Lara da Rocha Vaz Pato

Characterization of a SPECT pinhole collimator for optimal detector usage (the lofthole)

In single-photon emission computed tomography (SPECT), multi-pinhole collimation is often employed nowadays. Most multi-pinhole collimators avoid overlap (multiplexing) of the projections on the detector. This can be done by using additional shielding or by spacing the pinholes far enough apart. Using additional shielding has the drawback that it increases weight, design complexity and cost. Spacing the pinholes far enough apart results in sub-optimal detector usage, the valuable detector area is not entirely used. This is due to the circular projections of pinholes on the detector; these ellipses can not be tiled with high detector coverage. To overcome this we designed a new pinhole geometry, the lofthole, that has a rectangular projection on the detector. The lofthole has a circular aperture and a rectangular entrance/exit opening. Sensitivity formulae have been derived for pinholes and loftholes. These formulae take the penumbra effect into account; the proposed formulae do not take penetration into account. The derived formulae are valid for geometries where the field-of-view and the sensitivity of the aperture are solely limited by the exit window. A flood map measurement was performed to compare the rectangular projection of a lofthole with the circular projection of a pinhole. Finally, measurements were done to compare the amount of penetration of pinholes with the amount of penetration of a lofthole. A square lofthole collimator has less penetration than a knife-edge pinhole collimator that irradiates the same rectangular detector area with full coverage. A multi-lofthole collimator allows high detector coverage without using additional shielding. An additional advantage is the lower amount of penetration.

Evaluation of Fisher Information Matrix-Based Methods for Fast Assessment of Image Quality in Pinhole SPECT

The accurate determination of the local impulse response and the covariance in voxels from penalized maximum likelihood reconstructed images requires performing reconstructions from many noise realizations of the projection data. As this is usually a very time-consuming process, efficient analytical approximations based on the Fisher information matrix (FIM) have been extensively used in PET and SPECT to estimate these quantities. For 3D imaging, however, additional approximations need to be made to the FIM in order to speed up the calculations. The most common approach is to use the local shift-invariant (LSI) approximation of the FIM, but this assumes specific conditions which are not always necessarily valid. In this paper we take a single-pinhole SPECT system and compare the accuracy of the LSI approximation against two other methods that have been more recently put forward: the non-uniform object-space pixelation (NUOP) and the subsampled FIM. These methods do not assume such restrictive conditions while still increasing the speed of the calculations considerably. Our results indicate that in pinhole SPECT the NUOP and subsampled FIM approaches could be more reliable than the LSI approximation, especially when a high accuracy is required.

Evaluation of the local shift-invariance approximation in pinhole SPECT

The local shift-invariance approximation of the Fisher information matrix is commonly used in 3D PET and SPECT to speed up the analytical estimation of the contrast recovery coefficient and variance at voxels of interest in a reconstructed image. However, the approximation should only be used under specific conditions, which are usually assumed to be valid. In this work we investigate the two main assumptions behind this approach in a single-pinhole SPECT system, by analyzing Fisher Information Matrix column images. Our results show that the first assumption (localness) is less applicable when low angular sampling is used, and consequently also for highly non-uniform sampling strategies. With the second assumption (local shift-invariance), on the other hand, there seem to be issues in any angular sampling strategy. This indicates that the local shift-invariance approach might not be applicable in many other situations, even for very simple systems and phantoms, particularly in pinhole SPECT. Our findings are mainly intended to show that the local shift-invariance approximation should be used with care. One should always make sure that the assumptions made, as well as any additional approximations, are valid for a particular imaging system, phantom and protocol.

Efficient optimization for adaptive SPECT systems based on local shift-invariance

Adaptive SPECT systems automatically change some of their settings to maximize the image quality for a given subject and purpose. This approach has a lot of potential, and could lead to drastic improvements in performance. In particular, it would be very useful in high resolution pinhole SPECT, where the low sensitivity requires higher radiation doses or longer imaging times compared to other systems. In order to have adaptation in real-time, we need a fast method for optimizing the adaptive settings according to a given figure of merit. This is still a big challenge. Based on previous work, we address in this paper the issue of fast evaluation of image quality and optimization, for a class of adaptive SPECT systems. We evaluate the image quality in a voxel of interest, reconstructed using postfiltered MLEM, with the contrast-to-noise ratio (CNR). The CNR is computed analytically, using an approximation based on the Fisher information matrix and assuming local shift-invariance on the Fisher information matrices per adaptation parameter. We maximize the CNR with a gradient based optimization approach. We then test this method in the optimization of the angular sampling of a single-head SPECT system which rotates around a phantom. In this case, the method proved to be very efficient, and at the same time showed good agreement with previous results in literature and with the outcome from reconstructions.