Showing papers by "David H. Laidlaw published in 1998"

Partial-volume Bayesian classification of material mixtures in MR volume data using voxel histograms

Abstract: The authors present a new algorithm for identifying the distribution of different material types in volumetric datasets such as those produced with magnetic resonance imaging (MRI) or computed tomography (CT). Because the authors allow for mixtures of materials and treat voxels as regions, their technique reduces errors that other classification techniques can create along boundaries between materials and is particularly useful for creating accurate geometric models and renderings from volume data. It also has the potential to make volume measurements more accurately and classifies noisy, low-resolution data well. There are two unusual aspects to the authors' approach. First, they assume that, due to partial-volume effects, or blurring, voxels can contain more than one material, e.g., both muscle and fat; the authors compute the relative proportion of each material in the voxels. Second, they incorporate information from neighboring voxels into the classification process by reconstructing a continuous function, /spl rho/(x), from the samples and then looking at the distribution of values that /spl rho/(x) takes on within the region of a voxel. This distribution of values is represented by a histogram taken over the region of the voxel; the mixture of materials that those values measure is identified within the voxel using a probabilistic Bayesian approach that matches the histogram by finding the mixture of materials within each voxel most likely to have created the histogram. The size of regions that the authors classify is chosen to match the sparing of the samples because the spacing is intrinsically related to the minimum feature size that the reconstructed continuous function can represent.

MR microscopy of transgenic mice that spontaneously acquire experimental allergic encephalomyelitis

Abstract: Pathology of fixed spinal cords from transgenic mice with a myelin basic protein (MBP) specific T cell receptor was investigated. These mice spontaneously acquire the demyelinating disease experimental allergic encephalomyelitis (EAE). Several complementary imaging modalities, all on the same tissues, were used to visualize lesions; these included high-field (11.7-T) microscopic diffusion tensor imaging (DTI), T*_2- image-weighted imaging, and optical microscopy on histological sections. Lesions were predominantly in white matter around meninges and vasculature and appeared hyperintense in anatomical images. DTIs showed reduced diffusion anisotropy in the same hyperintense regions, consistent with inflammation and edema. Histology in the same tissues exhibited the characteristic pathology of EAE. Two techniques for visualizing the effective diffusion tensor fields are presented, which display direction, organization, and integrity of neuronal fibers. It is shown that DTI offers intriguing possibilities for visualizing axonal organization and lesions within white matter.

Visualizing diffusion tensor images of the mouse spinal cord

Abstract: Within biological systems, water molecules undergo continuous stochastic Brownian motion The diffusion rate can give clues to the structure of the underlying tissues In some tissues, the rate is anisotropic Diffusion-rate images can be calculated from diffusion-weighted MRI A 2D diffusion tensor image (DTI) and an associated anatomical scalar field define seven values at each spatial location We present two new methods for visually representing DTIs The first method displays an array of ellipsoids, where the shape of each ellipsoid represents one tensor value The ellipsoids are all normalized to approximately the same size so that they can be displayed simultaneously in context The second method uses concepts from oil painting to represent the seven-valued data with multiple layers of varying brush strokes Both methods successfully display most or all of the information in DTIs and provide exploratory methods for understanding them The ellipsoid method has a simpler interpretation and explanation than the painting-motivated method; the painting-motivated method displays more of the information and is easier to read quantatively We demonstrate the methods on images of the mouse spinal cord The visualizations show significant differences between spinal cords from mice suffering from experimental allergic encephalomyelitis and spinal cords from wild-type mice The differences are consistent with differences shown histologically and suggest that our new non-invasive imaging methodology and visualization of the results could have early diagnostic value for neurodegenerative diseases

Author's reply [to 'Comments on "Partial-volume Bayesian classification of material mixtures in MR volume data using voxel histograms" by Hamid Soltanian-Zadeh and Joe P. Windham']

Abstract: D.H. Laidlaw replies to comments made on his paper (D.H. Laidlaw et al., ibid., vol. 17, no. 1, p. 74-86, 1998) by H. Soltanian-Zadeh and J.P. Windham (ibid., vol. 17, no. 6, p. 1094, 1998). D.H. Laidlaw says that the concept of optimality always rests on a framework of assumptions. Eigenimage filtering is optimal under a certain set of assumptions. However, D.H. Laidlaw et al.'s voxel histogram method uses different assumptions and, as D.H. Laidlaw demonstrates, produces better results with fewer images. In this communication D.H. Laidlaw first discusses some of the assumptions leading to the eigenimage filtering method, then describe how his assumptions differ and how they lead to a different type of classification method. He then presents results comparing the two methods, addresses the four points raised in the communication by H. Soltanian-Zadeh and J.P. Windham (ibid., vol. 17, no. 6, p. 1094, 1998) and states his conclusions.

Visually representing multi-valued scientific data using concepts from painting

Abstract: Vector-valued and tensor-valued images are rich sources of information about many physical phenomena. Visually representing these images so that they can be understood is a challenge, however, because they contain so many inter-related components, all of which must be represented simultaneously and intuitively.