Does image processing improve diagnostic accuracy and differentiation between odontogenic keratocysts and simple bone cysts? A comparative study

Objective: The aim of this study was to evaluate the diagnostic accuracy of image processing in differentiating odontogenic keratocysts (OKCs) and simple bone cysts (SBCs) using panoramic radiographs (OPGs) and cone-beam computed tomography (CBCT) scans. Material and Methods: Five images of OKCs and five of SBCs were processed using enhancement features available in GIMP software (including edge enhancement and smoothing filters) and compiled into image panels for observer analysis. A total of 20 OPG and 20 CBCT images were used—10 processed and 10 unprocessed for each modality. Eight observers with prior knowledge in Oral and Maxillofacial Radiology assessed the images in two sessions spaced two months apart. The observers were blinded to the image processing status and the imaging modality (OPG or CBCT). In the first session, six diagnostic options were provided; in the second session, the options were narrowed to two. Results: Multivariate analysis revealed that image processing was not a significant predictor of diagnostic accuracy (p=0.642 and p=0.678). However, both the type of lesion (p<0.001) and the imaging modality (p=0.004) significantly influenced correct diagnosis. Notably, OPG images were more likely to lead to correct answers compared to CBCT (odds ratio=3.033; 95% confidence interval=1.418-6.487; p=0.004). Conclusion: while image processing did not improve diagnostic accuracy, the imaging modality and the type of lesion had a significant impact on observer performance.