Miniaturized coherent anti-stokes Raman scattering (Mini-CARS) microscopy for intraoperative peritoneal lavage cytology in gastric cancer

Peritoneal metastasis (PM) is a leading cause of mortality in gastric cancer. Peritoneal lavage cytology is often recommended for diagnosing PM; however, it is not suitable for intraoperative diagnosis due to the lengthy sample preparation time. In this study, we developed a rapid, quantitative molecular cytology platform based on miniaturized coherent anti-Stokes Raman scattering (mini-CARS) microscopy and a hybrid K-means cell clustering and principal component analysis algorithm (K-PCA). Utilizing the inherent advantages of CARS for label-free, high-speed, and high-resolution chemical imaging, we obtained images of 14,882 exfoliated cells from ascites samples of 16 gastric cancer patients, of which 5 are PM-positive and 11 are PM-negative. Then, the composition and morphology features were extracted and analyzed by K-PCA to classify the samples as PM-negative or PM-positive. Out of the 16 patient samples, 14 were correctly classified, achieving an accuracy of 87.5%, with a processing time of less than 25min per test. Compared to stimulated Raman molecular cytology (SRMC), our mini-CARS-based cytology platform offers comparable diagnostic accuracy, yet with a simpler system architecture and a reduced footprint. These outcomes highlight the potential of mini-CARS for the rapid and quantitative detection of PM from gastric cancer in the clinic.