Abstract: Cervical cancer is a common female malignant tumor. It has been increasing and rejuvenating in recent years. Early screening of cervical cancer is an effective control method to block cancer. In this study, a diffuse reflectance spectrum detection and analysis system based on LabWindows development software and MariaDB database was developed, which can acquire and save the spectral data to the database. The method of a neural network model based on spectral database was built to distinguish the cervical tissue and the normal tissue. The nude mouse tumor model test and human volunteer test were performed respectively, which verified that the system can distinguish between normal tissue and tumor tissue, and can be applied to the screening of cervical precancerous lesions.