Validation of an activity optimization method for nuclear medicine in planar studies

A method for optimizing the administered activity in Static Nuclear Medicine Studies is validated by comparison with ROC curve. Linear Discriminant analysis of image quality in gamma cameras was the applied statistical technique. The constructed linear discriminant function owns as dependent parameters, the differentiated levels of image quality obtained by observer's criterion. The independent parameters in the function were physical variables, as Signal-to Background ratios and Signal-to-Noise ratios. They were obtained from the selection of Regions of Interest in images obtained from a Jaszczak phantom, corresponding to lesion and background sites. The percentage of cases correctly classified by discriminant analysis was analyzed to grade the proposed discriminant method. The minimum value of the administered activity, which permits good image quality, (it means good results for the parameters selected by the discriminant function), can be proposed as an optimized value of activity for planar studies of Nuclear Medicine. The method was tested using images from a Jaszczak phantom, acquired under four activities (1088 MBq, 962 MBq, 740 MBq and 562 MBq) with a gamma camera equipped with a high resolution - low energy- parallel-hole collimator. The gamma camera was tested by a NEMA protocol. Image quality was graded by three expert observers who also developed a rated procedure which consist in analyzing the images for ROC analysis. Two of the six measured Background-to-Signal ratios were the parameters able to construct the linear discriminant function with high correlation respect to the observer criterion, from all the measured physical variables. The value of 740 MBq was the optimum after discriminant method application in this particular experiment. The results were coincident with the application of ROC-analysis. The optimal activity value obtained with the proposed discriminant procedure coincided with the activity value for which the area under the ROC curves was higher, the sensitivity was the best and the false-positives and false-negatives rates were the smallest. (Author)