Image analysis – Applications – Biomedical applications
Reexamination Certificate
2001-03-28
2004-08-10
Johns, Andrew W. (Department: 2621)
Image analysis
Applications
Biomedical applications
Reexamination Certificate
active
06775401
ABSTRACT:
FIELD OF THE INVENTION
This invention relates generally to the field of post-processing of images, such as trabecular bone images, obtained by MRI, CT, or other image technologies to increase apparent image resolution without partial volume blurring for structural dimensions in which image voxel size is larger than the typical structural element to be resolved, and to the use of these parameters to determine trabecular network strength as a risk factor in the assessment of osteoporosis.
BACKGROUND OF THE INVENTION
Magnetic Resonance Imaging (MRI) is a powerful tool for noninvasively quantifying tissue morphology. However, when voxel size is larger than the structures of interest, partial volume blurring complicates accurate measurement of structural parameters. On the other hand, acquisition of images at higher resolution often exacts an unacceptable signal-to-noise penalty, and thus, does not represent a viable alternative. For example, these constraints have been the major obstacle to the development of MRI as a means of quantifying trabecular bone architecture in vivo for the purpose of predicting fracture risk in osteopenic subjects (Link et al.,
J. Bone Miner. Res
. 13:1175-1182 (1998); Wehrli et al.,
Radiology
206:347-357 (1998); Majumdar et al.,
J. Bone Miner. Res
. 12:111-118 (1997); Gordon et al.,
Med. Phys
. 24:585-593 (1997)).
Trabecular bone (TB) (also known as cancellous bone), which makes up most of the axial skeleton and ends near the joints of the long bones, consists of a lattice of interconnected plates and rods that confer mechanical strength to the skeleton at minimum weight. However, there is growing evidence that in addition to the volume fraction of the bone (often quantified in terms of bone density), the three-dimensional (3D) arrangement of the trabecular network is a major determinant of elastic modulus and ultimate strength.
In general, characterization of the strength of trabecular lattices from three-dimensional (3D) images can be divided into three major categories: material, scale and topology (DeHoff et al.,
J. Microscopy
95:69-91 (1972)). ‘Material properties’ describe the bone material; ‘scale properties’ describe the size and thickness (local volume properties) of the trabecular elements; and the ‘topological properties’ describe the spatial arrangement of the bone material in the network. These parameters change characteristically with subject age.
A common diagnostic screening method for osteoporosis is based on ‘dual-energy X-ray absorptiometry’ (DEXA) (Wahner et al.,
The Evaluation of Osteoporosis: Dual Energy X
-
Ray Absorptiometry in Clinical Practice
, Cambridge: University Press, 1994) to measure integral bone mineral density (BMD). This method, however, does not distinguish between trabecular and cortical bone and ignores the role of structure as a contributor to mechanical competence.
Since trabecular thickness (80-150 &mgr;m) is typically less than the achievable voxel size in vivo (~150 &mgr;m), accurate structural information is difficult to obtain. The common approach toward quantifying trabecular structure has been to classify voxels as either “bone” or “marrow” via binary segmentation. In the low spatial-resolution regime, however, the “bone” voxels contain varying amounts of bone, usually with a higher proportion of marrow. Therefore, to avoid the loss of information inherent in binary classification, a means of estimating the bone volume fraction (BVF) in each voxel (Hwang et al.,
Int. J Imaging Syst. Technol
. 10:186-198 (1999)) was devised, referred to as BVF mapping. Linear interpolation has commonly been applied to increase the apparent resolution of digital images. In one dimension, for example, BVF at a spatial location between the centers of two adjacent voxels would be computed as the average of the two voxels. As a result, additional values calculated in this manner can never increase beyond the original values, and thus, contradict the axiom that smaller voxels are more likely to contain larger fractions of bone.
