JPS6131505B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an arithmetic processing device that is effective for use in performing arithmetic operations for image reconstruction, for example, in an X-ray CT apparatus. More specifically, the present invention relates to an arithmetic processing device that utilizes optical signals and is capable of processing a large number of operations in parallel by utilizing the spatial distribution of light.

In an X-ray CT device or the like, the amount of data for image reconstruction is extremely large, and processing this data to obtain an image requires an enormous amount of calculation. For this reason, a minicomputer equipped with an array processor for high-speed calculation is used for this arithmetic processing. however,
Such mini-computers are expensive, and the calculation time is several tens of seconds, which is time-consuming.

FIGS. 1 to 4 are explanatory diagrams showing an example of the procedure of CT image reconstruction. Here, Translate−
The Rotate method (TR scan method) will be shown as an example.

In Fig. 1, the analyte (absorption coefficient μ(x,
y)) When H is irradiated with a single spectrum X-ray of intensity I ₀ , the perspective data P(θ, x) can be expressed by equation (1).

P(θ, x) = ∫lμ(x, y) dl ...(1) Such 1-view fluoroscopic data is obtained by, for example, a 511-channel X-ray detector; By sequentially rotating the perspective data by, for example, 0.625° and obtaining data for all θ (O<θ<π), a total of 576 views of data are obtained.

Back Projection for image reconstruction
Prior to this, each view data is subjected to a convolution operation of a kernel function (correction function) as shown in FIG. 2, for example, in order to make a sharp image. Therefore, the reconstructed image g(x,y) is expressed by the convolution of the original image (x,y) and the kernel function h(x,y), and can be expressed by equation (2).

g(x, y)=∫ ^∞ _−∞ ∫ ^∞ _−∞ (x′, y′) ・h(x−x′, y−y′)×dx′dy′ ……(2) However, x′, y′: Distribution coordinate The convolution operation expressed by equation (2) is
It is a repeated sum-of-products operation, which is usually performed serially on a computer, and the calculations are huge and take a long time. For example, the kernel function h(x,
If the term y) is 1000 terms, then 511 (channel) x 1000 (term) x 576 (view)
This requires a product-sum operation.

FIG. 3 shows an example of a one-view function after performing a convolution operation.

A reconstructed image can be obtained by back projecting the data of each view subjected to convolution in this manner for all θ (576 views) as shown in FIG.

The present invention is an arithmetic processing device that is useful when performing the above-mentioned convolution calculations and uses optical signals, and is capable of performing high-speed calculations with a simple configuration. .

The present invention will be explained below by taking the apparatus shown in FIG. 5 as an example. Here, a case is shown in which a convolution operation is performed. In the figure, 1 is an LED array, in which n (12 in this case) rectangular LEDs are arranged, for example, in the x-axis direction. 2 is this LED
This is the input drive circuit for array 1. A projection signal as shown in the waveform P shown in the figure is given here to drive LED array 1. In LED array 1, the LED elements arranged in the x-axis direction are driven by the waveform P. It emits light with a brightness corresponding to each part of the x-axis of P. Therefore, the LED array 1 emits spatial light whose brightness is distributed in accordance with the projection signal in the x-axis direction.

3 is an optical gate array element, with respect to the x-axis direction.
m pieces arranged with a 45° inclination (here 9
A number of examples are shown below.
This optical gate array element 3 is constructed by arranging a plurality of pairs of electrodes corresponding to the number of optical gates on a PLZT substrate, for example, and using two deflection plates whose deflection planes are perpendicular to each other to form a sound switch. Each optical gate has a light transmittance corresponding to the magnitude of the applied signal. Reference numeral 4 denotes a drive circuit for this optical gate array element 3, to which a kernel function signal as shown by the waveform h shown in the figure is given, and the light transmittance of each optical gate is controlled correspondingly to this, x Each transmittance is distributed according to the kernel function in a direction inclined at 45 degrees to the axis. Here, the kernel function signal h shown in the figure has positive and negative polarities, whereas the light transmittance of the optical gate array element has one polarity, so the positive and negative times of the kernel function are different. The difference in polarity is determined by driving the optical gate array element.

Reference numeral 5 denotes an optical sensor array, which is arranged so as to be able to receive light from the LDE array 1 via the optical gate array element 3. This optical sensor array 5 is
Several strip-shaped optical sensors corresponding to the number n of LEDs in LED array 1 are arranged in the LED array direction (x-axis direction).
They are arranged in a direction perpendicular to the y-axis direction (y-axis direction). Reference numeral 6 denotes an output signal processing circuit which inputs the signal from the optical sensor array 5, and includes an analog memory circuit or an A/D conversion circuit and a digital memory circuit.

The operation of the device configured in this manner will be described next.

