Related Experiment Video
Updated: Jul 9, 2026
05:28
Clinical Imaging of Microwave Mammography
Published on: November 14, 2025
A Systematic Review and Meta-analysis of Two-Dimensional Breast Ultrasound Radiomics: Implications for Lesion
Omar Abd Al Mjed Allasasmeh1, Areej H Al-Sarairah2,3, Muath A Odeh Hurani4
1Centre of Diagnostic, Therapeutic and Investigative Studies (CODTIS), Faculty of Health Sciences, Universiti Kebangsaan Malaysia, Jalan Raja Muda Abdul Aziz, 56000, Kuala Lumpur, Malaysia.
Journal of Imaging Informatics in Medicine
|July 7, 2026
Summary
Radiomics using 2D breast ultrasound shows promise for distinguishing benign from malignant lesions. However, methodological limitations suggest pooled accuracy estimates may be optimistic, requiring further validation for clinical use.
Area of Science:
- Medical Imaging
- Radiology
- Artificial Intelligence in Medicine
Background:
- Radiomics offers a noninvasive method for breast lesion characterization using ultrasound.
- Current diagnostic accuracy and reliability of radiomics in breast ultrasound require further investigation.
Purpose of the Study:
- To systematically review and meta-analyze the diagnostic performance of radiomics in differentiating benign from malignant breast lesions on 2D ultrasound.
- To assess the methodological quality and identify limitations of existing studies.
Main Methods:
- PRISMA-DTA-compliant systematic review and meta-analysis of observational studies (2019-2026).
- Searched PubMed, Scopus, Web of Science, Cochrane Library for 27 eligible studies.
- Estimated pooled sensitivity, specificity, and HSROC AUC using bivariate random-effects models.
- Assessed heterogeneity, threshold effects, and publication bias using I², Spearman correlation, and Deeks' test.
- Evaluated risk of bias using QUADAS-2.
Main Results:
- Pooled sensitivity: 0.85 (95% CI 0.82-0.88), Pooled specificity: 0.87 (95% CI 0.83-0.91).
- Strong overall discrimination (HSROC AUC 0.92).
- Substantial heterogeneity (I² 74.0-85.5%) and frequent high/unclear risk of bias in patient selection and index test domains.
Conclusions:
- Radiomics models show clinically meaningful diagnostic performance but may overestimate accuracy due to methodological limitations.
- Prospective, multi-center, multi-vendor validation and standardized reporting are crucial for clinical translation.
- Further research needed to evaluate incremental utility within BI-RADS workflows.
