Fig. 4From: Automated detection of pulmonary embolism from CT-angiograms using deep learningThe figure shows example slices where both the slice-based and stack-based predictions are a true positives, b false positives or c false negatives. In the left panel (the first two images on each row from left to right) are images classified by Model A and in the right panel (the last two images on each row) are images classified by Model BBack to article page