JPS6044636B2 - Data smoothing method in radioisotope distribution imaging device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a data smoothing method in a radioisotope distribution imaging apparatus.

To obtain a radioisotope (hereinafter referred to as RI) distribution image, an RI substance is injected into the subject, and a detection device such as a scintillation camera is used to detect the RI substance flowing through the body with blood and accumulating in organs. Then, the detected values are used as data to construct an image on a matrix, and are processed and displayed.

At this time, the data (images) detected by the camera and stored on the matrix are used to obtain a plurality of RI distribution images separated at regular time intervals, like the time-lapse film of an 8TWL camera. From these multiple RI distribution images, the RI intensity that changes at regular intervals is determined for each position, and a number of time-dependent change curves are created and used as diagnostic data. Generally, these curves may contain statistical fluctuations, resulting in curves with dispersion, especially unnecessary unevenness (i.e., discontinuities), and as diagnostic data, data with a large amount of error should not be used. becomes. In order to eliminate this drawback, conventionally dedicated hardware is provided and processing (smoothing processing) is performed using this dedicated hardware to eliminate unevenness. However, this method requires a large amount of processing time. For example, length and width 64
Considering n images divided into ×64 matrices, 64 × 64 × 4096 curve data will be smoothed, and if it takes 100 msec to smooth n points/1 curve, the above dedicated hardware will need to smooth. According to 4
It took 09.6 seconds. An object of the present invention is to provide a smoothing method for a radioisotope distribution image/device that enables smoothing of a large amount of data in a short time.

The gist of the present invention is to average the data every m images without smoothing each individual curve data, and to calculate the resulting matrix data (images) of multiple images. It is meant to be remembered.

The details will be explained below. FIG. 1 is a diagram showing an embodiment of the present invention.

The RI processing device 1 stores a plurality of sheets of RI distribution image data, and each of the plurality of sheets has its own classification (0 to 6) in the vertical direction.
The data has a matrix-like data structure that is horizontally divided into six divisions (A) to 6 (A) to 6 (A). First, the first distribution image data of this RI processing device 1 is read out, and the first
64x64 matrix memory 2. Next 2
Read out the distribution image data of the first sheet. At this time, it is added to the previously written first sheet of distribution image data and written into the first matrix memory 2. This addition is realized by reading the previously written data from the memory 2, adding it to the data on the second sheet, and writing the resultant data. Therefore, in the matrix memory 2, instead of the first distribution image data, the result of adding the first and second distribution images is written. Apply the above method in the same way up to the 5th sheet. Therefore, as a result of reading and adding the data of the fifth sheet, the result of total addition of the corresponding data from the first sheet to the fifth sheet is stored in the first matrix memory 2. . When the reading, addition, and storage of these five sheets is completed, a process of dividing the added and stored data by a numerical value of 5 is performed. Division by the numerical value 5 is a so-called averaging process, and the division may be directly processed using the division function, but in this embodiment, each division of 64×64 read from the first matrix memory 2 For each data, it is checked how many times the numerical value 5 can be subtracted, the number of times is counted, and the number of times is considered as the result of division. This division process is 64×64
As a result, 64×64 pieces of matrix data are obtained. Such division processing is performed by register group 3. The 64×64 matrix data obtained by the register group 3 are stored in the second matrix memory 4. The data stored in the second matrix memory 4 has undergone smoothing processing, and when 64 x 64 matrix data have been stored, a specific area of the magnetic drum 5 (having a 64 x 64 matrix area). Next, as the next process for the 1st to 5th images,
The same process is performed for the second to sixth sheets. The result of this processing is also 64×64 matrix data, which is stored in an area different from the above area of the magnetic drum 5. Thereafter, the same process is performed for the 3rd to 7th sheets, the 4th to 8th sheets, etc., and each magnetic drum 5 is
Each memory is stored in a different area. In this way, smoothed total distribution image data is obtained on the magnetic drum 5, and data with high diagnostic accuracy can be obtained by passing it on to the subsequent diagnostic information calculation device. In the above embodiment, the unit is five sheets, but this is just an example, and the value will naturally vary depending on the diagnostic request and data classification. As for the number of sheets,
When handling around 100 sheets and applying the above example, the smoothing processing time was about several minutes, and a significant improvement was achieved. In the above embodiment, control of various elements, such as writing, reading, and address setting, can be realized by the control circuit 6. Furthermore, the above embodiment can also be implemented by a computer main body. At this time, the matrix memory may be the main memory, or may be a separately provided common matrix memory. FIG. 2 shows a concrete embodiment of the present invention.

