Application of artificial intelligence in toxicopathology

Abstract

Traditionally, pathologists examine tissue slides under a microscope to find pathological lesions, and have the burden of finding the lesions among so many histopathology slides. Furthermore, inconsistency of diagnoses results differ corresponding to training among researchers. Therefore, accumulated research experience has led to the use of novel tools for increasing accuracy and consistency of diagnoses. With rapid transition from analog to digital methods and new developments in digital pathology, it is possible to use whole slide imaging (WSI) by scanning glass slides. Artificial intelligence (AI), including machine learning and deep learning using WSI, is starting to be applied to automatically classify and count microscope images, and this method has been expanded to include the field of medical image analysis. This review aims to define current trends toward AI application in the biomedical area, especially in the field of toxicopathology, outline current future business trends, and discuss multiple issues of diagnosis, quantification, three-dimensional reconstruction, molecular pathological research, and the future direction of AI in toxicopathology. Big data systems including a large amount of welldefined toxicopathological information will be highly useful for accuracy and corrections of diagnoses. In addition, the need for critical peer review is profound in the continuing educational process. Taken together, it is highly promising that AI model based on big data in the toxicopathological field could classify, detect, and segment pathological lesions in numerous organs of experimental animals and could help explain various biological mechanisms. This promising approach will provide an accurate and fast analysis of tissue structure and biological pathways using AI algorithms and big data.