MRI Slice Picturing

We set up two coordinate systems, one in the object space and the other on the computer screen. We introduce six parameters to describe the slice plane, and we formulate the coordinate mapping from the screen to the object space. We designed six alternative algorithms that use the given data to estimate the density at any location in space and produce a slice of a three-dimensional array. Some of the algorithms exploit global information and some are selfadaptive; all but one have advantages in certain circumstances. We extended a well-known two-dimensional model of a human head model to build a three-dimensional model of a head, consisting of 10 ellipsoids of different size, orientation, and density. We produced the data sets by sampling in the object space (the head model) at evenly spaced intervals; the dimension of the data set is 128× 128× 128. We devised several test slices to test our model and algorithms. Some test slices have a complex shape, some are critical in their position, and some are really disasters to most algorithms. We also tried different sampling intervals to verify our ideas about the model. Based on subjective and objective comparisons, we summarize the strengths and weaknesses of the algorithms. For common use, we suggest the gradient algorithm and our GNP-integrated algorithm. In most cases, both can render well slices with both sharp and smooth edges.