The SSD technique for examining RAS has been studied in several
clinical research papers. In the paper of Rubin et al. not only are
accuracy of SSD and MIP defined quantitatively but also SSD drawbacks are
depicted [1].
Thirty one patients underwent conventional renal arteriography. CT
angiographic data were reconstructed to form an image with Shaded Surface
Display(SSD) and Maximum Intensity Projection(MIP). Main and accessory renal
arteries were seen at conventional arteriography. The accuracy of CT
angiography was found to vary with the rendering technique employed. Stenosis
was graded on a four-point scale(grades 0-3). SSD CT angiography was 59
sensitive and 82
specific for the detection of grade 2-3 stenoses. The
accuracy of stenosis grading was 55
with SSD CT angiography. MIP CT
angiography was 92
sensitive and 83
specific for the detection of
grade 2-3 stenoses. The accuracy of stenosis grading was 80
with MIP
angiography. This paper also depicts some SSD technique drawbacks.
First, partial volume effects tend to result in vessel discontinuity at a point
of high-grade stenosis, rather than showing the extreme narrowing seen with
conventional arteriography. Second, in the presence of calcium within a renal
artery, the vessel lumen is overestimated with SSD CT angiography because of
the difficulty of distinguishing between the calcium in the vascular
obstruction and the intra-vascular contrast material.
Theoretically, choosing threshold values correctly for CT
angiography should result in a clear image that corresponds to the conventional
arteriogram for determining the degree of stenosis. In a very recent paper,
Halpern et al. carefully chose threshold values. However, several
unsatisfactory results were still obtained [5]. (1) The
appearance of RAS changed as the segmentation threshold was raised from 80 HU
to 130 HU. Stenoses appeared discontinuous, rather than narrowing with higher
threshold values. (2) All five accessory renal arteries seen on conventional
arteriography were also identified by CT angiography. But even choosing the
``ideal'' SSD threshold value, as determined by the main renal artery
appearance on SSD versus conventional arteriography, artifactual stenoses were
suggested in each of these accessory renal arteries. (3) SSD could show 3D
images or objects only when they are not covered by other objects. This is one
of the principal drawbacks, inability to distinguish objects with different CT
density (i.e., vascular calcification from the enhanced vessel lumen). (4) The
appearance of stenoses of main renal artery and accessory renal arteries on
the SSD image depends on the threshold value chosen. Hence, there are no
standard criteria for choosing the threshold value, and different physicians
rendering an SSD image from the same data may produce different grades of
stenosis because of different threshold values. (5) Even though cited paper
presents a table concerning the relation between Spearman correlation
coefficients (r
) and ideal threshold value of visceral organ enhancement,
still it can not provide an accurate predictor of threshold value.
Rather than pick threshold values used in the SSD technique, Maximum Intensity Projection uses a connectivity algorithm, which need not choose threshold values, to create an image.