David L. Wilson

The authors are using a perceptual difference model (Case‐PDM) to quantitatively evaluate image quality of the thousands of test images which can be created when optimizing fast magnetic resonance (MR) imaging strategies and reconstruction techniques. In this validation study, they compared human evaluation of MR images from multiple organs and from multiple image reconstruction algorithms to Case‐PDM and similar models. The authors found that Case‐PDM compared very favorably to human observers in double‐stimulus continuous‐quality scale and functional measurement theory studies over a large range of image quality. The Case‐PDM threshold for nonperceptible differences in a 2‐alternative forced choice study varied with the type of image under study, but wasfor diffuse image effects, providing a rule of thumb. Ordering the image quality evaluation models, we found in overall Case‐PDMIDM (Sarnoff Corporation)SSIM [Wang et al. IEEE Trans. Image Process. 13, 600–612 (2004)]mean squared errorNR [Wang et al. (2004) (unpublished)]DCTune (NASA)IQM (MITRE Corporation). The authors conclude that Case‐PDM is very useful in MR image evaluation but that one should probably restrict studies to similar images and similar processing, normally not a limitation in image reconstruction studies.

Purpose: To develop and optimize a new modification of GRAPPA (generalized autocalibrating partially parallel acquisitions) MR reconstruction algorithm named “Robust GRAPPA.”Materials and Methods: In Robust GRAPPA, k‐space data points were weighted before the reconstruction. Small or zero weights were assigned to “outliers” in k‐space. We implemented a Slow Robust GRAPPA method, which iteratively reweighted the k‐space data. It was compared to an ad hoc Fast Robust GRAPPA method, which eliminated (assigned zero weights to) a fixed percentage of k‐space “outliers” following an initial estimation procedure. In comprehensive experiments the new algorithms were evaluated using the perceptual difference model (PDM), whereby image quality was quantitatively compared to the reference image. Independent variables included algorithm type, total reduction factor, outlier ratio, center filling options, and noise across multiple image datasets, providing 10,800 test images for evaluation.Results: The Fast Robust GRAPPA method gave results very similar to Slow Robust GRAPPA, and showed significant improvements as compared to regular GRAPPA. Fast Robust GRAPPA added little computation time compared with regular GRAPPA.Conclusion: Robust GRAPPA was proposed and proved useful for improving the reconstructed image quality. PDM was helpful in designing and optimizing the MR reconstruction algorithms. J. Magn. Reson. Imaging 2008;27:1412–1420.

We developed a three-dimensional (3D) registration method to align medical scanner data with histological sections. After acquiring 3D medical scanner images, we sliced and photographed the tissue using, a custom apparatus, to obtain a volume of tissue section images. Histological samples from the sections were digitized using a video microscopy system. We aligned the histology and medical images to the reference tissue images using our 3D registration method. We applied the method to correlate in vivo magnetic resonance (MR) and histological measurements for radio-frequency thermal ablation lesions in rabbit thighs. For registration evaluation, we used an ellipsoid model to describe the lesion surfaces. The model surface closely fit the inner (M1) and outer (M2) boundaries of the hyperintense region in MR lesion images, and the boundary of necrosis (H1) in registered histology images. We used the distance between the model surfaces to indicate the 3D registration error. For four experiments, we measured a registration accuracy of 0.96± 0.13 mm (mean±SD) from the absolute distance between the M2 and H1 model surfaces, which compares favorably to the 0.70 mm in-plane MR voxel dimension. This suggests that our registration method provides sufficient spatial correspondence to correlate 3D medical scanner and histology data.

Purpose: To investigate the hypothesis that the outer boundary of the hyperintense region observed in hyperacute (several minutes post‐ablation) T2 and gadolinium contrast‐enhanced (CE) T1‐weighted magnetic resonance (MR) lesion images is an accurate predictor of eventual cell death from radiofrequency (RF) thermal ablation.Materials and Methods: A low‐field, open MR imaging system was used to guide an ablation electrode into a thigh muscle of five rabbits and acquire in vivo T2 and CE T1‐weighted MR volumes. Ablation occurred by applying RF current for two minutes with the electrode's temperature maintained at 90° ± 2°C. After fixation, we sliced and photographed the tissue at 3 mm intervals, using a specially designed apparatus, to obtain a volume of tissue images. Digital images of hematoxylin and eosin (H&E) and Masson trichrome–stained histologic samples were obtained, and distinct regions of tissue damage were labeled using a video microscopy system. After the MR and histology images were aligned using a three‐dimensional registration method, we compared tissue damage boundaries identified in histology with boundaries marked in MR images.Results: The lesions have distinct zones of tissue damage histologically: a central zone of necrotic cells surrounded by an outer zone with cells that appeared non‐viable and associated with marked interstitial edema. In 14 histology images from five lesions, the inner and outer boundaries of the outer zone were compared with the boundaries of a hyperintense rim that surrounds a central hypointense region in the T2 and CE T1‐weighted MR images. For T2 and CE T1‐weighted MR images, respectively, the mean absolute distance was 1.04 ± 0.30 mm (mean ± SD) and 1.00 ± 0.34 mm for the inner boundaries, and 0.96 ± 0.34 mm and 0.94 ± 0.44 mm for the outer boundaries. The mean absolute distances for T2 and CE T1‐weighted MR images were not sufficiently different to achieve statistical significance (P = 0.745, 0.818, for the inner and outer boundary, respectively).Conclusion: In hyperacute T2 and CE T1‐weighted MR lesion images, observations strongly suggest that the outer boundary of the hyperintense rim corresponds to the region of eventual cell necrosis within a distance comparable to our ability to measure. This is good evidence that during RF ablation procedures, MR lesion images can be used to accurately localize the zone of irreversible tissue damage at the lesion margin. J. Magn. Reson. Imaging 2004;20:475–486.

