The effect of in-vivo-occurring errors in the reproducibility of radiographs on the use of the subtraction technique.

A device for serial radiography in maxillary and mandibular premolar and molar regions was developed. Radiographs of 9 patients were made during a period of 1 year. When testing the device, the influence of the time interval between the radiographs and the influence of the construction of the device on the reproducibility of the X-ray images were determined. An increase of the time interval between radiographs reduced the reproducibility. The use of bilateral bite blocks improved the reproducibility of the X-ray images in comparison with the images obtained with a device with a one-sided bite block. Errors in the reproducibility caused differences in the image geometry of radiographs. The in-vivo-found differences were studied for their effect on the detection threshold of observers using photographic subtraction radiography. An increase of the differences in image geometry between the radiographs resulted in a higher detection threshold and a higher % of false positive decisions. It appeared that very small differences between clinical radiographs, corresponding with an angulation error of 0.7 degrees, can be tolerated in order to prevent the occurrence of high numbers of type-I errors. The % of radiographs, which met this requirement and could be used for photographic subtraction radiography, was 55%, when taken at time intervals of 1 year.