contact author: matthias.benndorf[at]uniklinik-freiburg.de, homepage updated: 04/03/2017


Naive Bayes classification of mass lesions visualized with X-ray mammography

Disclaimer: The calculations provided at this site are a research-only project based on the data provided in Benndorf et al. 2015, they are not sufficient for medical decisions/diagnosis and do not replace the judgement of a radiologist. We do not take legal responsibility for any injury and/or damage inflicted to persons or property based on the suggestions provided herein.

The classifier’s methodological development and testing are detailed in Benndorf et al. 2015. A naïve Bayesian approach is employed to calculate posttest probabilities. The classifier is calibrated using a binning technique.
information about Naive Bayes classifiers| link to original publication | download threshold tables
BI-RADS descriptorsadditional information

shape
round
oval
lobular (4th edition)
irregular

margin
circumscribed
microlobulated
obscured
ill-defined
spiculated

density
hypodense
isodense
hyperdense
fat-containing

age
younger than 50
50 - 64
older than 64


BI-RADS category
not sure
0 (evaluation incomplete)
2
3
4
5