Solid tumors can be treated using radiofrequency (RF) thermal ablation under interventional magnetic resonance (MR) image guidance. We are investigating the ability of MR to accurately predict the region of cell death by comparing MR thermal lesion images to tissue damage as seen histologically. We developed a methodology using a three-dimensional computer registration for making spatial correlations. An open MR imaging system was used to guide an ablation electrode into rabbit thigh muscles and acquire in vivo post-ablation MR volumes. After MR and histology images were aligned with a registration accuracy of 1.32/spl plusmn/0.39 mm (mean/spl plusmn/SD), we compared a boundary of necrosis identified in histology with the outer boundary of the hyperintense rim in gadolinium contrast-enhanced (CE) Tl-weighted MR images. For 14 histology images from five lesions, the mean absolute distance between boundaries was 0.94/spl plusmn/0.44 mm. Since the small discrepancy is comparable to our ability to measure such errors, these boundaries may match exactly. Similar correlations to histology were obtained from experiments with rabbits sacrificed four days post-ablation. This is good evidence that during RF ablation treatments, MR images can accurately localize the zone of necrosis at the lesion margin.

Purpose To evaluate MR signal and lesion zone volume evolution through the sub‐acute phase following image‐guided radiofrequency (RF) thermal ablation. Materials and Methods For many tissues, including muscle and liver, thermal lesions that result from RF heating have a characteristic two‐boundary appearance featuring an inner core (zone I) surrounded by a hyper‐intense margin (zone II) and normal tissue (zone III), found in both T₂ and contrast enhanced (CE) T₁‐weighted MR images, both immediately post‐ablation and four days later. First, we compared corresponding points between manually segmented zone boundaries apparent in T₂‐ and CE T₁‐weighted images. Second, we examined the contrast‐to‐noise ratio (CNR) between all zone combinations. Third, we quantified the volume of zone I, zone II, and the entire lesion using a three‐dimensional lesion geometry model fitted to segmented images. Results On a slice‐by‐slice basis, no statistically significant differences were found between zone boundaries apparent in T₂ and CE T₁‐weighted images. The contrast to noise ratio (CNR) of zone I vs. zone II, zone I vs. background muscle, and zone II vs. background muscle was always equal or greater for T₂‐weighted images than for CE T₁‐weighted images. In addition, by day four, zone II significantly increased in intensity compared to background muscle. The median Zone I volume increase was 44.2% (42.6%) using T₂ weighted images and 55.5% (68.7% interquartile range) using CE T₁‐ weighted images. This expansion likely corresponds to an enlargement of the ablated, coagulative necrosis, region. The median Zone II volume increase was 15.0% (42.6%) using T₂‐ weighted images 1.5% (38.8%) using CE T₁‐weighted images. Conclusions 1) There are no significant differences between the apparent zone boundaries in T₂‐ and CE T₁‐weighted images; 2) CNR is equal or greater for T₂‐weighted images as compared to CE T₁‐images; and 3) both the inner and outer lesion zone volumes typically increase several days post‐ablation. J. Magn. Reson. Imaging 2003;18:353–359.

Spatial unsharp-mask processing and its variants are commonly used in x-ray radiography to enhance image contrast. We investigated the effect of three unsharp-masking filter kernels of different sizes on the detection of an advanced guidewire tip in simulated x-ray fluoroscopy image sequences. To isolate the effect of visual temporal processing, we repeated the experiments on single images. Filter gains were selected so that all three kernels increased the contrast of a 0.018-in. (0.457-mm) guidewire by a factor of 2 but had different effects on image noise and signal profiles. There was no statistically significant effect of unsharp masking on human-observer performance in single images. However, all three kernels significantly improved average performance in image sequences, and the guidewire contrast required for detection was reduced by 32%–40%. A prewhitening channelized observer model predicted the disparity between sequences and single images and fitted measurements at different kernel sizes well. A nonprewhitening observer model did not. We conclude that unsharp masking is a simple and effective method of improving guidewire visualization in fluoroscopically guided interventional procedures and that quantitative image quality studies are essential for evaluation of image-processing techniques in sequences such as x-ray fluoroscopy.