Luminal volume reconstruction from angioscopic video images of casts from human coronary arteries

Intravascular angioscopy has been hampered by its limitation in quantifying obtained images. To circumvent this problem, a lightwire was used, which projects a ring of light onto the endoluminal wall in front of the angioscope. This investigation was designed to quantify luminal dimensions of casts from stenotic human coronary arteries and to investigate the feasibility of performing volume reconstruction. Angioscopic video recordings were made during simultaneous motorized pullback (2 mm/s) of an angioscopic catheter and light-emitting fiber through 4 epoxy casts from human coronary arteries. Quantification of the images was performed using a computerized cross-sectional area measurement system. The coronary casts were divided into cross-sectional segments of 4 to 10 mm in thickness, with the true volume of each segment determined by means of a microdispenser. Because of vessel curvature and luminal narrowing, complete visualization of the ring of light at all 1-mm-distant locations was only possible in 19 of 40 segments. For these 19 segments, linear regression analysis showed a good correlation between measured and true segmental volume (r = 0.97, y = 0.88x + 6.58 mm3, standard error of estimate = 3.48 mm3). The relative error in the measured segmental volumes was 3.9 +/- 7.1% (mean +/- SD). These initial results of endoluminal volume reconstruction demonstrate the feasibility of this technique in vitro with high accuracy and low variability, but further technical improvements are necessary to increase the success rate, especially in the quantitative assessment of vessels with complex morphology.