Diagnosing gangrenous cholecystitis on computed tomography using deep learning: A preliminary study.

Area of Science:

• Radiology
• Artificial Intelligence
• Gastroenterology

Background:

• Gangrenous cholecystitis is a severe complication of acute cholecystitis requiring prompt surgical intervention.
• Accurate and timely diagnosis of gangrenous cholecystitis is crucial for patient outcomes.
• Computed tomography (CT) is a key imaging modality for diagnosing cholecystitis.

Purpose of the Study:

• To compare the diagnostic performance of a deep learning (DL) model against experienced physicians in identifying gangrenous cholecystitis using CT images.
• To evaluate the feasibility of DL as a diagnostic aid for acute cholecystitis necessitating emergency surgery.

Main Methods:

• A retrospective study analyzed CT images from 25 patients with gangrenous cholecystitis and 129 with noncomplicated acute cholecystitis.
• A convolutional neural network (CNN) model was trained and tested on CT images.
• The DL model's diagnostic performance (sensitivity, specificity, accuracy, AUC) was compared to that of three independent, blinded physicians.

Main Results:

• The CNN achieved higher diagnostic performance than the physicians, with an accuracy of 0.89 (95% CI, 0.81-0.95) and an AUC of 0.84 (95% CI, 0.68-1.00).
• Physicians' performance metrics included accuracy of 0.65 (95% CI, 0.57-0.72) and AUC of 0.63 (95% CI, 0.44-0.82).
• The difference in AUC between the CNN and physicians was statistically significant (P=0.048).

Conclusions:

• Deep learning models significantly outperform experienced physicians in diagnosing gangrenous cholecystitis based on CT imaging.
• DL shows potential as an assistive tool for identifying patients with acute cholecystitis who require emergency surgery.