AN ANALYSIS ON IMAGE PROCESSING INFLUENCING THE RESULTS OF MULTI-TUMOR MARKERS PROTEIN CHIP DETECTION SYSTEM

摘要

Objective: To improve the accuracy of multi-tumor marker protein chip and to provide more reliable evidence for the early diagnosis of cancer.Methods: Two groups of normal sample images and tumor sample images are selected randomly from the photos taken by the multi-tumor marker protein chip detection system.Each group includes 10 samples.1.We adjusted the brightness of each image and chose the appropriate display mode of brightness.Using the function of auto measurement in the system,the results of 12 tumor markers were detected directly from each image.2.After choosing the appropriate brightness of each image,the points of 12 tumor markers in each image were analyzed,segmented.Then detect the results.The results of two groups were compared.Results: In the group of normal samples images,there is no statistically significant difference in the results of 12 tumor markers before or after image analysis,segmentation (P> 0.05);while in the group of tumor samples images,the results of 12 tumor markers that were analyzed and segmented are significantly different with that were not processed (P<0.05).Conclusions: In the detection of multi-tumor marker protein chip,image analysis and segmentation have no influence on the measurement of the image of normal samples.But there are more effects on the measurement of the image of tumor samples.Different display modes of brightness and different segmentation methods always lead to different results,even the opposite results.Therefore,before the final measurement in the detection of multi-tumor marker protein chip,it is necessary to choose the proper display modes of brightness and image analysis,image processing methods.