Halpern et al. and Rubin et al. have shown that
MIP is more accurate than SSD in the grading of stenosis [5],
[1]. The advantage of MIP is that the gray scale reflects CT
attenuation rather than simulating light reflection as in SSD. Therefore,
objects of differing attenuation could be differentiated. However, MIP still
has the following problems:
(1) MIP does not convey depth relationships, nor does it depict overlapping vessels as a focal increase in opacity. (2) Partial volume effects tends to result in vessel discontinuity at the area of high grade stenosis, rather than showing extreme narrowing as seen with conventional arteriography. (3) Partial volume or beam hardening may cause gradual attenuation transitions instead of a steep attenuation transition between adjacent structures, and MIP may include other structures one wants to suppress. For example, the vessel may appear to be physically attached to the skull in the petrous carotid region, and if too low a threshold value is used, then the image may falsely include the iodine-filled vessels as part of the skull. For another example, the descending aorta lies against a portion of the spinal column, and therefore a calcification inside the vessel lumen may not be visualized as being separate from the bone. In these cases, one is forced to choose a threshold well above the attenuation of the contrast medium in the vessel, and this results in risking leaving bone edges unsuppressed. (4) Mural calcium can also interfere with the grading of stenoses by means of MIP CT angiography. Contrary to the overestimation of luminal patency with SSD, stenoses can be overestimated with MIP if an eccentrically located calcification overlies the luminal contrast material. In this case, multiple MIPs must be generated about both the x and z axes to ensure that a projection is rendered where the calcium is adjacent to the patent lumen, rather than overlying it.