Main Menu

Home
News
Research
Coding
Teaching
Articles
Action
AboutMe
On-Site

Annotation DB mammo PDF Print E-mail
Written by Harris Georgiou   
Sunday, 28 March 2010 23:29

1. Introduction

The goal of "Annotation DB Mammo" project, or "annotDB" for short, is to provide an easy and exportable electronic documentation database for BI-RADS mammographic annotation. It can be used as-is along with any breast imaging software (mammographic or ultrasound) or as a supplementary data tool for making the annotation process easier and faster for the doctors, before the data are processed later on.

The structure and contents of the core annotation tables (lists) are directly mapped to the BI-RADS annotation standard for mammographic and ultrasound breast imaging and diagnosis [see ref: 1], as well as two relevant works by researchers in the National & Kapodistrian University of Athens, Greece [see ref: 2-3]. It has been used successfully in research projects related to medical imaging diagnosis and machine learning.

The GUI forms are currently customized only for the BI-RADS annotation of mammographic images (not untrasound), but the data tables support both. There are two forms in the database, namely "_examination" and "_findings". The first one is the main GUI form that is displayed automatically on start. The second one is a supplementary subform that normally should not be called explicitly by the user, unless for reviewing the corresponding data series.

The "annotDB" package is based on the Microsoft Access platform, but the data tables are importable to other equivalent platforms like OpenOffice (Base). The data tables can also be exported from Access to a variety of text-based formats that can be imported by any application.

2. Quickstart

2.1  Installation

The "annotDB" package comes with a full installer for Windows platforms as well as a simple .zip package with no installer attached. In Windows, the installer creates the proper shortcuts in Start Menu and in the Desktop and provides additional options for installation folder. See below in "Download" for downloading links.

The package contains two main files:

  • annotDB_data.mdb  :  MS-Access file containing only the data tables
  • annotDB_mmForms.mdb  :  MS-Access file containing the GUI (optional)

In order to use the GUI file (forms), MS-Access version 2003 or later must be installed in the host computer. Alternatively, the MS-Access 2007 runtime can be downloaded and used instead from:  http://tinyurl.com/msacc2007rt

The file containing the data tables can be used as-is with any compatible platform like OpenOffice's Base. The full contents of the BI-RADS tables can be reviewed in the "BI-RADS_Tables.pdf" file.

2.2  Networked deployment

The "annotDB" package can be deployed in a networked, multi-user environment, using one central data storage and multiple GUI copies. In case where multiple users need simultaneous access to the database, the GUI forms file can be shared from a networked repository directly or copies of it can be distributed in every working station (locally).

In this second case, the local file must be updated so that the table links refer to the correct network path. This can be done within MS-Access by using:

  • (main menu) -> Tools -> Database Utilities -> Linked Table Manager

2.3  Other operating Systems

The "annotDB" package is typically used in Microsoft Windows systems with a compatible MS-Access suite or runtime installed, so that the GUI (forms) can be used as the front-end.

In case that no MS-Access suite or runtime is available, or if the host is a MacOS, Unix, Linux, etc, it is recommended that OpenOffice's Base is used instead. The data tables can be imported or linked externally, but the GUI forms have to be recreated within Base, as the scriping language is different and the format cannot be imported (at least until OpenOffice version 3.1).

For large-scale deployment, the data tables can be converted to a standard text-based format (CSV, XML, ...) and then imported to an on-line RDBMS platform, with a custom web-based GUI client.

3. Recommended practices

3.1  User access control

Please be advised that the "annotDB" package deals with sensitive data, namely patient's name, age and medical history. Thus, it is extremely important that access to the host computer and the data tables enforces strict policies for certified users.

No explicit access control is implemented in the MS-Access database files, in order to keep import/export overhead and compatibility issues to the minimum. The access policy should be employed in the lower level of the operating system itself, i.e., grant access only to specific system users.

3.2 Backup and archiving

In order to keep the database files secure and redundant, it is recommended that the "annotDB" files, especially the data tables (annotDB_data.mdb), should be frequently archived and stored in a secure place.

The simplest way to do this is using any compression tool to reduce the size of the data files, possibly with password protection, and email it to a certified user account for archiving. A very popular compression tool with freeware license and seamless integration to the Windows Explorer is 7-zip:

Using this, or any other similar tool, the admin user can simply go to the "annotDB" installation folder, point the data file (annotDB_data.mdb) and select mouse right-click and "compress and email". Make sure you first delete any older compressed files from the folder.

3.3  Uninstalling

The Windows installer of "annotDB" creates an uninstall shortcut in the package's group in the Start Menu. The uninstaller completely removes all files and shortcuts from the system.

CAUTION: Please not that the database files are completely removed by the uninstaller. Make sure you keep backups fo your files before uninstalling the "annotDB" package.

4. Support and feedback

The "annotDB" package is provided as-is, with no guarantee for support to users. However, the authors will be happy to respond to any feedback and bug reports, as well as suggested changes, in future versions.

Please read the "README.txt" file before asking questions with regard to installation, networked deployment or file organization.

The contact author can be reached at:   http://tinyurl.com/xgmailme

5. Change Log

  • version 2.3  (2010-04-15):  Minor changes in the GUI, added a new field for additional information about the pathology. 
  • version 2.2  (2010-03-28):  Minor changes in the GUI, added four fields for references to images 1-2-3-4 related to a finding.
  • version 2.0  (2010-03-26):  Redesigned some of the fields in the GUI, final structure and contents of data tables.
  • version 1.5  (2010-03-22):  Final beta release, various bug and GUI fixes.
  • version 1.0  (2010-03-18):  First beta release, various bug and GUI fixes.
  • version 0.9  (2010-03-15):  Finished alpha release, internal testing phase.

6. Copying, Licensing, Redistribution

The "annotDB" is a copyrighted work distributed and licensed under European Union's Public License (EUPL) v.1.1:

This means that it is freely redistributable as-is or modified by any user, provided that it is also distributed with the same or an equivalent license and the original authors are noted in the attribution section.

No registration or fee is required whatsoever by users and researchers that wish to use this software for non-commercial use. Please note the liability clauses, especially regarding privacy issues, since the data forms contain fields for some patient data.

Since the package is maintained and constantly updated by the authors, it is recommended that no unaltered package versions are "mirrored" elsewhere but instead provide a link to the original (author's) repository with the latest stable version here:

7. Download latest version

You can download the latest version of the "annotDB" here:

Please do not use direct links for download in your pages, use links to this page instead.

Thank you for using our software Laughing

References

[1]  American College of Radiology (ACR), Breast Imaging Reporting and Data System Atlas (BI-RADS Atlas). Reston, Va: (c) American College of Radiology; 2003.

[2]  Michael E. Mavroforakis, Geometric Approach to Statistical Learning Theory through Support Vector Machines (SVM) with Applications to Medical Diagnosis (PhD thesis), Dept. of Informatics & Telecomm., National & Kapodistrean Univ. of Athens, Athens, Greece, Aug. 2008.

[3]  Harris V. Georgiou, Algorithms for Image Analysis and Combination of Pattern Classifiers with Application to Medical Diagnosis (PhD thesis),
Dept. of Informatics & Telecomm., National & Kapodistrean Univ. of Athens, Athens, Greece, Feb. 2009.

 

Last Updated on Thursday, 15 April 2010 15:42