Radiomics-Based Differential Diagnosis of Radicular Cysts and Apical Granulomas on CBCT Images Using RadC-CNN...

Area of Science:

• Dentistry
• Medical Imaging
• Artificial Intelligence

Background:

• Differentiating radicular cysts (RC) from periapical granulomas (PG) is crucial for appropriate treatment.
• Histopathological diagnosis is the gold standard but is invasive.
• Cone-beam computed tomography (CBCT) offers a non-invasive imaging modality.

Purpose of the Study:

• To evaluate radiomic features from CBCT for differentiating RC from PG.
• To compare the diagnostic performance of traditional machine learning (ML) algorithms with a deep learning (DL) model, Radiomics Cyst Convolutional Neural Network (RadC-CNN).

Main Methods:

• Retrospective analysis of CBCT images from 98 patients with histopathologically confirmed RC or PG.
• Semi-automatic segmentation and extraction of 48 radiomic features.
• Comparison of traditional ML models (Decision Tree, KNN, SVM) with the proposed RadC-CNN architecture.

Main Results:

• 18 out of 34 reliable radiomic features showed significant differences between RC and PG (p < 0.05).
• Shape, first-order, and texture features (GLCM, GLRLM, GLSZM, NGTDM) were extracted.
• RadC-CNN achieved 90% accuracy, 90% sensitivity, and 91.3% precision, outperforming traditional ML algorithms.

Conclusions:

• CBCT-based radiomic analysis is a promising non-invasive method for distinguishing RC from PG.
• The RadC-CNN deep learning model demonstrates superior diagnostic performance compared to traditional ML algorithms.