A Review of Computer-Aided Detection of Pulmonary Embolism in CT Angiograms

pulmonary computed tomography angiography (CTA) is widely used in the detection of pulmonary embolism. Using these thin slices for studying clinical function of lung is necessary, but it is time-consuming for radiologists. Therefore, several systems have been designed for detection of embolism. The aim of this study was to review and evaluate systems of detection of pulmonary embolism.
In this study, all the articles in databases including IEEE, Science Direct, Medical Physics and SPIE in the field of processing pulmonary computed tomography angiograms and computer-aided detection systems were reviewed. A total of 16 articles between 2002 and 2015 were extracted. Each article presented one CAD system that was evaluated with quantitative and qualitative analysis. To assess systems, the number of images in database, sensitivity and false positive error rate were used.
Various systems are designed to detect embolism. Image database is affected by different factors such as pulmonary diseases, noise, the time of taking image, diversity of pulmonary embolism in main and peripheral branches and the number of data. The maximum number of data was 177 samples and the minimum was 3 samples. Sensitivity and the number of false positive errors are good criteria. Among the CAD systems, the sensitivity was 50% to 100% and the rate of false positive was 2- 22.6 per scan.
According to the results, in addition to having high sensitivity and low false positive error, detection systems must have large image database of embolism in various branches of vessels.