Currently, the cancer was examined by diagnosing the pathological changes of tumor. If the examination of cancer can diagnose the tumor before the cell occur the pathological changes, the cure rate of cancer will increase. This research develops a human-machine interface for hyper-spectral microscope. The hyper-spectral microscope can scan the specific area of cell and records the data of spectrum and intensity. These data is helpful to diagnose tumor. This study finds the hyper-spectral imaging have two higher intensity points at 550nm and 700nm, and one lower point at 640nm between the two higher points. For analyzing the hyper-spectral imaging, the intensity at the 550nm peak divided by the intensity at 700nm peak. Finally, we determine the accuracy of detection by Gaussian distribution. The accuracy of detecting normal cells achieves 89%, and the accuracy of cancer cells achieves 81%.