Quantitative characterization of signal-to-noise ratios in diagnostic imaging instrumentation

Different imaging modalities can operate on basically different object (biologic) parameters. In addition to an understanding of the intrinsic differences among modalities, it is useful to also understand how these differences translate into information on the output image. Thus, given a defined diagnostic situation, it is important to estimate the confidence (or signal-to-noise level) with which a particular abnormality is observed. For this purpose, we have extended the Rose model for characterization of signal to noise in images by taking into account the sources of degradation of contrast found in diagnostic imaging instrumentation. The model is not only useful in comparing different instruments applied to the identical physiologic state, but can also be used to compare signal to noise in images resulting from different instruments imaging the same subject under different conditions (for instance, positron and planar imaging in the head). The model can also be formulated so as to compare the signal to noise in images obtained from the same object by totally different modalities. A particularly useful feature of this model is its ability to yield an understanding of the intrinsic limitations of any one modality.