Dataset Open Access

MM-S2TXTR-F03 Textural features of mammographic masses, Greece, Feb.2003

Harris V Georgiou (MSc,PhD); Michael E Mavroforakis (MSc,PhD)

========================================================

       Dataset: MM-S2TXTR-F03

    Textural features of mammographic masses
          Greece, Feb.2003
 
            Release Notes

    Copyright (c) 2016 by Harris V. Georgiou

========================================================
   Release:     Aug 3, 2016

   - Version:  1.1a
   - Format:   .mat
========================================================


This file contains important information about the current version of the dataset package. Downloading and using this material hints that you accept the EULA/Terms-of-Use (please read carefully).

We welcome your comments and suggestions.

_______________________________________________
WHAT'S IN THIS PACKAGE?

-  Overview
-  Available file formats
-  Files and Datasets
-  License Agreement

_______________________________________________
OVERVIEW

Localized texture analysis of breast tissue on mammograms is an issue of major importance in mass characterization. This material aims to the establishment of a quantitative approach of mammographic masses texture classification based on these datasets of the extracted textural features.

This package contains an extensive set of textural feature datasets, in multiple configurations and scales, constructing compact sets of textural "signatures" for benign and malignant cases of tumors in mammograms. The datasets refer to 142 localized masses in mammograms, as well as a set of qualitative features that have been proven as of utmost importance regarding the true pathology in each case. For the construction of the raw material for these datasets, 130 of the 142 MLO mammograms containing a mass were selected by an expert physician for digitization, including a total of 46 benign mass cases and 84 mass malignancies of various types (12 were marked 'unreliable').

The files also include the full texture data of 7,000-13,000 sampling "boxes" of sizes 20 and 50 pixels, as well as complete-mass and borderline-only sampling areas.

The digitized mammograms were subsequently transformed digitally to a resolution of 63 mm (400 dpi) at 8-bit gray level, which is consistent with other typical image databases of digitized mammograms that are used as a reference in similar studies.

Detailed information about the files and the structure of the datasets are available in the manifest file (S2_data_texture_directory.pdf).

_______________________________________________

AVAILABLE FILE FORMATS

The datasets are available in the following formats (included):

*.mat    : Matlab/Octave native data file (workspace)

 

For further information see: * M. Mavroforakis, H. Georgiou, et.al. "Significance Analysis of Qualitative Mammographic Features Using Linear Classifiers, Neural Networks and Support Vector Machines", European Journal of Radiology, 54 (2005) 80-89. (doi:10.1016/j.ejrad.2004.12.015) * M. Mavroforakis, H. Georgiou, et. al. "Mammographic Masses Characterization, based on Localized Texture and Dataset Fractal Analysis, Using Linear, Neural and SVM Classifiers", Artificial Intelligence in Medicine, Vol 37 (2) (2006) 145-162. (doi:10.1016/j.artmed.2006.03.002)
Files (17.2 MB)
Name Size
EULA-TermsOfUse.txt md5:f5be4d66f1d3d7c5ccdebe0c8602caee 2.9 kB Download
MM-S2TXTR-F03.7z md5:3ed81f8043d6b61edd31068eef2832ec 17.2 MB Download
MM-S2TXTR-F03.7z.asc md5:cf78648f346de90b2ddc4c0a663b706b 851 Bytes Download
README.txt md5:2b452ffbf34d5e51c7bee071a87200d8 3.9 kB Download

Share

Cite as