Normal Database Creation

The concept of the normal database assumes that the perfusion and the metabolism of the myocardium in a healthy left ventricle shows a consistent pattern. It furthermore assumes that the segmentation model is a suitable standardization of the anatomy across persons. Under these premises, the results obtained with healthy hearts can be pooled for a statistical analysis to establish the average and the normal variation in each segment. This information represents the normal database for a certain type of cardiac scans. Results found in a patient scan can then be compared against this database resulting in an absolute deviation from the normal value per segment which can also be expressed as a fraction of the standard deviation (z-score).

The principle steps for building a normal database are:

• Recruitment of a sufficient number of healthy normal volunteers. Note that it would be desirable to form age and gender matched control groups.
• Performing the acquisition observing a strict protocol. The tracer should be applied in a consistent way, and the same stress condition used. The same type of attenuation correction should be applied, and the image reconstruction performed with the same algorithm and parameters. It is recommended use a reconstruction zoom and offset so that the reconstructed field-of-view is focused on the heart.
• Carefully analysing the data using PCARDP following a standard procedure. Note that the same kinetic and segmentation models should be used to pool the data. Kinetic models with fewer parameters are to be preferred to minimize the problem of fitting instabilities.
• The resulting rest, stress and reserve polar maps are the basis of the normal database. PCARDP includes a Norm Edit tool which allows to create new databases, and add normal polar plots to them. Finally, the tool calculates the average and the standard deviation of the normal values in each segment. These will be used in the evaluation of patient data.