AI and histology
The last decade has seen a dramatic increase in the development and application of artificial intelligence tools for use in pathology services. Many scientists are confident that this trend will continue and change the field of pathology in the coming years.
Artificial intelligence is a term that encompasses both machine learning and deep learning and is based on the idea that a machine can provide an optimal solution to a problem by passing information through layers of neural networks. Basic types of AI are trained through a labeled dataset, in the case of histopathology we are talking about sick and healthy samples, from which the machine will self-select some distinctive markers. In a classical histopathology laboratory, making a single diagnosis requires the labor of a trained technician who has spent more than 7 years learning how to correctly identify diseased tissue under the microscope. But artificial intelligence can assess tissue architecture and cell morphology instantaneously and can process an unlimited number of samples.
Artificial intelligence can already be used to detect and count cells, for example, in the mitotic process to identify cancer cells. It will also help in detecting segmentation and classifying tissue into healthy and diseased. A key advantage of this artificial technology is that it can be used to visualize an entire slide, where AI can automatically identify patterns across the entire slide.
Current AI models can detect breast cancer metastases to lymph nodes. And already during initial studies, it was found that AI easily achieves success rates comparable to those of a human professional and even more. It can detect tumors as small as 0.2 mm and neoplasms with a minimum number of cancer cells (less than 200 cells).
AI can also be useful for repetitive tasks, such as image analysis. After all, such work is quite monotonous and tedious for humans, so the risk of error due to human error is very high.
Despite the great potential of AI models in histopathology, artificial intelligence is unlikely to replace histopathologists. But it can be an excellent assistant.