Scholarlink

Objectives <p>This systematic review evaluates the literature to determine the usefulness and rules of artificial intelligence (AI) in the diagnosis and prediction of maxillary canine impaction (MCI).</p> Materials and methods <p>The search was performed through four electronic databases (PubMed, Scopus, and Web of Science, with Google Scholar). The protocol was registered at the International Platform of Registered Systematic Review and Meta-analysis Protocols (INPLASY202550102). Original English-language research articles utilizing AI in MCI were included. The diagnostic accuracy studies were evaluated using the QUADAS-2 tool, while the predictive studies were evaluated using the PROBASST tool.</p> Results <p>The research strategy yielded seven articles deemed suitable for data extraction and quality assessment. Of these, five were diagnostic accuracy studies, and two were predictive studies. Overall, the studies demonstrated a low risk of bias. AI demonstrated high diagnostic performance, including improved detection accuracy (up to 98.3%), efficient segmentation (Dice similarity coefficient 0.99), and moderate-to-high predictive performance for three-dimensional (3D) reconstruction or root resorption classification. These findings highlight the potential of AI to enhance localization, diagnosis, and treatment planning of impacted maxillary canines.</p> Conclusions <p>Across the seven included studies, AI demonstrated high accuracy in localizing impacted maxillary canines and showed predictive capability for eruption outcomes. These findings suggest that AI can meaningfully support clinicians in diagnosis, treatment planning, and management of MCI. Further research is necessary to elucidate the clinical applicability of AI by standardizing methodologies, including larger samples for better validity and generalizability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Artificial intelligence performance in maxillary canine impaction: a systematic review

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月16日

查看全文
  • Hassan Ahmed Assiri,
  • Budur Khalid Alsaanah,
  • Bushra Alshahrani,
  • Saeed Alassiri,
  • Hussain Almubarak,
  • Abdullah Alqarni,
  • Mohammad Shahul Hameed,
  • Zayed Assiri,
  • Bander Assiri

摘要

Objectives

This systematic review evaluates the literature to determine the usefulness and rules of artificial intelligence (AI) in the diagnosis and prediction of maxillary canine impaction (MCI).

Materials and methods

The search was performed through four electronic databases (PubMed, Scopus, and Web of Science, with Google Scholar). The protocol was registered at the International Platform of Registered Systematic Review and Meta-analysis Protocols (INPLASY202550102). Original English-language research articles utilizing AI in MCI were included. The diagnostic accuracy studies were evaluated using the QUADAS-2 tool, while the predictive studies were evaluated using the PROBASST tool.

Results

The research strategy yielded seven articles deemed suitable for data extraction and quality assessment. Of these, five were diagnostic accuracy studies, and two were predictive studies. Overall, the studies demonstrated a low risk of bias. AI demonstrated high diagnostic performance, including improved detection accuracy (up to 98.3%), efficient segmentation (Dice similarity coefficient 0.99), and moderate-to-high predictive performance for three-dimensional (3D) reconstruction or root resorption classification. These findings highlight the potential of AI to enhance localization, diagnosis, and treatment planning of impacted maxillary canines.

Conclusions

Across the seven included studies, AI demonstrated high accuracy in localizing impacted maxillary canines and showed predictive capability for eruption outcomes. These findings suggest that AI can meaningfully support clinicians in diagnosis, treatment planning, and management of MCI. Further research is necessary to elucidate the clinical applicability of AI by standardizing methodologies, including larger samples for better validity and generalizability.