Differential Diagnostics Focal Pancreatics Masses Based on the Analysis of the Echographic Texture of Endosonograms Using Fuzzy Mathematical Models

Abstract

Relevance: Differential diagnosis of focal formations of the pancreas is particularly difficult due to the similarity of their echosemiotics. One of the ways to objectify and improve the accuracy of ultrasound data is to use artificial intelligence methods to interpret images.Purpose: Improving the quality of diagnosis of focal pancreatic pathology according to endoscopic ultrasonography based on the analysis of the echographic texture using fuzzy mathematical models.Material and methods: In the Kursk Regional Clinical Hospital, endoscopic ultrasonography was performed in 272 patients for pancreaticobiliary diseases. The endoscopic video system of the company Olympus EVIS EXERA II with the ultrasonic processor EU-ME1 was used. Solid tumors of the pancreas were detected in 109 (40.1 %) patients, focal pancreatic masses were diagnosed in 40 (14.7 %) patients. Based on the research results, the main types of reference endosonograms corresponding to differentiable pathology were identified. On the basis of endosonograms, using hybrid fuzzy mathematical decision rul, a fuzzy hybrid model of differential diagnosis of chronic focal pancreatitis and ductal adenocarcinoma of the pancreas was obtained.Results: The selected source of information and the method of synthesis of hybrid fuzzy decision rules made it possible to obtain a fuzzy hybrid model of differential diagnosis of chronic focal pancreatitis and ductal adenocarcinoma of the pancreas. Based on the results of mathematical modeling and statistical tests on representative control samples, it was shown that the resulting model of differential diagnosis, using reference endosonograms reflecting the echographic texture of focal pancreatic masses, provides confidence in the desired diagnosis at the level of 0.6. The additional information used in the analysis of endosonograms increases the diagnostic confidence to a value of 0.9.Conclusions: The resulting model of differential diagnosis of chronic focal pancreatitis and ductal adenocarcinoma of the pancreas provides confidence in decision-making no worse than 0.9, which is an acceptable quality indicator in clinical practice in conditions of insufficient statistics with poorly formalized data structure.