Surgery – Diagnostic testing – Detecting nuclear – electromagnetic – or ultrasonic radiation
Reexamination Certificate
2000-12-28
2003-06-17
Pelham, Joseph (Department: 3742)
Surgery
Diagnostic testing
Detecting nuclear, electromagnetic, or ultrasonic radiation
C382S128000, C345S424000
Reexamination Certificate
active
06580936
ABSTRACT:
This application is based on Japanese Patent Application No. 2000-61244 filed Mar. 6, 2000, the content of which is incorporated hereinto by reference.
BACKGROUND OF THE INVENTION
Field of the Invention
The present invention relates to a coloring method and apparatus for multichannel MRI imaging process that generates a color image from multichannel MRI images.
Medical image diagnosis apparatuses based on CT (Computer Tomography) or MRI (Magnetic Resonance Imaging) can measure the three-dimensional form of a living body at an accuracy in the order of millimeters. Luminance information in these images appropriately reflects differences among living tissues. In the CT, the luminance information indicates an absorption coefficient for X rays, while in the MRI, it indicates amounts relating to magnetic relaxation time for nuclear spin. In particular, the MRI luminance information shows various contrasts depending on how radio pulses are applied during imaging and is thus effective on the diagnosis of living tissues. Images such as T1 (vertical relaxation time) weighted images, T2 (horizontal relaxation time) weighted images, or proton density weighted images which are often scanned in clinical fields are distinguished based on the magnitude of the repetition time (TR) or echo time (TE) of a pulse sequence observed during MRI imaging based on the spin echo method (see FIG.
1
).
The luminance information in these images is not observed only as a pure physical amount such as the spin relaxation time but as a mixture of several factors and thus varies even with a difference in magnetic field intensity among MRI apparatuses. That is, there is a rough rule of thumb that an image with a higher proton density is obtained by increasing the repetition time TR while reducing the echo time TE, but the TR or TE value for diagnosis is not strictly determined. Due to the time required for imaging, the types of images that can be used for diagnosis are naturally limited, and doctors or radiation technicians determine what images and how many types thereof will be scanned. A doctor who diagnoses a patient observes a plurality of scanned images arranged in parallel to determine the form of the subject and the state of the living tissue based on these images (see FIG.
2
).
A large amount of experience and training is required for the doctor to understand information on the living tissues of the patient from plural types of medical images. Additionally, with more images or more types thereof, it is more difficult for human beings to image the state of an object, so computers are desirably used for assistance. The recent advanced computer graphics technology has served to spread a technique for displaying a three-dimensional shape of an object from a plurality of images (volume data) continuously scanned by slightly changing a sectional position. A representative approach is surface rendering that displays boundary surfaces of the object, but this approach may totally discard the luminance information inside the surface.
In contrast, an approach called “volume rendering” provides each voxel with a color and opacity to generate an image transmitted from an arbitrary start point so as to visualize minor differences in luminance information. Volume rendering provides voxels with colors by generating transfer functions such as those shown in
FIG. 3
which provide correspondences between voxel luminances and color components (RGB or RGBA with opacity). No method for generating the transfer functions, however, has been established, and each user must empirically generate voxel colors. Thus, color images obtained are not general and cannot be applied to diagnosis or the like easily. Further, since the current coloring method is based on data with a single luminance, only one channel of information is obtained despite the coloring.
If one color image can be generated from multichannel MRI images, pieces of information that can be obtained only under individual imaging conditions can be incorporated in a single image; this image is assumed to be an effective tool for diagnosis. The MRI luminance information, however, may vary and is ambiguous as previously described, so that no definite criterion exists which states what images and how many types thereof can be used to generate color images.
As described above, the prior art has the following problems:
i. The conventional visualization technique principally displays boundaries of an object and fails to effectively use the MRI luminance information.
ii. No definite criterion exists which states what MRI images and how many types thereof can be used to generate color images from multichannel MRI images.
iii. There is no established method for generating transfer functions for assigning colors to voxels.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide a coloring method and apparatus for a multichannel MRI imaging process which effectively uses MRI luminance information to color multichannel MRI in order to incorporate information that can be obtained only under individual imaging conditions, in one image, so that the image can be used as an effective tool for diagnosis.
A coloring method for a multichannel MRI imaging process according to the present invention comprises imaging a plurality of (M-channel) MRI images while varying conditions for a sample such as a tissue sample for which colors can be determined and for which MRI imaging is possible, subjecting data on the M-channel MRI images to a first independent component analysis (ICA) to decompose the images into L (L≦M) independent component images, selecting N points on the sample to create a training sample that is a set of L independent component image luminances and color components. This training sample is used to generate as many transfer functions as the color components which output one color component for an arbitrary combination of the L independent component image luminances. For an object for which colored MRI is to be generated, M′-channel (M′≧L) MRI images are scanned while varying the conditions, and the data on the M′-channel MRI images are subjected to a second independent component analysis to generate L independent component images. Then, the second independent components are each calibrated so as to equal to those of the first independent component, and the transfer functions obtained using the training sample are applied to the calibrated independent component images to obtain a color image.
Additionally, a coloring apparatus for a multichannel MRI image process comprises an MRI device, a component extracting section, a transfer function generating section, a calibration section, and a converting component section into color. The MRI device scanes a plurality of (M-channel) MRI images while varying conditions for a sample such as a tissue sample for which colors can be determined and for which MRI imaging is possible, and for an object for which colored MRI is to be generated, further images M′-channel (M′≧L) MRI images while varying the conditions. The component extracting section subjects data on the M-channel MRI images to an independent component analysis (ICA) to decompose the images into L (L≦M) independent component images, and further subjects data on M′-channel MRI images to an independent component analysis to generate L independent component images. The transfer function generating section selects N points on the sample to create a training sample that is a set of L independent component image luminances and color components, and uses this training sample to generate as many transfer functions as the color components which output one color component for an arbitrary combination of the L independent component image luminances. The calibration section selects a transfer function from the transfer functions generating section and calibrates independent components of the object to be colored so that their scale equals that for independent components of the transfer function. The converting component s
Muraki Shigeru
Nakai Toshiharu
Cohen & Pontani, Lieberman & Pavane
National Institute of Advanced Industrial Science and Technology
Pelham Joseph
LandOfFree
Coloring method and apparatus for multichannel MRI imaging... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Coloring method and apparatus for multichannel MRI imaging..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Coloring method and apparatus for multichannel MRI imaging... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3125876