Skip to main content
. 2022 Jun 27;252:109278. doi: 10.1016/j.knosys.2022.109278

Fig. 1.

Fig. 1

Illustration of the pipeline of the proposed framework. For each CT image, a bag of instances (2D image slices) is constructed through the preprocess steps, which is fed into the explainable multi-instance multi-task network (EMTN) for the diagnosis of COVID-19 and multi-lesion segmentation. Following EMTN, explainability analysis endows the diagnosis results with explainability, lesion quantification and severity assessment provides the detailed information of patient status.