Radiomic texture mapping improves visualization of simulated condylar bone lesions in cone beam computed tomography.

Area of Science:

• Radiology and Imaging
• Dental Diagnostics
• Biomedical Engineering

Background:

• Cone-beam computed tomography (CBCT) is crucial for evaluating mandibular condyle integrity.
• Detecting small condylar defects remains challenging, impacting temporomandibular joint (TMJ) health.
• Texture feature maps (TFMs) offer a novel approach to enhance image analysis.

Purpose of the Study:

• To assess if TFMs derived from CBCT images improve the detection of small simulated condylar defects.
• To evaluate the influence of different voxel sizes on defect detection using TFMs.
• To determine the diagnostic performance of various texture features.

Main Methods:

• Simulated spherical defects (0.9-1.2 mm) were created in eight dry mandibles.
• CBCT scans were acquired at 0.20 mm and 0.40 mm voxel sizes.
• Gray-level co-occurrence matrix (GLCM)-based TFMs (contrast, entropy) were generated and analyzed by radiologists.

Main Results:

• Contrast-based TFMs significantly improved defect detection (Kappa: 0.78-1.00) compared to unprocessed images.
• Entropy maps showed lower diagnostic performance.
• Voxel size and defect dimension did not significantly affect TFM performance.
• TFMs enhanced interobserver agreement.

Conclusions:

• Contrast-based TFMs enhance the visualization of subtle condylar bone defects in CBCT.
• This post-processing technique aids in early TMJ diagnosis, independent of voxel size.
• TFMs represent a promising adjunct for improving diagnostic confidence in TMJ imaging.