulmonary embolism.
| S-EPMC2464412 | biostudies-literature
Project description:PurposeTo evaluate pulmonary embolism (PE) prevalence at CT pulmonary angiography in patients testing positive for coronavirus disease 2019 (COVID-19) and factors associated with PE severity.Materials and methodsA retrospective, single-center study evaluated 62 patients who tested positive for COVID-19 who underwent CT pulmonary angiography between March 13 and April 5, 2020. Another 62-patient cohort who underwent CT pulmonary angiography before the first reported local COVID-19 case was retrospectively selected. The relative rate of CT pulmonary angiography positivity was recorded. For the COVID-19 positive cohort, comorbidities, laboratory values, clinical outcome, and venous thrombosis of the patients were recorded. Two thoracic radiologists assessed embolic severity using the Mastora system and evaluated right heart strain. Factors associated with PE and arterial obstruction severity were evaluated by using statistical analysis. A P value < .05 was considered significant.ResultsOf the patients testing positive for COVID-19, 37.1% had PE, higher than 14.5% of pre-COVID-19 patients (P = .007). d-dimer levels closest to CT pulmonary angiography date correlated with the Mastora obstruction score. Receiver operating characteristic analysis identified optimal sensitivity (95%) and specificity (71%) for PE diagnosis at 1394 ng/mL d-dimer units. The mean d-dimer level was 1774 ng/mL and 6432 ng/mL d-dimer units in CT pulmonary angiography-negative and CT pulmonary angiography-positive subgroups, respectively (P < .001). One additional patient with negative results at CT pulmonary angiography had deep venous thrombosis, thus resulting in 38.7% with PE or deep venous thrombosis, despite 40% receiving prophylactic anticoagulation. Other factors did not demonstrate significant PE association.ConclusionA total of 37.1% of COVID-19 patients underwent CT pulmonary angiographic examinations diagnosing PE. PE can be a cause of decompensation in patients testing positive for COVID-19, and d-dimer can be used to stratify patients in terms of PE risk and severity.Supplemental material is available for this article.© RSNA, 2020.
| S-EPMC7336753 | biostudies-literature
Project description:BackgroundEarly diagnosis of the potentially fatal but curable chronic pulmonary embolism (CPE) is challenging. We have developed and investigated a novel convolutional neural network (CNN) model to recognise CPE from CT pulmonary angiograms (CTPA) based on the general vascular morphology in two-dimensional (2D) maximum intensity projection images.MethodsA CNN model was trained on a curated subset of a public pulmonary embolism CT dataset (RSPECT) with 755 CTPA studies, including patient-level labels of CPE, acute pulmonary embolism (APE), or no pulmonary embolism. CPE patients with right-to-left-ventricular ratio (RV/LV) < 1 and APE patients with RV/LV ≥ 1 were excluded from the training. Additional CNN model selection and testing were done on local data with 78 patients without the RV/LV-based exclusion. We calculated area under the receiver operating characteristic curves (AUC) and balanced accuracies to evaluate the CNN performance.ResultsWe achieved a very high CPE versus no-CPE classification AUC 0.94 and balanced accuracy 0.89 on the local dataset using an ensemble model and considering CPE to be present in either one or both lungs.ConclusionsWe propose a novel CNN model with excellent predictive accuracy to differentiate chronic pulmonary embolism with RV/LV ≥ 1 from acute pulmonary embolism and non-embolic cases from 2D maximum intensity projection reconstructions of CTPA.Relevance statementA DL CNN model identifies chronic pulmonary embolism from CTA with an excellent predictive accuracy.Key points• Automatic recognition of CPE from computed tomography pulmonary angiography was developed. • Deep learning was applied on two-dimensional maximum intensity projection images. • A large public dataset was used for training the deep learning model. • The proposed model showed an excellent predictive accuracy.
| S-EPMC10281920 | biostudies-literature
Project description:We report the case of a middle-aged woman on a background of severe respiratory disease who presented with hemoptysis. Investigations with CT pulmonary angiogram showed evidence of a large pulmonary arterial filling defect which was initially treated as a pulmonary embolism. However, despite being anticoagulated, she experienced ongoing hemoptysis. Further imaging several days later using aortic phase CT confirmed that the filling defect was a false positive; the artifact was due to retrograde filling from a systemic arterial-pulmonary arterial shunt and anticoagulation was stopped. Although PE is the most common cause of filling defects, this case suggests that in the setting of patients with chronic inflammatory parenchymal pulmonary disease, clinicians should consider alternative causes such as systemic arterial-pulmonary arterial shunts.
| S-EPMC10023907 | biostudies-literature
Project description:Online supplemental material is available for this article.
| S-EPMC7233384 | biostudies-literature
Project description:BackgroundThe aim of this study was to develop and evaluate a deep neural network model in the automated detection of pulmonary embolism (PE) from computed tomography pulmonary angiograms (CTPAs) using only weakly labelled training data.MethodsWe developed a deep neural network model consisting of two parts: a convolutional neural network architecture called InceptionResNet V2 and a long-short term memory network to process whole CTPA stacks as sequences of slices. Two versions of the model were created using either chest X-rays (Model A) or natural images (Model B) as pre-training data. We retrospectively collected 600 CTPAs to use in training and validation and 200 CTPAs to use in testing. CTPAs were annotated only with binary labels on both stack- and slice-based levels. Performance of the models was evaluated with ROC and precision-recall curves, specificity, sensitivity, accuracy, as well as positive and negative predictive values.ResultsBoth models performed well on both stack- and slice-based levels. On the stack-based level, Model A reached specificity and sensitivity of 93.5% and 86.6%, respectively, outperforming Model B slightly (specificity 90.7% and sensitivity 83.5%). However, the difference between their ROC AUC scores was not statistically significant (0.94 vs 0.91, p = 0.07).ConclusionsWe show that a deep learning model trained with a relatively small, weakly annotated dataset can achieve excellent performance results in detecting PE from CTPAs.
| S-EPMC8919639 | biostudies-literature
Project description:Time and dose related expression profiles of rat right heart tissue in microsphere bead model for Pulmonary embolism Keywords: Time course and dose response in experimental PE
2008-07-01 | GSE6104 | GEO
Project description:PurposeSince 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.MethodsThe 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.ResultsThe 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.ConclusionsThe DL-based algorithm has a high degree of diagnostic accuracy with balanced sensitivity and specificity for the detection of PE on CTAs.
| S-EPMC10093638 | biostudies-literature
Project description:Pulmonary embolism is a potentially fatal condition with increased mortality if anticoagulation is delayed. This study aimed to find influencing factors on the duration from requesting a computed tomography (CT) pulmonary angiography (CTPA) to performing a CTPA in suspected acute pulmonary embolism. In 1849 cases, automatically generated time data were extracted from the radiological information system. The impact of the distance to the scanner, case-related features (sector of patient care, triage), and workload (demand for CTs, performed CTs, available staff, hospital occupancy) were investigated retrospectively using multiple regression. The time to CTPA was shorter in cases from the emergency room (ER) than in inpatients and outpatients at distances below 160 m and 240 m, respectively. While requests from the ER were also performed faster than cases from regular wards (< 180 m), no difference was found between the ER and intensive care units. Compared to "not urgent" cases, the workflow was shorter in "urgent" (- 17%) and "life-threatening" (- 67%) situations. The process was prolonged with increasing demand (+ 5%/10 CTs). The presented analysis identified relevant in-hospital influences on the CTPA workflow, including the distance to the CT together with the sector of patient care, the case triage, and the demand for imaging.
| S-EPMC11021441 | biostudies-literature
Project description:To evaluate dual-layer dual-energy computed tomography (dlDECT)-based characterization of thrombus composition for differentiation of acute pulmonary embolism (PE) and chronic thromboembolic pulmonary hypertension (CTEPH). This retrospective single center cohort study included 49 patients with acute PE and 33 patients with CTEPH who underwent CT pulmonary angiography on a dlDECT from 06/2016 to 06/2022. Conventional images), material specific images (virtual non-contrast [VNC], iodine density overlay [IDO], electron density [ED]), and virtual monoenergetic images (VMI50KeV) were analyzed. Regions-of-interest (ROIs) were manually placed in pulmonary artery thrombi, and morphological imaging characteristics for acute and chronic PE were assessed. Area under the receiver operating characteristics curve (AUC) of ROI measurements, morphological imaging features, and their combination in distinguishing between acute PE and CTEPH were evaluated. Compared to PE, thrombi in patients with CTEPH had lower attenuation on conventional images (Median [inter-quartile range]: 40 [35-47] HU vs 64 [52-83] HU) and VMI50keV reconstructions (59 [46-72] HU vs 101 [80-123] HU) as well as decreased iodine uptake (IDO: 0.5 [0.2-1.0] vs 1.2 [0.5-1.8]; p for all < 0.001). Conventional images and VMI50keV reconstructions were the most accurate for differentiating between acute and chronic thrombi (conventional: AUC 0.92, 95% CI 0.86-0.98; VMI50keV: AUC 0.91, 95% CI 0.85-0.97). Main pulmonary artery (MPA) diameter combined with thrombus attenuation significantly increased the AUC compared to MPA diameter alone (p = 0.002 respectively). Thrombi in patients with CTEPH exhibit lower attenuation and reduced contrast enhancement. Analyzing attenuation in pulmonary thrombi may add diagnostic information to established morphological parameters in differentiating acute PE from CTEPH.
| S-EPMC11811430 | biostudies-literature