Figure 6 | Scientific Reports

Figure 6

From: Training calibration-based counterfactual explainers for deep learning models in medical image analysis

Figure 6

(a–d) Explanations generated using TraCE by introducing gender-specific attributes into the counterfactuals synthesized for changing the diagnosis state of a normal subject to be abnormal. In contrast to the age attribute, image manipulations associated with change in gender (female \(\rightarrow\) male) do not cause any apparent change to the likelihood of being assigned to the abnormal group. In each case, we highlight the changes \(\Delta _A({\text {x}}), \Delta _D({\text {x}}),\Delta _A({\text {x}})+\Delta _D({\text {x}})\) and show the likelihood \(P(state = {{abnormal}})\).

Back to article page