EVALUATION OF THE FACTORS AFFECTING THE ACCURACY AND PRECISION OF A TECHNIQUE FOR QUANTIFICATION OF VOLUME AND ACTIVITY IN SPECT

The use of the technique of maximization of interclass variance to determine the threshold level defining the volume of an object in single photon emission computed tomography (SPECT) and hence its activity has been evaluated in a variety of conditions. The influence of varying activity on volume assessment and varying volume on activity assessment were shown to be negligible. The effect of object shape for the range of shapes likely to be found in the body was also small. Non-uniformity of activity in the object caused an underestimation of volume and to a lesser extent its activity. However, with activity varying by 100% within the volume, volume was only underestimated by 10% and activity by 3%. The presence of activity surrounding the object caused an overestimation of volume which increased linearly with the relative level of surrounding activity. The object to surrounding activity concentration ratio could be estimated and therefore its effect corrected. The influence of nearby objects and methods of dealing with their effect on the technique are described. Activity concentrations in objects with uniform distribution assessed using total activity divided by total volume were shown to be more accurate than assessments from maximum voxel count rate for objects with a volume below 150 ml.