Artificial intelligence

Early diagnosis of deep venous thrombosis is essential for reducing complications, such as recurrent pulmonary embolism and venous thromboembolism.

202205980001) and the Gachon Program (GCU-202106290001). [5](/articles/s41598-022-25849-0#Fig5) shows examples of the FP case. [4](/articles/s41598-022-25849-0#Fig4) shows the images of the detection result for each backbone based on the type of data used. This study attempted to prove the positive aspects of the proposed method by comparing and analyzing the results by applying the method to different AI algorithms for the effect of the proposed algorithm on the AI model. Figure [5](/articles/s41598-022-25849-0#Fig5)b shows a case of successful localization and unsuccessful classification. [2](/articles/s41598-022-25849-0#Tab2), the two models that used the suggested synthesized data performed better in terms of Sn and mAP values. [4](/articles/s41598-022-25849-0#ref-CR4). The sensitivity (Sn and recall), FPs per image (FPPI), and precision were calculated using the model’s evaluation indicators. [1](/articles/s41598-022-25849-0#ref-CR1). In the model based on the ResNet50 backbone using the same proposed data, the performances approached 0.843 (± 0.037) Sn, 0.608 (± 0.139) FPPI, 0.610 (± 0.061) precision, and 0.807 (± 0.040) mAP. The RetinaNet has a feature pyramid network combined with the ResNet backbone [14](/articles/s41598-022-25849-0#ref-CR14). Axial images were obtained in digital imaging and communication in medicine (DICOM) format with a 5-mm slice thickness and 5-mm slice interval.