Comparison between single and serial computed tomography images in classification of acute appendicitis, acute...

Area of Science:

• Radiology and Medical Imaging
• Artificial Intelligence in Medicine
• Computational Pathology

Background:

• Accurate and timely diagnosis of acute abdominal conditions like appendicitis and diverticulitis is crucial for patient outcomes.
• Computed tomography (CT) is a primary imaging modality for evaluating these conditions, but interpretation can be challenging.
• Automated diagnostic tools can potentially improve efficiency and accuracy in clinical practice.

Purpose of the Study:

• To develop and evaluate a convolutional neural network (CNN) utilizing the EfficientNet algorithm for automated classification of acute appendicitis, acute diverticulitis, and normal appendix.
• To compare the diagnostic performance of single image versus RGB serial image methods in this classification task.

Main Methods:

• Retrospective enrollment of 715 patients undergoing contrast-enhanced abdominopelvic CT.
• Development of a CNN model based on the EfficientNet algorithm.
• Training and validation using 4,078 CT images, with data augmentation for unbalanced datasets.
• Comparison of classification performance using single image and RGB serial image methods.

Main Results:

• The RGB serial image method showed higher sensitivity, accuracy, and specificity for classifying normal appendix and acute diverticulitis compared to the single image method.
• Mean areas under the receiver operating characteristic curve (AUCs) were significantly higher for all three conditions (acute appendicitis, acute diverticulitis, normal appendix) using the RGB serial image method.
• The developed model accurately distinguished between acute appendicitis, acute diverticulitis, and normal appendix on CT images, especially with the RGB serial image method.

Conclusions:

• An EfficientNet-based CNN model can accurately classify acute appendicitis, acute diverticulitis, and normal appendix from CT images.
• The RGB serial image method offers improved diagnostic performance over the single image method for these conditions.
• This AI approach holds promise for enhancing diagnostic accuracy and efficiency in abdominal imaging.