Affine adaptive filtering of CT data

Westin, Carl‐Fredrik; Richolt, Jens A.; Moharir, Vik M.; Kikinis, Ron

doi:10.1016/s1361-8415(00)00011-6

Cited by 40 publications

(27 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In three dimensions, local structure has been used to describe landmarks, rotational symmetries, and motion [7,3,8,12,13]. In addition to isotropy, it describes geometrical properties which have been used to guide the enhancement and segmentation of blood vessels in volumetric angiography datasets [10,15], bone in CT images [18], and to the analysis of white matter in DTI [17].…”

Section: Local Image Structurementioning

confidence: 99%

Interface Detection in Diffusion Tensor MRI

O’Donnell

Grimson²,

Westin

2004

Medical Image Computing and Computer-Assisted Intervention – MICCAI 2004

View full text Add to dashboard Cite

Abstract. We present a new method for detecting the interface, or edge, structure present in diffusion MRI. Interface detection is an important first step for applications including segmentation and registration. Additionally, due to the higher dimensionality of tensor data, humans are visually unable to detect edges as easily as in scalar data, so edge detection has potential applications in diffusion tensor visualization. Our method employs the computer vision techniques of local structure filtering and normalized convolution. We detect the edges in the tensor field by calculating a generalized local structure tensor, based on the sum of the outer products of the gradients of the tensor components. The local structure tensor provides a rotationally invariant description of edge orientation, and its shape after local averaging describes the type of edge. We demonstrate the ability to detect not only edges caused by differences in tensor magnitude, but also edges between regions of different tensor shape. We demonstrate the method's performance on synthetic data, on major fiber tract boundaries, and in one gray matter region.

show abstract

Section: Local Image Structurementioning

confidence: 99%

Interface Detection in Diffusion Tensor MRI

O’Donnell

Grimson²,

Westin

2004

Medical Image Computing and Computer-Assisted Intervention – MICCAI 2004

View full text Add to dashboard Cite

show abstract

“…They are defined as the dual tensor basis to the tensor basis defined by the outer products of these filter directions, {n kn T k }. More details about how to calculate the elements M k can be found in [8]. In order to evenly cover the local spectrum of the dataset, Knutsson proposed to use filter directions defined by the vertexes of the hexagon in 2D, and the icosahedron in 3D.…”

Section: Local Structure Estimation In Vector and Tensor Datamentioning

confidence: 99%

Anisotropic Interpolation of DT-MRI

Castaño-Moraga

Rodríguez-Florido

Álvarez

et al. 2004

Medical Image Computing and Computer-Assisted Intervention – MICCAI 2004

View full text Add to dashboard Cite

Abstract. Diffusion tensor MRI (DT-MRI)is an image modality that is gaining clinical importance. After some preliminaries that describe the fundamentals of this imaging modality, we present a new technique to interpolate the diffusion tensor field, preserving boundaries and the constraint of positive-semidefiniteness. Our approach is based on the use of the inverse of the local structure tensor as a metric to compute the distance between the samples and the interpolation point. Interpolation is an essential step in managing digital image data for many different applications. Results on resampling (zooming) synthetic and clinical DT-MRI data are used to illustrate the technique. This new scheme provides a new framework for anisotropic image processing, including applications on filtering, registration or segmentation.

show abstract

“…An orientation estimation algorithm is employed to obtain information about local structures (24,(27)(28)(29)(30)(31). A suitable vector representation for orientation in 2D has been described previously (25,26).…”

Section: Orientation Estimationmentioning

confidence: 99%

Three‐dimensional adaptive filtering in magnetic resonance angiography

Westin

Wigström

Loock

et al. 2001

Magnetic Resonance Imaging

View full text Add to dashboard Cite

In order to enhance 3D image data from magnetic resonance angiography (MRA), a novel method based on the theory of multidimensional adaptive filtering has been developed. The purpose of the technique is to suppress image noise while enhancing important structures. The method is based on local structure estimation using six 3D orientation selective filters, followed by an adaptive filtering step controlled by the local structure information. The complete filtering procedure requires approximately 3 minutes of computational time on a standard workstation for a 256 ؋ 256 ؋ 64 data set. The method has been evaluated using a mathematical vessel model and in vivo MRA data (both phase contrast and time of flight (TOF)). 3D adaptive filtering results in a better delineation of small blood vessels and efficiently reduces the high-frequency noise. Depending on the data acquisition and the original data type, contrast-to-noise ratio (CNR) improvements of up to 179% (8.9 dB) were observed. 3D adaptive filtering may provide an alternative to prolonging the scan time or using contrast agents in MRA when the CNR is low. J. Index terms: angiography; adaptive filtering; noise reduction; tensor; image enhancement MOST MAGNETIC RESONANCE ANGIOGRAPHY (MRA) techniques produce a 3D data set, which is processed to selectively display the vasculature of interest. Angiograms based on both phase contrast (1) and time of flight (TOF) (2) can be reconstructed from the 3D data sets using maximum intensity projection (MIP), which creates a 2D image by recording the intensity of the brightest voxel along each projection ray through the 3D volume (3). The MIP technique is sensitive to highintensity values that tend to obscure regions of slow or recirculating flow, is unable to separately identify arterial and venous structures within the same data, and cannot discriminate high-intensity artifacts from true vessel signal (4). A high-resolution data set acquired while maintaining a short scan time often has a noise level high enough to obscure important diagnostic information in the angiograms. Several methods for reduction of noise in MRA data have been published (5-19). Song et al. presented a method to reduce the noise in 3D phase-contrast MRA (10). The method exploits the fact that the velocity field is divergence-free. The noise contribution, which is not divergence-free, can be removed using a projection operator method in Hilbert space. Du and coworkers showed that 3D TOF images could be improved by using a nonlinear second difference spatial filtering technique (11). Comparison with a Laplacian filtering technique demonstrated that the contrast-to-noise ratio (CNR) increased when the nonlinear second difference spatial method was employed. Reducing noise without blurring the lines and edges encourages intraregion smoothing in preference to smoothing across regions. Perona and Malik developed a multiscale smoothing and edge detection scheme, mathematically formulated as an anisotropic diffusion process that effectively performed intraregi...

show abstract

Affine adaptive filtering of CT data

Cited by 40 publications

References 40 publications

Interface Detection in Diffusion Tensor MRI

Interface Detection in Diffusion Tensor MRI

Anisotropic Interpolation of DT-MRI

Three‐dimensional adaptive filtering in magnetic resonance angiography

Contact Info

Product

Resources

About