A projection signal P of one view is applied to the LED array 1 via an input drive circuit 2, and each LED
emits light with a brightness corresponding to each signal. Here, it is assumed that the relationship between the input signal and the brightness, the span, etc. are determined through an appropriate correction circuit and feedback circuit, but the detailed circuit configuration and its operation are not directly related to the present invention and will therefore be omitted.

The light from LED array 1 is transmitted to light gate array 3.
is irradiated. A signal from the gate drive circuit 4 is applied to the optical gate array 3, and the corresponding gate is first opened by a positive kernel function.
It becomes transparent according to the corresponding signal. At this time, the light from the LED array 1 that is irradiated onto the optical gate array element 3 passes through the light-transmissive gate portion of the optical gate array element 3 and is received by the optical sensor array 5. Here, in the optical sensor array 5, the magnitude of the signal output from each optical sensor corresponds to the intensity of the received light, and the brightness of the light from the LED array 1 and the intensity of the light from the optical gate array element 3 correspond to the intensity of the received light. It is related to the product of light transmittance of corresponding parts. The signals from the optical gate array 5 are stored in memory means within the output signal processing circuit 6. Next, in the same way, the corresponding gate of the optical gate array element 3 is opened by the negative kernel function, and the optical sensor array 5 receives the light that has passed through the optical gate array element 3. do. The signal output from the photosensor array 5 at this time is subtracted from the signal previously stored in the memory means in the output signal processing circuit. This subtracted signal P(θ, τ) can be represented by equation (3), and is obtained by performing a convolution operation on the input signal.

P(θ, τ) = ΣP(θ, x)・h(τ−x)・Δ
x...(3) FIG. 6 is a perspective view of the configuration of a second optical gate array element used in place of the LED array 1 in the device shown in FIG. This optical gate array element 7
is composed of n optical gates arranged in the x-axis direction, and the light transmittance of each optical gate is controlled according to a given signal. The surface of the optical gate array element 7 is irradiated with uniform light, and the light transmitted through the optical gate array element 7 becomes spatial light having a brightness distribution corresponding to the projection signal in the x-axis direction.

In addition, in the above embodiment and explanation, it is assumed that convolution calculation is performed in a CT apparatus, and for this purpose, the arrangement directions of the LED array 1, optical gate array 3, and optical sensor array 5 are not shown in the diagram. In other words, the LED array 1 and the optical sensor array 5 are arranged at right angles to each other, and the optical gate array 3 in the middle is arranged in the opposing line direction. You may arrange it so that it becomes. In this case, the product of two functions can be calculated, and the idea of the present invention includes such a case as well. Furthermore, various optical systems for parallelizing, enlarging, or reducing light beams may be interposed between the respective elements.

As explained above, the device according to the present invention can process a large number of operations in parallel by utilizing the spatial distribution of light, and is extremely effective when applied to CT devices and the like.

[Brief explanation of the drawing]

1 to 4 are explanatory diagrams and diagrams showing an example of a known CT image reconstruction procedure, FIG. 5 is a perspective view of the configuration of the arithmetic processing device according to the present invention, and FIG. 6 is the device shown in FIG. FIG. 2 is a perspective view of a configuration of an optical gate array element used in place of an LED array in FIG. 1...LED array, 2...Input drive circuit, 3
...Optical gate array element, 4...Drive circuit, 5...
...Photosensor array, 6...Output signal processing circuit.

Claims (1)

[Scope of Claims] 1 A plurality of strip-shaped light emitting elements or a plurality of strip-shaped light gates are arranged one-dimensionally and bright in the one-dimensional direction in response to a given first signal. An array of light emitting elements or light gates that output spatial light with a spatial distribution, the plurality of light gates being one-dimensional and having a predetermined angle with the arrangement direction of the light emitting elements or light gates of the array means. The second gate is arranged and its transmittance is given in the direction of arrangement of the light gate.
an optical gate array element distributed according to the signal of
A plurality of strip-shaped optical sensors are one-dimensional,
An arithmetic processing device comprising a photosensor array that is arranged at a predetermined angle with respect to the arrangement direction of the light gates and receives spatial light from the light emitting elements or the array means of the light gates via the light gate array element. 2. Arranging the light emitting elements or light gates of the light emitting element or light gate array means so that the arrangement direction of the light emitting elements or light gates and the arrangement direction of the light sensors of the light sensor array are perpendicular to each other, and arranging the light gates of the light gate array element. The arithmetic processing device according to claim 1, wherein the arithmetic processing device is arranged in a diagonal direction with respect to the arrangement direction of the other two arrays. 3. The first signal is a projection signal, the second signal is a kernel function signal, and the first signal is a kernel function signal. An arithmetic processing device according to claim 2 which is used.