Skip to main content
Fig. 4 | Insights into Imaging

Fig. 4

From: A deep learning model using chest X-ray for identifying TB and NTM-LD patients: a cross-sectional study

Fig. 4

Chest radiography and the class activation heatmaps for tuberculosis (TB), nontuberculous mycobacterial lung disease (NTM-LD), and Imitators. The colours on the heatmap represent the diagnostic weights of determining the class in interest. The hotter the colours are (red and yellow), the more important the areas are to the final decision in the deep neural network (DNN). a demonstrates patchy and poorly defined consolidation with cavities at the left upper and lower lobes that are highly suggestive for pulmonary TB. In b, the chest radiography reveals bilateral bronchiectasis with nodular infiltrations, which termed nodular bronchiectasis is a typical presentation of NTM-LD. c shows bronchiectasis at bilateral lung fields

Back to article page