Figure 2 | Scientific Reports

Figure 2

From: Using a convolutional neural network for classification of squamous and non-squamous non-small cell lung cancer based on diagnostic histopathology HES images

Figure 2

Prediction procedure for a squamous tumour biopsy section. (a) Representation of original slide. (b,c) Steps for prediction analysis based on (b) Whole Slide Image (WSI) or (c) virtual TMA strategy. Heatmaps show for each tile probability of being predicted as squamous tumour (red) or non-squamous tumour (green); the highest probability was kept for coloration; grey was used for tiles non selected by the pathologist or removed by denoising step (tile containing more than 2/3 of white background).

Back to article page