Scoping Review of AI-Based Automatic Landmark Identification in 3D CT/CBCT

A scoping review by You Wu, published on figshare in May 2026, systematically examines 37 studies on automatic landmark identification in three-dimensional computed tomography and cone-beam computed tomography. The review analyzes the accuracy, efficiency, and clinical applicability of AI-based methods, focusing on studies published until January 2026. It includes studies that validated automatic methods against manual expert annotations.

Use Cases

• Benchmarking new automatic landmarking algorithms against the accuracy trends and clinical standards identified in the review.
• Designing clinical validation studies based on the reported sample size limitations and landmark types (hard vs. soft tissue) from the included studies.
• Informing the development of standardized evaluation metrics for 3D landmarking systems, addressing the lack of standardization noted in the conclusions.

Strengths

• Provides a structured analysis of 37 specific studies, offering a clear scope of existing research.
• Quantifies key findings, such as 75.7% of studies having fewer than 200 cases and 56.8% meeting a clinical correctness standard (MRE < 2mm).
• Explicitly details the search methodology and inclusion criteria, enhancing reproducibility.

Limitations

• The dataset is a 36.8 KB document (DOCX) containing a review article, not a primary data collection; its utility is as a meta-analysis.
• The review notes that the included studies had limited samples and lacked standardized evaluation, which may affect the generalizability of its conclusions.
• Column-level documentation for any underlying data from the reviewed studies is absent; the file contains only the review's text and analysis.

Provenance

Source

figshare

Collection Method

Scoping review of literature from PubMed, Web of Science, and the Cochrane Library up to January 2026.

Time Range

Studies published until January 2026.

Freshness

Last updated 2026-05-29 06:07:36.