Impact of Computer-Aided Detection Prompts on Mammography Decisions
Author Information
Author(s): Fiona J Gilbert, Susan M Astley, Caroline RM Boggis, Magnus A McGee, Pamela M Griffiths, Stephen W Duffy, Olorunsola F Agbaje, Maureen GC Gillan, Mary Wilson, Anil K Jain, Nicolas Barr, Ursula M Beetles, Miriam A Griffiths, Jill Johnson, Rita M Roberts, Heather E Deans, Karen A Duncan, Geeta Iyengar
Primary Institution: University of Aberdeen
Hypothesis
The study investigates how the presence and size of computer-aided detection (CAD) prompts affect reader behavior in breast screening mammograms.
Conclusion
Mammograms with CAD prompts, especially larger ones, were more likely to be recalled for further assessment.
Supporting Evidence
- 64% of all cases contained at least one CAD prompt.
- Larger prompts were more likely to be recalled for cancer cases.
- 58% of cancer cases prompted at any location were marked for recall.
- 82% of cancer cases with prompts in the region of interest were recalled.
- Significant associations were observed with prompting and breast density for cancer cases.
Takeaway
When doctors look at breast X-rays, prompts from a computer can help them notice more problems, especially if the prompts are bigger.
Methodology
The study analyzed mammograms from women attending routine screenings, comparing reader decisions with and without CAD prompts.
Potential Biases
Reader variability and the influence of study participation on decision-making may introduce bias.
Limitations
The study's retrospective design limits the ability to separate the effects of prompt size from the appearance of cancer on mammograms.
Participant Demographics
Women aged 50 years or older attending routine mammography.
Statistical Information
P-Value
p<0.001
Statistical Significance
p<0.001
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website