A technique for digital image registration used prior to subtraction of lung images in nuclear medicine.

The aim of the approach to image registration described is to find a general relationship, described by polynomials, for the distribution of activity in the perfusion image compared to the distribution in the composite image of perfusion and ventilation. This technique was used because the actual spatial relationship between the lungs in both images will never be known prior to re-alignment. For each pair of images the spatial relationship is unique and is characterised by the coefficients of the polynomial expressions. The coefficients for each polynomial approximation are calculated using picture element locations (control points). These corresponding pairs are located manually at similar anatomical positions in both images with a user-guided screen cursor. The appropriate grey level for each pixel in the registered perfusion image is found in the original at coordinate locations which are returned by the polynomial expressions.