Related Experiment Video
Updated: Dec 12, 2025
15:48
Tracking the Mammary Architectural Features and Detecting Breast Cancer with Magnetic Resonance Diffusion Tensor Imaging
Published on: December 15, 2014
22.9K
Mixed-Methods Study to Predict Upstaging of DCIS to Invasive Disease on Mammography.
Lars J Grimm1, Benjamin Neely2, Rui Hou1
1Department of Diagnostic Radiology, Duke University Medical Center, 2301 Erwin Rd, Box 3808, Durham, NC 27710.
AJR. American Journal of Roentgenology
|August 14, 2020
Summary
Radiologists can improve their ability to predict invasive disease in ductal carcinoma in situ (DCIS) using a two-stage observer study. This training helps identify patients who should not undergo active surveillance, improving patient care.
Area of Science:
- Radiology
- Oncology
- Breast Imaging
Background:
- Increasing incidence of ductal carcinoma in situ (DCIS) raises concerns about overtreatment.
- Active surveillance is a treatment option for DCIS, but identifying patients with underlying invasive disease is crucial.
- Radiologists' ability to predict upstaging of DCIS to invasive disease is an area for improvement.
Purpose of the Study:
- To evaluate if a mixed-methods, two-stage observer study can enhance radiologists' prediction of DCIS upstaging to invasive disease on mammography.
- To identify specific imaging features and criteria that improve the prediction of invasive disease in DCIS cases.
Main Methods:
- A retrospective study involving 300 DCIS cases (240 pure DCIS, 60 upstaged to invasive disease) from stereotactic biopsies (2010-2015).
- Nine breast radiologists reviewed mammograms in two blinded rounds, with a focus group discussion between rounds to establish consensus criteria.
- Performance was assessed using the area under the receiver operating characteristic curve (AUC) to compare prediction accuracy before and after the intervention.
Main Results:
- Initial reader performance (Round 1) had a mean AUC of 0.623.
- Consensus criteria identified included the presence of associated masses, asymmetry, architectural distortion, larger areas of densely packed calcifications, and focusing on suspicious rather than common features.
- Reader performance significantly improved in Round 2 (mean AUC, 0.765; p = .045) after the intervention.
Conclusions:
- A mixed-methods, two-stage observer study effectively improved radiologists' ability to predict upstaging of DCIS to invasive disease.
- Training and consensus criteria development can enhance diagnostic accuracy in breast imaging, potentially impacting treatment decisions and patient management.
