Related Experiment Videos
Interreader variability and predictive value of US descriptions of solid breast masses: pilot study.
P H Arger1, C M Sehgal, E F Conant
1Department of Radiology, University of Pennsylvania Medical Center, Philadelphia 19104, USA.
Academic Radiology
|April 11, 2001
Summary
This study assessed ultrasound characteristics for solid breast lesions, finding high specificity and positive predictive value (PPV) for predicting biopsy results. The developed US recommendation system accurately predicts histologic findings, showing good interreader agreement.
Area of Science:
- Radiology
- Medical Imaging
- Oncology
Background:
- Solid breast lesions require accurate characterization to guide management.
- Ultrasound (US) is a primary imaging modality for breast lesion evaluation.
- Standardized interpretation of US characteristics can improve diagnostic accuracy.
Purpose of the Study:
- To evaluate the specificity of specific ultrasound (US) characteristics for solid breast lesions.
- To assess interreader variability in the interpretation of these US characteristics.
- To validate a US recommendation system for predicting histologic findings.
Main Methods:
- Prospective evaluation of 70 solid breast masses in 61 patients scheduled for biopsy.
- Three readers independently interpreted sonograms using established US characteristics.
- Specificity and positive predictive value (PPV) calculated against biopsy results.
- Interreader variability assessed, and a 5-category assessment system developed.
Main Results:
- High average specificities for key US characteristics: spiculation (97%), duct extension (95%), taller than wide (91%).
- Overall average specificity across readers was 88.5%.
- Positive predictive values (PPVs) for assessment categories II-V ranged from 5% to 94%.
- Readers demonstrated similar interpretations with good correlation.
Conclusions:
- The proposed US recommendation system accurately predicts histologic findings of solid breast lesions.
- A standardized sonographic classification lexicon enhances diagnostic accuracy and clinical decision-making.
- High specificity and interreader agreement support the utility of US characteristics in breast lesion assessment.