Loughborough University
Browse
7 Evaluation of a CAD enhanced synthetic mamamogram (v2).pdf (80.8 kB)

Evaluation of a computer-aided detection (CAD)-enhanced 2D synthetic mammogram: comparison with standard synthetic 2D mammograms and conventional 2D digital mammography

Download (80.8 kB)
journal contribution
posted on 2018-06-15, 08:31 authored by Jonathan James, E. Giannotti, Yan Chen
Digital breast tomosynthesis (DBT) when combined with standard 2D digital mammography has been shown to improve the performance of breast cancer screening by increasing cancer detection rates [1-5]. The 2D component remains an important part of the examination and is used to facilitate assessment of symmetry between the breasts, aid comparison with prior mammograms and identify the presence of breast microcalcifications where the evidence for detection with DBT is less robust [1]. The mean glandular dose per view of a DBT image is around 2.3 mGy, which is between 1 - 1.5x more than the dose of standard 2D digital mammography [6]. Acquiring both a DBT and standard 2D digital mammogram on each woman leads to at least a doubling of the radiation dose, which may not be considered acceptable in an asymptomatic screening population. Consequently there has been much interest in the generation of synthetic 2D mammograms from the DBT data set eliminating the additional radiation burden of a separate 2D digital mammogram. There is evidence from prospective and retrospective studies to support the use of synthetic 2D mammograms [5,7-9]. Several retrospective multi-reader studies, including the UK TOMMY trial, have demonstrated comparable performance between synthetic and conventional 2D mammography [7,8]. The Oslo and Storm-2 prospective studies of DBT in breast cancer screening found equivalent cancer detection rates regardless of whether the conventional 2D or the synthetic mammograms were read, concluding that synthetic mammograms were an acceptable replacement for directly acquired conventional 2D mammograms [5,9]. Another approach to improve performance is to combine the synthesised image with a Computer Aided Detection (CAD) algorithm. CAD has been used over the years to assist with the interpretation of 2D mammography. CAD software places marks or prompts on the images to draw the reader’s attention to potential areas of concern, reducing observational oversights. A CAD algorithm has been developed with machine learning technology (iCAD Inc., Nashua , NH, USA and GE Healthcare, Buc, France) to assist in the detection of breast cancer on DBT images. Unlike a conventional CAD system which places marks on the image, areas of concern are automatically identified on each tomosynthesis slice and then blended onto a 2D synthetic image to provide a single CAD enhanced 2D synthetic image for each mammographic projection. The aim of this study was to evaluate the diagnostic performance of the CAD enhanced synthetic mammogram in comparison with standard 2D synthetic mammograms generated from the DBT data set and standard 2D digital mammography.

History

School

  • Science

Department

  • Computer Science

Published in

Clinical Radiology

Citation

JAMES, J., GIANNOTTI, E. and CHEN, Y., 2018. Evaluation of a computer-aided detection (CAD)-enhanced 2D synthetic mammogram: comparison with standard synthetic 2D mammograms and conventional 2D digital mammography. Clinical Radiology, 73(10), pp. 886-892.

Publisher

© The Royal College of Radiologists. Published by Elsevier

Version

  • AM (Accepted Manuscript)

Publisher statement

This work is made available according to the conditions of the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) licence. Full details of this licence are available at: https://creativecommons.org/licenses/by-nc-nd/4.0/

Acceptance date

2018-05-30

Publication date

2018-06-30

Notes

This paper was accepted for publication in the journal Clinical Radiology and the definitive published version is available at https://doi.org/10.1016/j.crad.2018.05.028

ISSN

1365-229X

Language

  • en

Usage metrics

    Loughborough Publications

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC