Table 1.
Some studies related to the classification and segmentation of pulmonary embolism.
Fig 1.
Illustration of CT pulmonary angiography exams: (a) axial C+ classified as normal and (b) axial C+ classified with the presence of PE indicated in the markings.
Fig 2.
Flowchart of the proposed methodology.
Table 2.
Characteristics of Universal Repository patients.
Table 3.
Dataset information.
Fig 3.
Application of automatic windowing and segmentation: (a) original image, (b) windowing for PE, (c) windowing for lungs, and (d) automatic segmentation.
Fig 4.
CNN-LSTM model metrics for PE classification at slice and stack levels: (a) precision curve, (b) ROC curve, and (c) loss.
Table 4.
Confusion matrices for internal validation.
Table 5.
Precision, recall and f1-score metrics for the internal validation set.
Fig 5.
Classification of exams from the Universal Repository database: (a) Tp, (b) Fp, (c) Tn, and (d) Fn.
Table 6.
Accuracy and confidence interval for age groups for the external database.
Fig 6.
Universal Repository exam heatmaps: (a) Tp, (b) Fp, (c) Tn e (d) Fn.
Table 7.
Comparison between exam reports.
Table 8.
Model and expert performance compared to reports.