Image Processing Reference
In-Depth Information
arterial phase and large and small veins in the venous phase of the 4D CTP data
as well as in the background. The SNR was determine for large and small vessels
by subtracting the background mean from the vessel mean and dividing by the
standard deviation of the background. The average SNR over all 20 CTP scans
increased significantly ( p< 2 . 8
10 4 paired two-tailed T-test) from 14.6 ( ˃ :4.1)
to 29.5 ( ˃ :9.4) for large vessels ( > 1 . 5 mm ) and from 3.6 ( ˃ :1.1) to 6.8 ( ˃ :3.0)
for small vessels ( < 1 . 5 mm ).
·
(a)
(b)
(c)
(d)
Fig. 2. Subimages of maximum intensity projections of an arteriogram and venogram
(W/L 320/160) from two different patients used in the evaluation. The arrows indicate
locations where visualization of smaller vessels improved due to filtering. (a) Arteri-
ogram derived from the original CTP data. (b) Arteriogram derived from the CTP
data filtered with the TIPS anisotropic diffusion filter. (c) Venogram derived from the
original CTP data. (d) Venogram derived from the CTP data filtered with the TIPS
anisotropic diffusion filter.
4
Discussion and Conclusion
In this paper a TIPS anisotropic diffusion method was proposed for enhancing
vessels and reducing noise in 4D CT perfusion (CTP) data, to improve the qual-
ity of arteriograms and venograms derived from this data. The proposed filter
uses the 4th dimension to distinguish between structures after which diffusion
is performed on each 3D sequential scan in the 4D dataset. Therefore, intensity
values are not mixed over time, preserving the time-intensity profiles. Even when
Search WWH ::




Custom Search