The present embodiment includes a data detection section 10, a data processing section 11, a control section 12, and a diagnostic processing section 13. Data detection section 1
At 0, data of multiple matrices are obtained.
The data processing unit 11 includes a first matrix memory 110,
Second matrix memory 11, addition register 112
, and a division register 113. The control unit 12 includes a processing number control unit 120, a read number control unit 121, an address and access control unit 122, and a write number control unit 123. The data processing section 11 operates under the control of the control section 12. The overall operation will be explained.

First, one sheet of matrix data from the data detection unit 10 is written in the first memory 110 in the DMA format in the matrix address format. After this data is stored, the data in the first memory 110 and the data in the second memory 111 are read out for each matrix intersection and are stored in the addition register 111.
is sent to and added to. The result of this addition is sent to the second memory 11 and stored therein. In addition, the first memory 110
, 111, the contents of the memory 11 are all zeros, so the addition is just a formality, and essentially the contents of the memory 110 are written as they are to the second memory 111. It turns out. Next, the data of the second matrix is read out to the first memory 110, added to the data of the previously stored memory 11, and then transferred to the second memory 11 again. Remember. Thereafter, data is similarly read, added, and stored up to m sheets. As a result, the memory 111 finally has 1
This means that the data addition results for the total data of ~m sheets are stored. After storing the result of addition of 1 to m pieces of total data in the memory 111, the contents of the memory 111 are sent to the division register 113 and divided by the numerical value m, and the result is memo! Memorized by River 11. The result is averaged data, ie, smoothed data. This smoothed data is sent in DMA format to the diagnostic processing section 13 and subjected to diagnostic processing. In the above process, the read number control unit 121 has the function of specifying the number m of reading matrix data from the data detection unit 10 and sending the number m to the division register 113.

The address and access control unit 122 includes the memories 110 and 1
11 addresses and access control thereof. The write count control unit 123 specifies and controls the write count from the memory 111. This completes the first data smoothing.

Next, a second data smoothing is performed. This second data smoothing is performed in the same way between the second matrix data and m+1 matrix data. The same goes for the third and subsequent times. In this way, the diagnostic processing unit 13 receives all the data that has been smoothed. According to the present invention, high-speed smoothing processing has become possible.

[Brief explanation of the drawing]

FIG. 1 shows an embodiment of the present invention, and FIG. 2 shows a more specific embodiment of the invention. 1... RI processing device, 2, 4, 110, 111
... Matrix Memo I83 ... Lujista group, 1
0... Data detection section, 11... Processing section, 12... Control section, 13... Diagnosis processing section.

Claims (1)

[Claims] 1 A plurality of sheets of radioisotope distribution image data obtained from each radioisotope distribution image after radioisotope injection from a subject as a distribution image in a matrix format are stored, and m sheets from the first sheet among the plurality of sheets are stored. 2nd to (m+1)th, 3rd to (m+2)
For each of m sheets, data is added for each division of the matrix over m sheets, the addition result is divided by m, the division result is stored in each division of the corresponding matrix, and each m sheet is A data smoothing method in a radioisotope distribution imaging apparatus, which obtains smoothed data for each m images and stores new smoothed matrix data obtained for each m images in separate memory areas.