Table 3 Patch-level performance of the deep learning approach.

From: Quantitative analysis of prion disease using an AI-powered digital pathology framework

Method

Subset

DSC (%)

ACC (%)

UPerNet-Twins

Train

73.7 ± 9.1

93.6 ± 8.5

Val

70.2 ± 7.5

93.7 ± 6.8

  1. The performance of the network was assessed in terms of the Dice coefficient (DSC) and accuracy (ACC) for both the training and validation sets.
Back to article page