Abstract: A series of nuclear images of a heart is acquired and stored in a memory device. One of the images is assumed to have occurred at the diastole and the perimeter of the left ventricle in this image is found. This perimeter is applied to the image series to produce a sum of the radiation counts within the perimeter for each image. If the maximum count sum is not produced by the image from which the perimeter was found, a new perimeter and a new maximum count sum are produced specifically for that maximum count sum image. A group of five consecutive images, which include the image from which the minimum count sum was produced, is defined. For each image in this group the left ventricle perimeter and the sum of enclosed radiation counts are individually produced. The smallest of these latter sums is used with the maximum count sum to derive the cardial ejection fraction.