TERSE: Illustrative Results

This page illustrates the type of image reconstruction results that one can obtain using the statistical methods available in the TERSE software package.

We are currently using a Picker PRISM 3000 triple-head SPECT system for cardiac scans, with an Americium transmission source opposing a fan-beam collimator, with parallel-beam collimators on the other two heads. All three heads collection emission data.

One key component of our approach is that we use regularized methods to reconstruct the attenuation maps from the fan-beam transmission scans. The regularization helps compensate for the missing transmission data. The following transmission images are of a normal volunteer (see 1996 JNM paper.) The top row of images (attenuation maps) is from FBP reconstruction; the ring artifact is due to fan-beam truncation. The middle row of images was reconstructed using a penalized weighted least-squares method from the fan-beam transmission data. Note that the ring artifact is eliminated. The bottom row of images was acquired subsequently on a Siemens Multispect system with an offset line source (and hence no truncation) and reconstructed by FBP. The images shown were manually selected to correspond to approximately the same slice levels. Note that the TERSE software has done a reasonable job of "filling in" the truncated region.


The following images (taken from Fig. 12 of our 1996 Circulation paper) illustrate the improved uniformity obtained in images reconstructed with compensation for nonuniform attenuation.

Notice that in the leftmost "uncorrected" image the "donut" (the myocardium) is not uniformly red; the decreased red at the bottom could be interpreted as cardiac disease since it suggests that part of the heart has inadequate blood supply. In fact, the In comparison, the leftmost image on the bottom, reconstructed by the attenuation correction method that is used routinely at the UM hospital, is nearly uniformly red all around the myocardium, suggesting that the patient's blood supply is normal, which indeed is the case in this patient as was confirmed by X-ray angiography.
Back to Fessler home page