Image analysis – Applications – Biomedical applications
Reexamination Certificate
2000-08-30
2004-03-30
Boudreau, Leo (Department: 2621)
Image analysis
Applications
Biomedical applications
Reexamination Certificate
active
06714668
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention concerns the field of image processing, particularly of two- and three-dimensional radiological images.
As is known, radiological apparatuses comprise a means of emission such as an X-ray tube and a means of reception of the emission, such as a solid state detector or even a scintillator and a video camera, of CCD type, for example.
The means of emission and the means of reception of X-rays are generally supported by a mobile system with one or more axes, making it possible to take pictures at different angles of incidence. The means of reception is connected to image processing means making possible the generation of three-dimensional images from a series of two-dimensional images picked up by the means of reception.
It is important to obtain a good match between a three-dimensional image and a two-dimensional image taken upon a stage in which the patient undergoes a particular procedure such as placement of a catheter in the field of angiography, in order to be able to follow the movement of the catheter in the two-dimensional image, but also in the three-dimensional image.
A method of automatic registration of subtracted angiography images is known from the article “Fully automatic 3D/2D Subtracted Angiography Registration” by KERRIEN, LAUNAY, BERGER, MAURINCOMME, VAILLANT and PICARD, published in CARS'99, Proceedings of the 13
th
International Congress and Exhibition, Paris, Jun. 23-26, 1999. This article indicates that reconstructed three-dimensional images of cerebral vascularization can be obtained from a large number of subtracted angiography images and that in interventional neurology it is worthwhile to register the three-dimensional volume reconstructed on the instantaneous two-dimensional images.
In interventional radiology, it is desirable for the practitioner to be able to know at any time that the catheter is in the patient's body with the utmost possible precision. The practitioner can at the present time deduce this information from digital subtracted angiography images called DSA, which he associates with preoperative magnetic resonance thanks to his anatomical knowledge. DSA images offer, among other things, real time imaging with very high space resolution. Three-dimensional reconstructions of blood vessels called “3DXA” have also been used recently from rotational angiography sequences made by rapid rotation of the X-ray tube and of the camera over half a turn and the taking of about fifty DSA images, which are the projections on input of a tomography algorithm producing the 3DXA image on output. For more information on this technique, the reader is invited to refer to the thesis of LAUNAY, “Localization and 3D reconstruction from stereotaxic angiograms,” doctoral thesis, National Polytechnic Institute of Lorraine, Nancy, France, 1996.
These reconstructions make possible a very good appreciation of angioarchitecture. Furthermore, those three-dimensional images can be used in real time according to several types of visualization, such as maximum intensity projection, isosurface, volume melting, virtual endoscopy or even reformatted cross-section, and are a further assist to the diagnoses of practitioners.
SUMMARY OF THE INVENTION
The present invention concerns an improved method of registration.
The present invention also concerns a method of registration of millimetric or submillimetric precision at short calculation time.
The method of automatic registration is intended for two- and three-dimensional angiography images by comparison of a two-dimensional digital subtracted angiography image with data on a three-dimensional image reconstructed from rotational angiography sequences, in which a field of distortions in the image is estimated, a conical projection matrix is estimated and an approximation is made of a rigid transformation in space equal to the difference between an initial registration based on the field of distortions and on the conical projection matrix and a perfect registration. With the rigid transformation comprising a translation part and a rotation part, an approximation is made of the translation part by considering the rotation part as known, the optimal transformation being attained when the correlation of the digital subtracted angiography image and the reconstructed three-dimensional image data is at a maximum.
It is assumed preferably that the residual registration error is due to an error of positioning on translation and rotation of the reconstructed three-dimensional image in the mark of an imaging means.
Preferably, the difference between the reconstructed three-dimensional image data and the two-dimensional digital subtracted angiography image is considered due to a slight displacement following a rigid transformation, the rigid transformation being determined by a modified optical flux technique.
The reconstructed three-dimensional image data preferably relate to a two-dimensional image obtained by conical projection of maximum intensity of the reconstructed three-dimensional image. This type of projection links a single voxel to each pixel. The depth of the voxel chosen as value of one of the space coordinates at a given instant can be used for each pixel on projection.
Advantageously, on each displacement of the digital subtracted angiography image, the new two-dimensional image is calculated from the old two-dimensional image, without new calculation of projection of the reconstructed three-dimensional image. The duration of calculation necessary is considerably reduced from several hours to one or two minutes.
In an embodiment of the invention, the initial registration is given by a marker unit of known three-dimensional position.
In another embodiment of the invention, the initial registration is given by calibration of the angiography machine.
Thus, starting from an initial registration, the initial error in translation is reduced by maximization of the correlation between the digital subtracted angiography image and the reconstructed three-dimensional image data, and by refinement of registration by a modified optical flux technique, it is possible to obtain a registration of millimetric or inframillimetric precision perfectly sufficient for real use in the course of an operation, notably, of interventional neurosurgery or radiology, at the end of a calculation time limited to a few minutes, and even to a few tens of seconds.
The invention applies, in particular, to radiology with three-dimensional images in which the blood vessels are visible. The informational content of such images is slight.
REFERENCES:
patent: 6493415 (2002-12-01), Arai et al.
Kerrien et al., “Machine Precision Assessment for 2D-3D Digital Subtracted Angiographic Images Registration”, Proc. SPIE. Conference on Image-Processing, Feb. 1998, V. 3338, pp 39-49, San Diego, CA.
Feldmar, J et al, “3D-2D Projective Registration of Free-Form Curves and Surfaces”, Computer Vision and Image Understanding, US, Academic Press, V. 65, No. 3, Mar. 1, 1997, pp403-424.
Kerrien et al, “Fully Automatic 3D/2D Subtracted AnGiography Registration”, Cars'99, Proceedings of the 13th International Congress and Exhibition, Paris Jun. 23-26, 1999.
Launay, “Localization and 3D Reconstruction From Stereotaxic Angiograms” Doctoral Thesis, National Polytechnic Institute of Loranne, Nancy, France 1996.
Kerrien Erwan
Launay Laurent
Maurincomme Eric
Vaillant Regis
Boudreau Leo
Cantor & Colburn LLP
Chaskin Jay L.
GE Medical Systems SA
Lu Tom Y.
LandOfFree
Method of automatic registration of images does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of automatic registration of images, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of automatic registration of images will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3283303