Comparative Evaluation of Imaging Modalities for Complex Oral and Maxillofacial Pathologies: A Systematic Review and Semi-Quantitative Synthesis

Abstract

It has been established that a significant section of the population is affected by oral and maxillofacial pathologies. This systematic review and semi-quantitative synthesis, therefore, aim to identify different imaging modalities that are used to diagnose these pathologies and their relative effectiveness. This study also evaluates the impact of technologies such as artificial intelligence (AI) on the efficacy of the imaging modalities. The method applied in the study was a systematic review and semi-quantitative synthesis, and different quantitative studies were selected and their findings analyzed. The selection process was non-biased in nature since objective inclusion and exclusion criteria, such as the year the study was published and the rating of the methodology, were used in the selection process. The study found that certain imaging modalities, such as ultrasonography and scintigraphy, have more accuracy and sensitivity compared to conventional radiology. Ultrasonography was also found to have more advantages due to its lack of use of ionization, thus reducing the impact associated with ionization during the imaging process. AI was also found to have a positive impact on different imaging modalities, as it increased the sensitivity and the accuracy of the modalities. One of the recommendations is continued research on the impact of AI tools; technologies are continually being improved, and therefore, their effectiveness is likely to increase over time. The use of ultrasonography should be advocated for, and the necessary resources should be made available, as ultrasonography not only has a higher rate of success but also reduces the risks associated with conventional imaging processes.

Keywords: artificial intelligence; diagnostic accuracy; imaging modalities; oral and maxillofacial pathologies; ultrasonography.

Figure 1. PRISMA flow diagram illustrating the study selection process.

PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Figure 2. Sensitivity, specificity, and accuracy of different imaging modalities for the diagnosis and management of complex oral and maxillofacial pathologies, as reported in included studies.

CLE, confocal laser endomicroscopy; AI, artificial intelligence; CBCT, cone-beam computed tomography.

Figure 3. Sensitivity of different imaging modalities for the diagnosis and management of complex oral and maxillofacial pathologies, as reported in included studies.

CLE, confocal laser endomicroscopy; AI, artificial intelligence; CBCT, cone-beam computed tomography.

Figure 4. Specificity comparison across different imaging modalities.

CLE, confocal laser endomicroscopy.

Figure 5. Accuracy comparison across different imaging modalities.

CLE, confocal laser endomicroscopy.

Figure 6. Comprehensive diagnostic performance comparison.

Ultrasonography consistently demonstrated high sensitivity and accuracy across studies, with sensitivity ranging from 88.2% to 100% and accuracy up to 98.4%. It was particularly effective for soft tissue and bone pathology detection. Conventional radiography showed good sensitivity (94.1%) but lower specificity (50%) compared to ultrasonography and was less accurate for certain pathologies. Scintigraphy had the highest sensitivity (100%) but very low specificity (14.3%), indicating it is useful for initial detection but not for definitive diagnosis. AI-enhanced modalities (CBCT and CLE) showed improved accuracy and specificity compared to traditional methods, with AI-based CBCT improving diagnostic accuracy by 11% and deep learning increasing CLE accuracy and specificity for oral cancer detection. Other modalities, such as tissue segmentation by ultrasound (Poul et al. [15]) and tissue differentiation (Di Stasio et al. [13]), provided valuable insights but did not report diagnostic accuracy for disease detection. CLE, confocal laser endomicroscopy; CBCT, cone-beam computed tomography.