Label-Free Virtual Peritoneal Lavage Cytology via Deep-Learning-Assisted Single-Color Stimulated Raman Scattering Microscopy

Abstract

Clinical guidelines for gastric cancer treatment recommend intraoperative peritoneal lavage cytology to detect free cancer cells. Patients with positive cytology require neoadjuvant chemotherapy instead of instant resection, and conversion to negative cytology results in improved survival. However, pathologists’ or artificial intelligence's accuracy of cytological diagnosis is disturbed by manually produced, unstandardized slides. In addition, the elaborate infrastructure makes cytology accessible to a limited number of medical institutes. This work develops CellGAN, a deep learning method that enables label-free virtual peritoneal lavage cytology by producing virtual hematoxylin–eosin-stained images with single-color stimulated Raman scattering microscopy. A structural similarity loss is introduced to overcome the challenge of unsupervised virtual pathology techniques that cannot accurately present cellular structures. This method achieves a structural similarity of 0.820 ± 0.041 and a nucleus area consistency of 0.698 ± 0.102, indicating the staining fidelity outperforms the state-of-the-art method. Diagnosis using virtually stained cells reaches 93.8% accuracy and substantial consistency with conventional staining. Single-cell detection and classification on virtual slides achieve a mean average precision of 0.924 and an area under the receiver operating characteristic curve of 0.906, respectively. Collectively, this method achieves standardized and accurate virtual peritoneal lavage cytology and holds great potential for clinical translation.