Clearly, there remains a need in the art to increase the apparent resolution of the BVF map, and to overcome the partial volume blurring which presently precludes accurate measurement of structural dimensions in the limited-resolution regime in which image voxel size is larger than the typical structural element to be resolved. Since acquiring images at increased resolution often exacts an unacceptable signal-to-noise penalty, methods to alleviate the adverse effects of partial volume blurring are instrumental for the accurate measurement of architectural parameters in applications, such as predicting the mechanical competence of trabecular bone networks. Once they have been accurately measured, parameters for quantifying the visible differences in architecture have recently been derived and applied to predict fracture risk (Wehrli et al.,
J. Bone Miner. Res
. (2001) in press).
Prior to the present invention, even with the most advanced technology, it has not been possible to obtain images of trabecular architecture in vivo at a resolution better than trabecular thickness. In the distal radius, voxel sizes reported by MR range from 5.5×10
−3
mm
3
(Song et al,
Magn. Reson. Med
. 41:947-953 (1999)) to 12×10
−3
mm
3
(Majumdar et al., 1997), corresponding to a mean linear resolution of 187 and 257 &mgr;m, respectively. In peripheral quantitative computed tomography (p-QCT) of the wrist, the smallest reported voxel size is 160 &mgr;m (Laib et al.,
Technol. Health Care
6:329-337 (1998)). However, the point-spread function is considerably wider in CT than in MR, resulting in an effective resolution closer to 300 &mgr;m. Realizing the limitations of image resolution, Majumdar et al., 1997 refer to the derived histomorphometric measures as “apparent.” Müller et al. demonstrated excellent serial reproducibility in histomorphometric parameters obtained in vivo in the distal radius of test subjects by p-QCT (Müller et al.,
J. Bone Miner. Res
. 11: 1745-1750 (1996)), even though the reported trabecular thickness values were overestimated by at least a factor of 2. While apparent histomorphometric measures may still be useful as long as they track true variations in these parameters, analytical approaches aimed at deriving topological parameters (Pothuaud et al.,
J. Microse
. 199:149-161 (2000); Saha et al.,
Int. J. Imaging Sys. Technol
. 11:81-90 (2000); Gomberg et al.,
IEEE Trans. Med. Imaging
19:166-174 (2000); Hildebrand et al,
J. Bone Miner. Res
. 14:1167-1174 (1999)) are unlikely to yield meaningful parameters at in vivo resolution without enhancement of the apparent resolution.
SUMMARY OF THE INVENTION
The present invention comprises a novel image post-processing method, system and device for increasing apparent image resolution of images, such as trabecular bone images, obtained by MRI, CT, or other image technologies. Referred to as “subvoxel processing,” the method and system are applicable to volumes of interest containing material phases of two discrete signal intensities.
The accuracy of the method has been demonstrated by nonlimiting example, using a micro-computed tomography (&mgr;-CT) image of human trabecular bone, to show that subvoxel processing is significantly more accurate than trilinear interpolation in decreasing apparent voxel size, especially in the presence of noise. In addition, the effectiveness of the method has been illustrated with MR images of human trabecular bone acquired in vivo, although it is also applicable ex vivo. The subvoxel-processed images were shown to have architectural features characteristic of images acquired at higher spatial resolution.
In a preferred embodiment of the invention, the method is illustrated with images of trabecular bone; however, the algorithm may easily be applied to images of other materials, which may be considered to locally contain only two phases. The method of Bayesian subvoxel classification (Wu et al.,
Magn. Reson. Med
. 31:302-308 (1994)) also divides voxels into subvoxels; however in contrast t
Hwang Scott N.
Wehrli Felix W.
Dilworth Paxson LLP
Johns Andrew W.
McConathy Evelyn H.
Nakhjavan Shervin
The Trustees of the University of Pennsylvania
LandOfFree
Subvoxel processing: a method for reducing partial volume... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Subvoxel processing: a method for reducing partial volume..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Subvoxel processing: a method for reducing partial volume... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3278631