Comparative diagnostic accuracy of radiomics, deep learning, and hybrid AI for invasiveness stratification…

AI / informaticsFrontiers in oncology· 2026

Comparative diagnostic accuracy of radiomics, deep learning, and hybrid AI for invasiveness stratification of pulmonary ground-glass nodules: a systematic review and meta-analysis.

Ning Dong, Zhongwei Li, Zhenjie Cong … Peng Liang

Medical Imaging AIAI / informaticsDiagnosisQ2IF 3.1

Hybrid AI models most accurately distinguished invasive from non-invasive ground-glass lung nodules

Comparative diagnostic accuracy of radiomics, deep learning, and hybrid AI for invasiveness stratification of pulmonary ground-glass nodules: a systematic review and meta-analysis.

Ning Dong … Peng Liang

Frontiers in oncology · 2026

Background

Accurate preoperative stratification of ground-glass nodules (ground-glass nodules) into invasive versus non-invasive adenocarcinomas is essential for optimizing surgical management and preventing overtreatment.

Methods

A comprehensive search of PubMed, Embase, Cochrane Library, and Web of Science was conducted through 24 September 2025 (PROSPERO CRD42025115051).

n = 9,768 nodules

Results

hybrid models best separated invasive from non-invasive nodules, closer to 1.0 is better

Hybrid

0.94AUC

Deep learning

0.92AUC

Radiomics

0.89AUC

More results

At a 20% pre-test probability, a positive hybrid AI result increased post-test probability of invasiveness to approximately 59%, while a negative result decreased it to approximately 5% (approximate pooled LR + 6.0; LR- 0.22).

However, effect estimates showed substantial heterogeneity (I² > 60%) and publication bias suggested sensitivity inflation (3-5%).

More results

Most studies were single-center and retrospective, with 82% conducted in East Asia and limited external validation.

“

Conclusion · 1 of 3

Hybrid AI showed higher pooled accuracy compared with radiomics or deep learning alone, but the certainty of evidence is low-moderate due to inconsistency, publication bias, and restricted population diversity.

Conclusion · 2 of 3

AI-based ground-glass nodules risk stratification should currently complement, rather than replace, multidisciplinary clinical decision-making until validated in prospective multicenter trials.

Conclusion · 3 of 3

Systematic review registration: https://www.crd.york.ac.uk/prospero/display_record.php? RecordID=CRD42025115051, identifier CRD42025115051.

Comparative diagnostic accuracy of radiomics, deep learning, and hybrid AI for invasiveness stratification of pulmonary ground-glass nodules: a systematic review and meta-analysis.

Comparative diagnostic accuracy of radiomics, deep learning, and hybrid AI for invasiveness stratification of pulmonary ground-glass nodules: a systematic review and meta-analysis.

Related papers