Fully Automated Gating of Optical Coherence Tomography Data
K Sihan1, C Botha2, F Post2, S de Winter1, E Regar1, PJWC Serruys1, R Hamers1, N Bruining1
1Thoraxcenter, Department of Cardiology, Erasmus MC, Rotterdam, The Netherlands
2TU-Delft, Delft, The Netherlands
Abstract
Intra-coronary optical coherence tomography (OCT)
provides ultra-high resolution imaging of coronary ves-
sel wall structures. However, during image acquisition the
OCT catheter is affected by cardiac motion. These motion-
induced artifacts not only complicate longitudinal image
reconstructions, it results in a saw-tooth shaped appear-
ance of the coronary vessel wall, but more importantly it
affects the accuracy of quantitative analysis (QOCT). To
overcome this problem we propose to perform image-based
gating applying a genetic algorithm (GA) that automati-
cally selects a subset of OCT cross-sections that are rela-
tively unaffected by the catheter displacement during the
cardiac cycle. The gated subset contains cross-sections
(frames) acquired in the near end-diastolic phase, during
which the heart is relatively motionless. We evaluated the
GA in a comparison test with a different gating method
(Simulated Annealing (SA)) and with manual frame selec-
tion (MFS) and found promising results.
1. Introduction
Optical coherence tomography (OCT) is a new intra-
coronary imaging technique capable of cross-sectionally
imaging the coronary arteries at high image resolution,
close to that of histopathology. This resolution is much
higher than that of intra-coronary ultrasound (ICUS) the
current de-facto reference intra-coronary imaging method
and shows much more morphological details. For later
quantitative analysis, especially within studies evaluat-
ing new treatment methods, the lumen areas and conse-
quently lumen volumes are measured. The current stan-
dard method for quantitative OCT analysis involves man-
ual selection of frames at fixed distances, since analysis of
all frames (several hundred per patient) is impractical. This
method is referred to as the manual frame selection (MFS).
However, cardiac motion induced catheter displacement
could cause a reduced quantification accuracy. The mo-
tion effects appear as a saw-tooth shaped representation
of the coronary vessel wall in reconstructed longitudinal
views, which are often used during quantification. Within
the MFS method the time interval between the manually
selected frames is 1-to-1 translated to distance; however,
this assumption is incorrect as longitudinal catheter mo-
tion influences the distance (and thus the anatomical posi-
tion) between the locations where the frames are acquired.
Furthermore, during the cardiac cycle the vessel dimen-
sions vary, under influence of the blood pressure, where
the MFS does not account for as the frame selection is ran-
dom with respect to the cardiac cycle. Therefore, the ac-
curacy of the coronary dimensions as quantified by MFS
is diminished. By applying gating, these problems may be
prevented, analogous to what has been proposed for ICUS
previously [1].
2. Materials
OCT imaging was performed with a commercially avail-
able system (Lightlab imaging, Westford, MA, USA). This
system uses a 1310-nm broadband light-source generated
by a super luminescent diode with an output power of
8.0mW. The average tissue penetration depth is approxi-
mately 1.5 mm with an axial and lateral resolution of 15
µm and 25 µm, respectively. The imaging probe has the
size of a guide-wire with a maximum outer diameter of
0.019 inch (ImagewireTM, LightLab Imaging). The wire
contains a single-mode fibre optic core within a translu-
cent sheath. It is connected to an imaging console, sim-
ilar to ICUS, that is responsible for real-time image data
processing, visualization and image storage. Systematic
imaging of a coronary segment is also similar to ICUS by
an automatic continuous speed pullback (between 1 and 3
mm/s) of the imaging wire. OCT images are generated at
a rate of 5-20 per second (ICUS 30 frames/s).
3. Automated image-based OCT gating
Image-based gating methods rely on information re-
trieved from the images themselves (e.g. called features).
Retrospective image-based gating methods are feasible as
has been published for ICUS [1]. Unfortunately, the ap-
proach used for ICUS cannot be applied for OCT, mainly
due to the fact that there are much less details of the coro-
nary wall visible because the tissue penetration depth of
ISSN 0276−6574 9 Computing in Cardiology 2010;37:9−12.