| Summary: | 碩士 === 國立臺北科技大學 === 自動化科技研究所 === 103 === Recently, requirements of image processing techniques used for helping biomedical diagnosis have become popular. In traditional systems, they’re time consuming because a lot of observations and reorganizations of human, and after analyzing multiple case, human may less focused lead to results fall. About this, in this thesis, an image analysis system for cancer cell is proposed. This system will help human easier to observation cancer cell, and save resources. The advantages of this are summarized as follows: 1) the proposed system used image processing and machine learning to simulate human analysis cells sizes, and used watershed to calculate cells correlation coefficient. The results of system (normal/cancer=19/73, Artificial recognition AUC=0.78, System AUC=0.65, System and Artificial recognition AUC=0.74), 2) the system process should only take about 20-30 minutes, and manual follow-up screening just at the average is less than 400 cells (5 minutes), proves system effectively and accelerate human count processes.
|
|---|
