Analyzing pancreatic characteristics in diabetic patients: A texture-based CT investigation with volume assessment

Abstract

The current study aimed to analyze the texture feature of the pancreas in diabetic patients at King Faisal Medical Complex (KFMC) in Taif, KSA through Cutting-Edge Computed tomography (CT) image analysis and volume metrics. Material and method: 78 participants were involved, 39 patients with normal pancreas acted as a control group, while the remaining 39 were diagnosed with diabetes, out of which 20 had type 2 diabetes (T2D) and 19 had type 1 diabetes (T1D). These patients underwent CT scans using two specific CT scanners and were compared to normal subjects with the same age, gender and CT scanner. The entire pancreas was subjected to computerized texture analysis, which involved the extraction of 31 variable features. A multivariate logistic regression analysis was used to identify factors that could predict diabetes. A Receiver Operating Characteristic (ROC curve) was generated to determine the optimal cut-off values for statistically significant variables. Findings: Analysis of various pancreatic texture features revealed sphericity to be the most informative for diagnosing diabetes. Diabetic patients, in general, exhibited lower sphericity compared to healthy controls (p = 0.0246). For individuals with diabetes, the sphericity feature demonstrated a sensitivity of 74.36%, a specificity of 69.23%, and an Area Under the Curve (AUC) of 0.7538. Interestingly, subgroup analysis did not show any significant differences among patients with type 2 diabetes. ConclusionThis study demonstrates the potential of using computerized 3D CT texture analysis of the pancreas to assess diabetes mellitus through the measurement of quantitative parameters.