Figure 5
From: Weakly-supervised learning for lung carcinoma classification using deep learning
Examples of two carcinoma false positive prediction outputs using the weakly-supervised method. (a) Is a case of chronic interstitial pneumonia: there are prominent appearances of macrophages in the alveolar space and fibrosis of the alveolar septum as a background of mild to moderate chronic active inflammation. Infiltrating inflammatory cells include lymphocytes, monocytes, and neutrophils. (b) Is a case of emphysematous bulla: there is a cystic emphysematous space in the lung beneath the pleura. Chronic inflammation, congestion and hemorrhage are seen. The inflammatory tissues in (a,b) is the main cause of false positive prediction due to the resemblance to cancer stroma.
