AI combined hyperspectral cell image recognition method and system

• Main Authors: XIE XINGRAN, PAN YUNDI, DENG YINGJIAO, RAN TAOJING, GAO LILI, ZHOU CHUNHUA, ZOU DUOWU, LI QINGLI, ZHANG JIA'NAN, WANG YAN
• Format: Patent
• Language: Chinese; English
• Published: 26.04.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention discloses an AI combined hyperspectral cell image recognition method, which comprises the following steps: S1, collecting hospital data, screening the data by a doctor, and selecting a pancreatic neuroendocrine tumor patient and a pancreatic benign disease patient; s2, constructing a medical hyperspectral microscopic imaging system to perform hyperspectral cell image acquisition; s3, carrying out ROI (Region of Interest) labeling, image segmentation and spectral analysis on the hyperspectral cell image; s4, inputting the multispectral information of the pancreatic neuroendocrine tumor cells and the pancreatic benign cells into the ResNET18 network for training and learning; and S5, verifying the effectiveness and generalization of the liquid-based cytology identification of the pancreatic neuroendocrine tumor and the pancreatic benign disease by combining the hyperspectrum with the ResNET18 network by using clinical data. By adopting the scheme, cell pathology doctors with different qualificatio