High-Resolution Determination of Soft Tissue Deformations Using MRI and First-Order Texture Correlation

Abstract

Mechanical factors such as deformation and strain are thought to play important roles in the maintenance, repair, and degeneration of soft tissues. Determination of soft tissue static deformation has traditionally only been possible at a tissue's surface, utilizing external markers or instrumentation. Texture correlation is a displacement field measurement technique which relies on unique image patterns within a pair of digital images to track displacement. The technique has recently been applied to MR images, indicating the possibility of high-resolution displacement and strain field determination within the mid-substance of soft tissues. However, the utility of MR texture correlation analysis may vary amongst tissue types depending on their underlying structure, composition, and contrast mechanism, which give rise to variations in texture with MRI. In this study, we investigate the utility of a texture correlation algorithm with first-order displacement mapping terms for use with MR images, and suggest a novel index of image "roughness" as a way to decrease errors associated with the use of texture correlation for intra-tissue strain measurement with MRI. We find that a first-order algorithm can significantly reduce strain measurement error, and that an image "roughness" index correlates with displacement measurement error for a variety of imaging conditions and tissue types.

Measurement of Strain in the Left Ventricle during Diastole with cine-MRI and Deformable Image Registration

“…A realistic representation of the material behavior helps to improve predictions in these areas. Other techniques for image-based strain measurement, such as optical flow [41] and texture correlation [42], use only an image-based energy term. These techniques do not ensure physically reasonable deformations in regions that lack image contrast or texture and they are sensitive to noise [42].…”

Measurement of intracellular strain on deformable substrates with texture correlation

“…Strains were measured using a 2D texture correlation algorithm described previously (Gilchrist et al, 2004). Briefly, texture correlation is a subset of digital image correlation where pixel displacements are determined by comparing intensity patterns between two images, a ''reference'' and ''deformed'' image set (Bay, 1995;Bey et al, 2002;Wang et al, 2003).…”

“…magnitude of cell strains not different from underlying substrate strains) to cytoplasmic and nuclear regions of attached cells over a range of applied substrate stretch. Fibroblast-like cells were isolated from the fibrous anulus region of intervertebral disc tissues and labeled with fluorescent probes specific to mitochondria or nuclei for visualization with confocal microscopy and tracking using two-dimensional (2D) texture correlation (Gilchrist et al, 2004). A secondary objective was to evaluate the dependence of strain transfer on cell alignment relative to the direction of substrate stretch.…”

Measurement of all orthogonal components of displacement in the volume of scattering materials using wavelength scanning interferometry

“…In all these cases, spatial features which are part of the material microstructure are tracked to sub-pixel accuracy. Magnetic resonance imaging (MRI) has also been used to evaluate 3-D deformation inside bones and tendons [16,17] in conjunction with 2-D DIC. An alternative method based on high-resolution 3-D MRI volumes and a surface registration algorithm based on finite element optimization was used to evaluate the growth of cortical surfaces during the folding of cerebral cortex of human and animal brains [18].…”