Manas Dave discusses a topic featured in our sister journal Evidence-Based Dentistry.

‘Applications of artificial intelligence in the analysis of dental panoramic radiographs: an overview of systematic reviews' was published in the journal Dentomaxillofacial Radiology (DMFR) in 2023.1 An evidence-based summary of this article, ‘Artificial intelligence and dental panoramic radiographs: where are we now?' was published in Evidence-Based Dentistry this year.2

figure 1

©belterz/E+/Getty Images Plus

Background

Artificial intelligence (AI) encompasses a broad range of technologies that enable machines to perform tasks that typically require human intelligence, such as learning, reasoning and problem-solving.2 The subfields include machine learning where systems can improve their performance based on data, and deep learning (which is a subset of machine learning) involving neural networks with many layers.3 Common public applications of AI include nature language processing tools like OpenAI's ChatGPT4 and Google's Gemini,5 which facilitate human-computer interactions and information retrieval. In radiology, AI is being increasingly utilised to enhance diagnostic accuracy and efficiency. Advanced algorithms are being used in the detection and analysis of medical images (such as MRI and CT scans), contributing to improved diagnostic efficacy and patient outcomes.6

This review shows how AI is integrating into dentistry and possible tools that may become available in the future.

There have been numerous studies on the use of AI in analysing dental panoramic radiography (DPR) images. Therefore, the aim of this systematic review was to establish the current state of knowledge on the suitability of AI in DPR analysis.

Methods

An electronic database search of Bielefeld Academic Search Engine, Google Scholar, Association for Computing Machinery: Guide to Computing Literature and PubMed.

Only systematic reviews that included panoramic radiography with diagnoses made by AI algorithms compared to human diagnoses were included.

Results

  • Twelve systematic reviews were included that were published between 2019-2023

  • The risk of bias was high in one systematic review, and it was subsequently excluded from further analysis

  • Teeth identification and numbering: four systematic reviews showed the sensitivity ranged from 96-98% and specificity 97-94%

  • Detection of periapical lesions: two systematic reviews showed the sensitivity ranged from 48-65% and specificity ranged from 87-99.95%

  • Detection for periodontal bone loss: six articles from two systematic reviews showed the sensitivity ranged from 76-84% and specificity from 81-93.75%

  • Detection of osteoporosis: 13 reports from two systematic reviews showed the sensitivity ranged from 76.8-99.1% and specificity ranged from 43.8-98.4%

  • Detection of maxillary sinusitis: two systematic reviews showed the sensitivity ranged from 77.6%-86.7% and specificity ranged from 69.4-88.3%

  • Detection of dental caries: two systematic reviews were included, neither reported sensitivity and one reported specificity at 86%

  • Other metrics included the diagnosis of temporomandibular joint osteoarthrosis with a sensitivity of 39-94% and specificity of 77-91%.

Conclusions

The authors stated:

‘…The latest analyzed AI models achieve high accuracy in detecting caries - 91.5%, osteoporosis - 89.29%, maxillary sinusitis - 87.5%, periodontal bone loss - 93.09%, and teeth identification and numbering - 93.67%. The detection of periapical lesions is also characterized by high sensitivity (99.95%) and specificity (92%). The above results indicate that AI applications can significantly support dentists. However, due to the small number of heterogeneous source studies synthesized in systematic reviews, the results of this overview should be interpreted with caution. As systematic reviews in AI become outdated quickly, their regular updating is recommended.'

Commentary

This systematic review comprehensively extracted accuracy, sensitivity, specificity and precision parameters, where reported, for the AI detection of lesions. The authors acknowledge how quickly such reviews become out of date and the need to use the latest information available as AI improves over time. Additionally, individual studies would have used different AI tools and datasets to train and test their models so results will inherently be heterogeneous until a common AI tool becomes commonly used. This review shows how AI is integrating into dentistry and possible tools that may become available in the future.