Figure 2 | Scientific Reports

Figure 2

From: Deep learning systems detect dysplasia with human-like accuracy using histopathology and probe-based confocal laser endomicroscopy

Figure 2

Confusion matrices for the two pCLE models (Attn and MultiAttn) designed to detect dysplasia given its relative rarity. Attn (top left, top right) and the MultiAttn (bottom left, bottom right) models for pCLE classification. The matrices on the left represent normalized data in the form of percentages (e.g. the MultiAttn model correctly classified 71% of true dysplasia sequences). Darker colors indicate higher percentages in a given square. The matrices on the right represent the same data in the form of raw numbers of sequences in each class. Darker colors indicate higher numbers of sequences. The top-left to bottom-right diagonal represents accordance between the true classification and the classification predicted by the model. Created using matplotlib 3.3.2 (https://matplotlib.org/).

Back to article page