Machine Vision and Machine Learning in Digital Pathology

Computational approaches to the quantitative evaluation of histological and cytological images have been with us in cell biology and pathology for a long time. In biology, the stereological approaches that rely on scene sampLing and the projection of quantitative information from 2D to 3D have been used for over 80 years. The introduction of computer science into biology is, however, relatively recent. Moreover, in the last few years, there had been a sharp acceleration of the computing power used in biological investigations. In cellular imaging, this has translated into rapid computer-assisted image acquisition and image analysis and the integration of high-resolution cellular imaging data into the investigations of complex biological processes. The principles of artificial intelLigence appLied to cellular imaging are beginning to make their way into the mainstream of biologist's and pathologist's daily work. These tools will enable, not replace, the thought processes and methods already extant in the communities of cell biology and pathology. This article reviews some of the principles and practice of machine vision and machine learning in digital pathology.