Table 6 Confusion matrix of the canine test dataset, which shows the classification performance of three pathologists based on ten nuclei selected by each pathologist, which should represent the overall nuclear size predicted by the individual pathologist. The labels on the left indicate the class labels as defined by the consensus of our experts based on diagnostic experience. The mean accuracy of our pathologist was 74.22%.

From: Histological classification of canine and feline lymphoma using a modular approach based on deep learning and advanced image processing