Figure 6
From: Improve the performance of CT-based pneumonia classification via source data reweighting
Two failure cases of our method. Each row contains some slices of a target CT scan. The first CT scan is from the CP class, but is incorrectly predicted as being from the NCP class. The second CT scan is from the NCP class, but is incorrectly predicted as being from the CP class. The cause of failure is that the two types of pneumonia have similar findings in CTs, which makes it challenging to distinguish them.
