Diabetes is a general disease often infected in humans. Many ways to detect diabetes, one of them is checking blood pressure, but this way is not effective, because it takes blood first and take a lot of time. Iridology is one way analysis health based on the iris. Therefore we need a tool used to identify pancreatic damage as an indication of diabetes through iridology. Load image is the first step to identify pancreatic organs based on the iris. The eye image that we used as the input system comes from the eye clinic database. The next step is adaptive median filtering used in the process preprocessing to reduce the noise on the image. After that the next step is segmentation process using hough circle transform method. The results of segmentation will be normalized and take the Region of interest. ROI will be done feature extraction by using GLCM (Gray Level Co-Occurrence Matrix). To know the condition of pancreas organ using backpropagation method.