Computation of a location shift between two subsequent intravascular ultrasound registrations by cross-correlation analysis of the lumen area functions

For longitudinal studies on atherosclerosis and restenosis after angioplasty by intravascular ultrasound (IVUS), it is essential that repeated studies are performed at exactly the same location along an arterial section. In human femoral arteries, lumen and plaque area functions of two subsequent IVUS pullback maneuvers were compared by cross-correlation analysis. In cross-correlation analysis of two functions with equal abscissa values, the data sets are repetitively correlated after incremental shifts of the two functions along the abscissa. This results in multiple correlation coefficients with a maximum at the relative position where the two functions show the closest match. In group A (12 patients), both pullbacks were performed after angioplasty and in group B (17 patients) one pullback was performed before angioplasty and the second immediately after angioplasty. In group A, cross-correlation showed a shift between lumen area functions of 5 mm in one patient and no shift in the other patients. Maximum correlation coefficients in group A ranged from 0.644 to 0.978. Four patients from group B showed shifts from 2.5 to 35 mm. Maximum coefficients were significantly smaller than in group A: 0.259-0.864 (p < 0.01). Plaque area functions in group B showed higher correlations (0.468-0.862, p = 0.034) and only two shifts. Cross-correlation of lumen and plaque area functions may he used to compute location shifts between two subsequent IVUS registrations and to correct such shifts.