Selected article for: "attentional region and deep network"
Author: Cheng Jin; Weixiang Chen; Yukun Cao; Zhanwei Xu; Xin Zhang; Lei Deng; Chuansheng Zheng; Jie Zhou; Heshui Shi; Jianjiang Feng
Title: Development and Evaluation of an AI System for COVID-19
  • Document date: 2020_3_23
    • ID: k1lg8c7q_29
    Snippet: After proper training of the deep network, Guided gradient-weighted Class Activation Mapping (Guided Grad-CAM) [18] was exploited to explain the "black box" system and extract attentional areas which is connected to the back end of the diagnostic model. Figure 5 shows some representative cases for the visualization of Guided Grad-CAM to determine the attentional regions. The original CT slices are in the first column. The second column is the res.....
    Document: After proper training of the deep network, Guided gradient-weighted Class Activation Mapping (Guided Grad-CAM) [18] was exploited to explain the "black box" system and extract attentional areas which is connected to the back end of the diagnostic model. Figure 5 shows some representative cases for the visualization of Guided Grad-CAM to determine the attentional regions. The original CT slices are in the first column. The second column is the result of pseudo-color display of the feature map. The third column is the gradient map in the region of attention. We found that the spatial distribution of the attentional region, morphology and the texture within it are consistent with the characteristics of COVID-19 as reported in previous manual diagnosis studies [4, 19] .

    Search related documents:
    Co phrase search for related documents
    • black box system and box system: 1, 2, 3, 4, 5
    • CT slice and deep network: 1, 2, 3
    • CT slice and diagnosis study: 1
    • deep network and diagnosis study: 1, 2, 3, 4, 5, 6
    • deep network and diagnostic model: 1, 2, 3, 4, 5, 6
    • deep network and feature map: 1, 2, 3, 4, 5, 6
    • deep network and gradient map: 1, 2
    • feature map and gradient map: 1