Automated Readout (CDCOM)
CDCOM version 1.6
In Spring 2011 problems concerning the physico-technical quality control using CDMAM3.4 and CDCOM have been observed on some mammography systems. As stated in our letter of 18th May, 2011 EUREF and Artinis Medical Systems B.V. were aware of the importance of these problems and therefore combined forces to analyze and solve the problem. This has resulted in a new version of CDCOM (version 1.6)
Changes in CDCOM version 1.6 compared to version 1.5.2.
In CDCOM version 1.5.2 the position of the gold discs of the CDMAM phantom were estimated using the gridlines of the phantom. Gridline variations resulted in an inaccurate estimation of the gold disc positions. With the small gold disc search area used in CDCOM version 1.5.2 this led to incorrect scores for some specific phantom images.
In CDCOM version 1.6 the position of the gold discs is estimated using the discs on the phantom which are clearly visible in combination with a template with the actual position of all discs.
This template is translated, rotated and skewed for an optimal fit to the phantom.
Evaluation of CDCOM version 1.6
We have performed an extensive evaluation and have received data from several third parties for this purpose. The new version solves the problem concerning incorrect estimated gold disc position. We have evaluated this version using the following method, which is suggested in the forthcoming supplement to the European Guidelines:
- The individual CDCOM scores are added up to a detection matrix
- This matrix is smoothed using a Gaussian function
- A psychometric curve has been constructed for each diameter and the 62.5% points are determined on this curve
- A third order polynomial function has been fitted to all 62.5% points, coefficients >0
Due to (small) differences in the search area which is used in the old and new version of CDCOM, scores of individual CDMAM images are very much alike but not 100% identical. However, for series of phantom images which did not have problems with the inaccurate estimate of gold disc positions, scores of series are similar. For series of phantom images which did have problems the score should be as expected using CDCOM version 1.6.
More information about the evaluation of CDCOM which has led to version 1.6 can be found in the downloadable pdf-file below.
Remarks: In the evaluation of CDCOM the presence of artefacts may influence the score. Furthermore care has to be taken that the 62.5% points on the psychometric curve are actually on the phantom, extrapolation might introduce inaccurate results.
In spite of our extensive evaluation of this new version, in special cases it might happen that unexpected bugs appear. We kindly ask users of this version to report bugs, so proper action can be taken.
CDCOM software (version 4.1.0) for CDMAM4.0
The Guildford CDMAM Analyser version 2.1 (from the NCCPM in Guildford) including software and instruction manual for automated determination of threshold contrast can be obtained using the link below.
Please note that this software has not been developed by EUREF. The functioning and correctness of the software is the responsibility of the developers as mentioned on their website.