Skip to main content
. 2020 Oct 22;8:194158–194165. doi: 10.1109/ACCESS.2020.3033069

FIGURE 2.

FIGURE 2.

Deep convolutional neural network (DCNN) architecture for the machine diagnosis of NCIPs. The training of this DCNN is a two-stage processes. In the first stage, we train the DCNN to distinguish the region of interests (ROIs) with the likelihood of NCIP into two classes: NCIPs or otherwise. In the second stage, we fine-tune the DCNN to predict the case-based probability of COVID-19 by ensembling the most significant location proposals self-produced in the DCNN.