Truace
Subscribe

diagnostic accuracy of imaging AI for detecting avascular necrosis of the femoral head

Source article: Applications and performance of imaging artificial intelligence in detecting and staging avascular necrosis of the femoral head: a systematic review and meta-analysis

Objective To evaluate the diagnostic accuracy of imaging artificial intelligence (AI) models for detecting avascular necrosis of the femoral head (AVNFH) and to examine whether diagnostic performance differs according to imaging modality, disease-stage focus, and validation strategy. Materials and methods In accordance with PRISMA-DTA, we conducted a systematic search in four main databases. Two independent reviewers evaluated the studies, extracted relevant data, and assessed the quality following the QUADAS-2…

By the Truvace desk1 min read1 sourcePeer-reviewedTRV-2026-1280· cite & verify
Applications and performance of imaging artificial intelligence in detecting and staging avascular necrosis of the femoral head: a systematic review and meta-analysis
Experimental data for development of finite element models : head/thoraco-abdomen/pelvis by Nusholtz, Guy S Kaiker, Patricia S United States. National Highway Traffic Safety Administration University of Michigan. Transportation Research Institute. Biosciences Division. Public domain
Trace impact readingNegative state
P 74The P score combines the specificity and measured human impact of the grounded problem claim with the strength of this Trace’s cited sources.

Both sides are scored from claims and sources, not community votes.

G 69The G score combines the specificity and measured human impact of the grounded gain claim with the strength of this Trace’s cited sources.

In brief

A systematic review and meta-analysis of imaging AI for avascular necrosis of the femoral head pooled nine estimates and reported sensitivity of 0.888, specificity of 0.937, and SROC AUC of 0.967, with MRI-based models showing higher sensitivity than radiography-based models.

The findings suggest promising research-stage performance, but the authors emphasize that risk of bias, heterogeneity, and reliance on only five external test-set estimates limit confidence for clinical use, leaving prospective multicenter validation and demonstrated clinical benefit as unresolved steps.

Main points

  1. Systematic review and meta-analysis followed PRISMA-DTA with QUADAS-2 quality assessment across four databases.
  2. Primary analysis pooled nine estimates for binary AVNFH detection using bivariate random-effects model.
  3. MRI-based estimates showed higher sensitivity than radiography-based estimates in study-level comparison (p = 0.027).
  4. Only five external test-set estimates contributed to primary analysis, limiting precision of subgroup comparisons.

The gain

Imaging AI models achieved high pooled diagnostic accuracy for detecting avascular necrosis of the femoral head in research datasets.

The problem

Risk of bias, heterogeneity, and limited external testing restrict confidence that reported accuracy will translate to routine clinical use.

The rundown

The review searched four databases under PRISMA-DTA, with two independent reviewers extracting data and assessing quality via QUADAS-2. Nine estimates were pooled for binary detection, yielding sensitivity 0.888 and specificity 0.937 with SROC AUC 0.967.

Subgroup analyses found MRI-based estimates had higher sensitivity than radiography (p = 0.027), while no significant differences were found by disease-stage focus or validation strategy. Authors noted prospective, multicenter validation and evaluation of clinical benefit are needed before routine implementation.

What this doesn’t fix

Confidence in clinical applicability is limited by risk of bias, between-study heterogeneity, and sparse external testing, with only five external test-set estimates in primary analysis.

Sources

  1. Peer-reviewedSkeletal Radiology2026-10-04

The debate