[17a-WL2_101-1~9] 3.2 Information photonics and image engineering

11:00 AM - 11:15 AM JST(2:00 AM - 2:15 AM UTC)Presentation by Applicant for JSAP Young Scientists Presentation Award

[17a-WL2_101-6] All-Optical Classification of Cell Transmitted-Light Data Using Diffractive Neural Networks: A Simulation Study

〇(M1)Norihide Sagami¹, Yueyun Weng², Cheng Lei², Ryosuke oketani¹, Kotaro Hiramatsu¹ (1.Kyushu Univ., 2.Wuhan Univ.)

Keywords:

optical neural network,bioimaging

Diffractive neural networks (DNNs) have attracted attention as an ultrafast signal processing technology that exploits optical diffraction and interference for computation. Although DNN-based classification has been demonstrated for simple images such as MNIST, DNN analysis of cellular images has not yet been validated. In this study, we performed a DNN-based cell classification simulation using amplitude–phase pairs of transmitted light acquired by large-scale quantitative phase imaging, and demonstrated a classification accuracy of 96.1% for lung cancer cells, breast cancer cells, and leukocytes (N = 4800).