Mammographic mass eigendetection

Arnau Oliver, Jordi Freixenet, Robert Marti, Erika R. E. Denton, Reyer Zwiggelaar

Research output: Contribution to conferencePaperpeer-review

Abstract

A new algorithm for the detection of masses in mammographic images is presented. The algorithm has been designed in two steps. Firstly, the regions likely to be a mass are detected by using a deformable template matching approach, where the template is constructed using the eigenimages of a set of manually detected masses. Subsequently, an algorithm adapted from the eigenfaces approach is used to assure that the detected regions really depict true masses (false positive reduction). The evaluation uses a leave-one-out methodology and is based on a database of 120 mammograms, which include 40 masses and 80 normals. ROC and FROC analysis is used to demonstrate the potential of the developed approach.
Original languageEnglish
Pages71-75
Number of pages5
Publication statusPublished - 04 Jul 2006
Event10th UK Conference on Medical Image Understanding and Analysis - MIUA: 4-5 th July 2006, University of Manchester - University of Manchester, Manchester, United Kingdom of Great Britain and Northern Ireland
Duration: 04 Jul 200605 Jul 2006

Conference

Conference10th UK Conference on Medical Image Understanding and Analysis - MIUA
Abbreviated titleMIUA 2006
Country/TerritoryUnited Kingdom of Great Britain and Northern Ireland
CityManchester
Period04 Jul 200605 Jul 2006

Fingerprint

Dive into the research topics of 'Mammographic mass eigendetection'. Together they form a unique fingerprint.

Cite this