Background: Maximum intensity projection (MIP) is a volume rendering technique that determines the pixel intensity as the maximum of all values sampled along the viewing direction. MIP has been successfully applied to diagnose bone fractures in computed tomography (CT) and the stenosis of vascular structures in magnetic resonance angiography (MRA). However, MIP has a major drawback in that the depth and occlusion information cannot be perceived in the output images. The most universal way to alleviate this problem is to occasionally change the viewpoint for depth perception. To support this function in real time, MIP should be performed at an interactive frame rate.
Objective: We develop an efficient rendering algorithm for MIP so that MIP is performed at an interactive frame rate without a loss of image quality.
Methods: The proposed method predicts the position of the maximum intensity for each ray using blockwise maximum bounds, after which it performs bidirectional compositing toward both ends of the ray from this predicted position. During the compositing process, block skipping is used as an acceleration method.
Results: The proposed method outperformed the block skipping method using the sequential compositing with a speed-up factor of 2.2 ∼ 2.8 depending on the data set without any degradation of the image quality.
Conclusion: We proposed an efficient rendering technique for MIP. Our method was superior to the conventional block skipping method with respect to the rendering speed and degree of performance consistency.
Keywords: Maximum intensity projection; acceleration techniques; bidirectional compositing; block skipping; prediction of the starting position.