Imaging bile duct tumors: pathologic concepts, classification, and early tumor detection
Background: To compare fat-suppressed three-dimensional gradient echo (3D-GRE) and two-dimensional gradient echo (2D-GRE) techniques for hepatic magnetic resonance imaging. Methods: For 107 patients with suspected focal hepatic lesions, unenhanced 2D-GRE multislice in-phase and opposed-phase images and 3D-GRE images were obtained. Two radiologists independently and in random order reviewed the images. Quantitative analysis compared contrast-to-noise ratio (C/N) of liver masses and of the liver to spleen and signal-to-noise ratio (S/N) of the liver. Subjective grading of lesion conspicuity, confidence for lesion detection, artifacts, and overall image quality were also evaluated. Results: Conspicuity and confidence for focal lesions was statistically significantly higher on 3D-GRE than on 2D-GRE images. Liver-to-spleen C/N was higher on 3D-GRE images than on in-phase and opposed-phase 2D-GRE images (p < 0.01). Liver S/N was higher on in-phase 2D-GRE images than on 3D-GRE and opposed-phase 2D-GRE images (p < 0.001). Lesion-to-liver C/N was higher on 3D-GRE images than on 2D-GRE images for nonsolid lesions (p < 0.05), but there was no significant difference for solid lesions. Motion and pulsation artifacts were significantly fewer on 3D-GRE images, but graininess and shading were significantly higher. Overall image quality was not significantly different for the three sequences. Conclusion: Fat-suppressed 3D-GRE sequences can afford comparable or better image quality than 2D-GRE images and may have the potential to detect more focal hepatic lesions.