Histopathological diagnosis is the ultimate method of attaining the final diagnosis; however, the observation range is limited to the two-dimensional plane, and it requires thin slicing of the tissue, which limits diagnostic information. To seek solutions for these problems, we proposed a novel imaging-based histopathological examination. We used the multiphoton excitation microscopy (MPM) technique to establish a method for visualizing unfixed/unstained human breast tissues. Under near-infrared ray excitation, fresh human breast tissues emitted fluorescent signals with three major peaks, which enabled visualizing the breast tissue morphology without any fixation or dye staining. Our study using human breast tissue samples from 32 patients indicated that experienced pathologists can estimate normal or cancerous lesions using only these MPM images with a kappa coefficient of 1.0. Moreover, we developed an image classification algorithm with artificial intelligence that enabled us to automatically define cancer cells in small areas with a high sensitivity of ≥0.942. Taken together, label-free MPM imaging is a promising method for the real-time automatic diagnosis of breast cancer.
Keywords: artificial intelligence; breast cancer; multiphoton excitation microscopy; rapid diagnosis; surgical margin.
© 2022 The Authors. Cancer Science published by John Wiley & Sons Australia, Ltd on behalf of Japanese Cancer Association.