A Novel Method of Measuring Pathological Conditions: Using Machine Learning in Tissue Stretch Response Patterns

Abstract

Pathological diagnosis is an important diagnostic technique to determine a medical treatment policy. In the standard method, diagnosis of tissue slices has been investigated based on the visual inspection by microscope. However, it is difficult to evaluate a state of disease in a quantitative manner using the current methodology. Here, we propose a novel pathological diagnosis method focusing on the physical characteristics of tissue sections depending on the difference of disease state. We have found that the cracking pattern caused by applying tension to tissue sections depends on the pathological condition. By adapting such cracking pattern as a quantitative index for pathological diagnosis, it becomes possible to perform pathological diagnosis in a reliable and quantitative manner.