Endoscopic OCT Image Texture Analysis for Identifying Early-Stage Pancreatic Cancers
Presentations
PO-GePV-M-326 (Sunday, 7/10/2022) [Eastern Time (GMT-4)]
ePoster Forums
Purpose: We proposed a new index called Roughness to quantify the texture contexts of the endoscopic OCT images of the pancreatic ducts, which can be used to quantitatively identify early-stage pancreatic cancers in pancreatic ducts.
Methods: We proposed and defined Roughness (x) to describe the level of Roughness for a region of interest (ROI) x on the endoscopic OCT images, which was defined by the following:Roughness (x)=∑_(n=1)^∞ (〖|g〗_n (x)-g_mean (x)|)/g_mean (x)g_n (x): the grey value of pixel #n for the ROI x; g_mean (x): the mean grey value for the entire ROI x. Based on the above formula we calculated Roughness for ROIs on the endoscopic OCT images to perform the texture analysis using MATLAB and ImageJ software. The endoscopic OCT image samples were acquired from our IRB approved clinical trials on pancreatic cancer patients using the endoscopic OCT imaging system that we built for early detection pancreatic cancers. We calculated and compared the results of Roughness for the endoscopic OCT images of the cancerous and healthy pancreatic ducts.
Results: The difference of Roughness between the cancerous pancreatic duct and the normal one was significant, which indicated that the defined quantity Roughness could be used as an index to distinguish cancerous pancreatic ducts from normal healthy pancreatic ducts.
Conclusion: The proposed index Roughness can be applied to quantitatively identify early-stage pancreatic cancers expressing in the endoscopic OCT images of pancreatic ducts.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email