Rapid diagnosis using sequential Raman measurements

Abstract

Pathological diagnosis using images obtained by Raman microscopy has the drawback of long acquisition times. To address this challenge, we employed a multi-armed bandit approach. Our proposed method dynamically decides the next point to irradiate, focusing measurements on areas necessary for diagnosis, thereby accelerating the diagnostic process. This approach allows for a significant reduction in the number of samples needed for measurements while theoretically guaranteeing the accuracy of sample identification, such as distinguishing between different cell phenotypes. In practical applications, we demonstrated the effectiveness of our method by showing that, compared to the conventional raster scanning method, the number of illuminations required to distinguish between human follicular thyroid cells and follicular thyroid carcinoma cells could be reduced to between 1/31683 and 1/3333 , thereby proving its efficacy.