Machine Learning-Based Radiological Features and Diagnostic Predictive Model of Xanthogranulomatous Cholecystitis

Area of Science:

• Radiology
• Oncology
• Gastroenterology

Background:

• Xanthogranulomatous cholecystitis (XGC) is a rare, benign gallbladder condition.
• XGC can be radiologically indistinguishable from gallbladder cancer (GBC).
• Accurate differentiation is crucial for appropriate treatment decisions.

Purpose of the Study:

• To analyze radiological characteristics of XGC and GBC.
• To develop a diagnostic prediction model for differential diagnosis.
• To improve clinical decision-making in distinguishing XGC from GBC.

Main Methods:

• Radiological features were analyzed using RandomForest and Logistic regression.
• Computed tomography (CT) and magnetic resonance imaging (MRI) models were established.
• Receiver operating characteristic (ROC) curve analysis was performed to validate the model's effectiveness.

Main Results:

• The CT/MRI model achieved a mean Area Under the Curve (AUC) of 0.897 with 90.6% accuracy.
• The diagnostic prediction model demonstrated a mean AUC of 0.888 and accuracy of 89.8%.
• These models showed high diagnostic efficiency in distinguishing XGC from GBC.

Conclusions:

• The developed diagnostic prediction model offers good preoperative diagnostic accuracy.
• This model can assist clinicians in differentiating XGC from GBC.
• Improved preoperative discrimination aids in optimal clinical management.