The overwhelming amount of image-based information in modem medicine makes it crucial to develop methods to handle and analyze images and make them comprehensible for users. The aim of this thesis was to study the radiological practice of three-dimensional (3D) visualization of tubular structures in the body with CT and MRI. All the studies cancern 3D imaging of tubular structures with camputed tomography (CT) and magnetic resonance imaging (MRI). The first three studies examine the abdominal aorta; the two later ones the, bile ducts.

Study I compared measurements of aorta diameters taken from MR images presented using two visualization methods - maximum intensity projection (MlP) and volume rendering (VRT) - with invasive angiography (DSA) and CT as reference methods. Mean diameters of MR images were smaller than those from DSA and CT when MlP was used, but in general not when VRT was used.

Study II evaluated the dependence on the observer and the choice of method and settings during rendering using the same material as in Study 1. In both MlP and VRT, the choice of settings had significant influence on the results. With DSA as the reference method, VRT gave larger measurement errors than MIP when the rendering parameters were set to fixed values, but not if the user was allowed to select the settings freely.

Study III evaluated three new techniques for standardizing VRT protocols for MRA. Inter-reader variability and agreement with DSA were studied by comparing diameter measurements of the abdominal aorta obtained by the three new techniques, by VRT with freely chosen parameters and by MlP. All three new methods were significantly better than MlP and VRT with freely chosen parameters conceming inter-observer agreement. Agreement with DSA was significantly better for one of the methods. Standardized protocols seem to have a potential to make VRT a clinically useful alternative to MlP for MR angiography measurements.

Study IV evaluated CT imaging of the bile ducts after drip intravenous infusion of the contrast medium iotroxate (CT cholangiography) in terms of adverse effects and visibility. With infusion time adjusted for individual variation in serum bilirubin concentration, a total side-effect frequency of less than 1% was found. A systematic review of previously published studies indicated a frequency of 2.3%. Good contrast excretion and visualization of bile ducts even in patients with elevated bilirubin levels were noted.

Study V evaluated the diagnostic benefits of the same imaging method by comparing it with findings from surgery and endoscopic retrograde cholangiopancreatography (ERCP). The consensus sensitivity and specificity for diagnosing biliary stones was 88% and 94%, respectively. The use of VRT improved diagnostic certainty in 14% of the evaluatians, and the visualization of ductal stones was improved in 38% of the positive cases.

In conclusion, volume rendering technique with standardized parameters may become a clinically useful tool in the clinical MRI environment. DIC-CT with bilirubin-governed infusion time and volume rendering post-processing produces detailed images of the biliary tree, resulting in good sensitivity and specificity. Moreover the safety is acceptable.