A New automatic Method for Producing Maximum Intensity Venograms From Pre and Post Gd 3D Data

Several methods are available to produce cerebral MR venograms including time of flight, phase contrast, BOLD venography and MP-RAGE subtraction venography. Each method has advantages and disadvantages over the others. One of the major disadvantages of the MP-RAGE subtraction technique is the need for segmentation of extracerebral tissue to produce a venogram with good signal to noise. This is required to remove other structures within the head which enhance with Gd, including skin, nasal sinuses, and extracerebral vessels. Segmentation is further hindered by the need to include enhancing veins as well as cerebral tissue in the segmented subtraction data but not in the data used for data registration, which corrects misaligned data. All the currently available cerebral segmentation algorithms are designed to include brain only. Thus, the only method of producing these venograms requires manual segmentation, which is skilled and time consuming. Therefore, a new automatic segmentation, registration and subtraction technique has been established which can easily be implemented using morphological operators to produce cerebral venograms. The output from this was compared to manually segmented and registered data. Signal to noise was lower using the automatic method, however, all vessels and pathology were visible. The automatic method was substantially faster than the manual method.