JPS62189582A - Ct image aligning system - Google Patents

Abstract

PURPOSE:To obtain a high picture quality three-dimensional re-constitution image by extracting a position deviation vector in a position deviation detecting area set optionally between adjoining slices, smoothing the extracted position deviation vector in the slicing direction, thereafter, using the vectors before the smoothing and after the smoothing and executing the registration processing such as polynomial approximation. CONSTITUTION:When a position dislocation occurs, position deviation detecting areas 201 and 211 are set to adjoining slicing images 20 and 21, and the relative position deviation of central picture element 202 and 212 of respective areas is extracted. Concretely, while moving the area 201 in the area 211, a position deviation vector 32 is obtained by a template matching method, etc. For a full line graph 43 obtained by arranging the detected position deviation vector by obtaining the (x) component of the vector for the ordinate and a slicing position (slice No.) for the abscissa, a smoothing processing is executed, and the smooth position deviation vector is obtained like a dotted line 42. As the smoothing system at such a time, the processing to remove a peculiar action such as points 44 and 45 is executed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to three-dimensional reconstruction processing of CT images, and particularly relates to a correction method that is effective when positional displacement occurs due to body movement, such as in a whole-body CT image. .

[Background of the invention]

As a body motion correction method between CT slice images, the above patent application
8-209064 extracts the contour from each slice image, smooths the contour line in the slice direction, and performs coordinate transformation using data before and after smoothing to eliminate one position shift. The 0-line method that we are trying to correct is
Although positional deviations on the body surface can be significantly reduced, there is a problem in that positional deviations are detected in only one contour, and correction cannot be completed in other areas.

[Purpose of the invention]

An object of the present invention is to provide a three-dimensional reconstructed image with higher image quality by performing the alignment between the slices in a large number of regions of the slice image.

[Summary of the invention]

The present invention focuses on the fact that images are similar between adjacent CT slice images and that the shapes of bones, Wa organs, etc. in the human body change smoothly, and the position is set arbitrarily between adjacent slices. Extract the positional deviation vector within the deviation detection area,
After smoothing the extracted positional deviation vector in the slice direction, registration processing such as polynomial approximation is performed using the vectors before and after smoothing to realize positioning between slices.

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to FIGS. 1 to 6.

FIG. 1 shows how the slice images 10 to 14 are displaced due to various body movements.

When a positional shift occurs in this way, as shown in FIG. 2, a positional shift detection area 20.
1,211, and center pixel 202,212 of each area.
Extract the relative positional deviation of Specifically, as shown in FIG. 3, while moving the region 201 within the region 211, the positional deviation vector 32 is determined by a template matching method or the like.
seek.

The above processing is performed between all adjacent slices as shown in FIG. In the figure, it is now assumed that among the plurality of target slice images 40, a shift occurs in slice 401 due to body movement. If we number each slice from O to 8, between slices 3 and 4 and 4 and 5, there will be 411,41
2) A deviation as shown in Fig. 2) is detected. When the detected positional deviation vectors R' are arranged with the X component of the vector on the vertical axis and the slice position (slice Nα) on the horizontal axis, a solid line graph 43 is obtained. Therefore, a smoothing process is applied to 43 to obtain a smooth positional deviation vector as shown in the dotted line graph 42. The smoothing method in this case is point 44.4.
Remove specific movements like 5 (e.g. Median
filter) processing.

The above discussion was about a specific positional deviation detection area, but the above process is performed by specifying a large number of areas as shown in Figure 5.
As shown in FIG. 6, between adjacent slice images 60° and 61
Matching processing is performed using combinations of 0 and 601 and 610 and 611 to extract positional deviation vectors. Hereinafter, similar processing is performed for slices 62 and 63.

The positional deviation vector extracted in this way is
For each y-direction component, smoothing is performed in the slice direction as shown in FIG.
y3' component 70, 72, smoothed Xp y component 71.
Combinations like 73 (ub, Vb) (k=1~
n), the following m-th order high-order polynomial can be used as the coordinate transformation formula, for example.

The coefficients of the above coordinate transformation formula are determined by the least squares method. That is, the coefficient that minimizes Q expressed by the following equation is calculated by solving the following normal equation.

F3nxs aazz abzt EJbst As a result, a slice image after smoothing (that is, a registered image) can be obtained from a slice image before smoothing.

〔Effect of the invention〕

As described above, according to the present invention, alignment between slices can be performed with high accuracy using a plurality of misalignment detection areas, and therefore a high-quality three-dimensional reconstructed image can be generated. be.

[Brief explanation of drawings]

Figure 1 shows body movement between slices, Figure 2 shows the positional deviation detection area between slices, Figure 3 shows the positional deviation detection method using template matching, Figure 4 shows smoothing of the positional deviation vector, and Figure 5 6 is a diagram explaining the setting method of the displacement detection area, FIG. 6 is a diagram explaining the relative positional relationship of the displacement detection area between slices, and FIG. 7 is a diagram explaining the coordinate transformation method using the displacement vector before and after smoothing. It is. 211... Search area, 201... Template area, 32... Misalignment vector, 401... Slice that caused misalignment, 41... Misalignment vector,
411゜412...Singular vector, 43...Change before smoothing, 42...Change after smoothing, 600,601
... Set of positional deviation detection areas, 70.72 ... Positional deviation amount before smoothing (x+y), 71, 73... Positional deviation amount (X, y) after smoothing. Agent Patent attorney Katsuma Ogawa □″ / Nijika Riju 7 Figure

Claims (1)

[Claims] In the process of reconstructing a three-dimensional image from a plurality of transverse tomographic images (slice images), a means for detecting positional deviation between adjacent slice images from an arbitrary positional deviation detection area set in each slice image, and a detected positional deviation 1. A CT image positioning method comprising means for smoothing a vector and means for performing registration processing between a positional deviation vector before smoothing and a positional deviation vector after smoothing.