JPS5856588A - Method and apparatus for picture analysis - Google Patents

Abstract

PURPOSE:To perform analysis with high accuracy and speed, by obtaining the density difference between a picture of blood corpuscle from a image pickup device and a picture element measured previously, analyzing the picture of the position of detected dip from the density histogram as a threshold value and splitting the region. CONSTITUTION:A picture obtained from a image pickup device 1 is corrected at a shading correction circuit 2 to output an analog signal corresponding to the density of picture element, a horizontal synchronizing pulse and a vertical synchronizing pulse. Each picture element density is outputted from an AD converter 4 at a sampling pulse generator 3 and given to a picture memory 8, a read access memory 6 and a register 10. The previous data is stored in a register 11 and a differential value eliminating dh or over of threshold value is given to an adder 15. The position of dip of the density histogram stored in the memory 6 is detected at a computer 16 and the picture element density is processed at a new threshold value. Next, the processing of classification is done with the computer 16, allowing to attain the analysis processing with high accuracy and speed.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention detects the difference between the background and the object in an image where the difference from the background is not clear among the objects to be analyzed such as blood cells and urine sediment obtained in an optical field of view. It relates to a method and device that enable accurate judgment and processing at a much higher speed than conventional methods, and in particular, an image analysis method and device for performing high-speed region segmentation of an object to be measured in a microscopic image. We aim to provide the following.

Traditionally, when classifying and counting objects such as white blood cells in a smear sample based on morphological differences or differences in size, density, etc., as in the case of white blood cell classification, this method is used to provide clinical data. analyzes microscopic images using various algorithms and performs classification and counting. In cases such as white blood cells that are stained at a relatively high concentration, Vglm is the optical difference between the cells and the background, whereas Vglm is the optical difference between the cells and the background. Since there is almost no difference between the object to be measured and the background, it is difficult to analyze all images using a fixed threshold set in advance. In particular, when performing classification counting, it is necessary to divide regions into areas such as the background and object, relatively high concentration parts such as relations in the target object, and low concentration parts such as protoplasm. , for example, the degree of maroon color, the quality of the smear, the drift of the optical system, and the position of the smear.Since the density of the area is detected differently depending on the position of the smear, it is necessary to set the optimal threshold for each image. It is now necessary to find and set a value. Therefore, the density information of each pixel is stored in the memory address corresponding to each pixel of the image, and then the density information of each pixel is used to create density and plots using a computer, or a similar method is used. The optimal threshold value for connecting each pixel and performing region segmentation focusing on the continuity of features was determined by calculation, that is, by computer software. In any case, in order to obtain the optimum value, the conditions for the same image must be changed little by little to perform arithmetic processing, which has the disadvantage that the calculation time is extremely long. In other words, the method of inputting information into a computer's memory and calculating the threshold value for analysis allows much more accurate analysis than the method of using a pre-fixed threshold value. Although it is possible to add precise correction calculations, it takes a considerable amount of time to process the calculations.
Therefore, the analysis speed for a single case was extremely slow, making it impractical for use in hospital laboratories that require high-speed processing of multiple samples.

The method and apparatus of the present invention can determine the optimal threshold value and capture the image into memory even when the difference between the object in the image and the background is not clear, such as a glass cylinder in urine sediment. The process is processed almost simultaneously, and the time required to achieve the same accuracy as conventional methods, which used to take several minutes to several tens of minutes, can be significantly shortened to about 1/60 to 1/30 seconds. It is something.

In other words, if the end is f:lfl, the density difference value between each pixel of the image obtained by the imaging device and the pixel of the previous measurement is obtained, a density histogram of pixels having density difference values below a predetermined value is obtained, and this density histogram is calculated. It is characterized by detecting the position of the valley in the histogram, analyzing the image using the density value at the position of this valley as a new threshold, and dividing the @ area. The present invention is as follows:
It is done in a similar way. First, a smear of urine or other sample prepared in Preparation 2 is photoelectrically converted by a microscope camera, and the concentration of light is collected as an electrical signal.

These electrical signals correspond to pixels, that is, each dot on the photoelectric conversion surface.The intensity of these electrical signals is proportional to the strength of the light. Each point of the image shown in Figure 1,
In other words, if the density of a pixel is fi, j, then this scanning advances in one direction by 1÷mf, advances by 1 in the J direction, and returns to 1.
Proceeding in the direction 1->m, signals for a total of Xn pixels can be obtained. When using a general television camera, the time to obtain mXn pieces of information, that is, the time to scan the information once, is about 1/60 second. The electrical signal thus obtained is converted into a digital logical value by performing AD conversion on the pulse signal of each pixel. Conventionally, this logical value was directly stored in memory and the computer read each address and performed all arithmetic processing, which took a very long time, but in the present invention, the speed amplifier is achieved as follows. Ta. In other words, does each pixel in the image belong to the background?
Should it be divided by setting a predetermined threshold to determine whether it belongs to the protoplasm or the nucleus? As mentioned in W1, this threshold must be set for each image.
In the past, all of this was calculated using a computer, and then dividing the information in memory required two layers of effort.

In the present invention, the AD conversion value of the image signal is stored in the memory, and at the same time, the difference in density between each pixel is stored in another memory as a histogram. That is, if the density of a certain point in the image is fi, j, the number stored in is read out, 1 is added thereto, and it is stored again at the same address. Therefore, the first memory stores mere density information corresponding to the image 1, and the second memory stores histogram information of the density and difference value of each point of the image. Either the address of the first memory corresponds to the two-dimensional surface of the image, or the address of the second memory 1-i' does not correspond at all to the two-dimensional surface of both images, and the density has been further transformed. This is an address related to distribution information. The contents stored in this second memory are:
It is as shown in FIG. 2, and indicates how many pixels there are corresponding to a predetermined density value and density difference value. Note that the numbers in FIG. 2 indicate degrees. A difference value of zero indicates that there is no density difference between adjacent pixels, which occurs, for example, on a uniform background or inside a cell. In Figure 2, the part that is near the center and shifted to the right corresponds to pixels on the boundary between the cell and the background, and is simply a histogram of the concentration distribution] - Obtaining a remarkable histogram that could not be obtained with Durham. I can do it. Furthermore, in FIG. 2, if a threshold value is set in advance for the density difference value and pixels having a difference value greater than a predetermined value, that is, a density gradient are removed, the histogram for pixels below the threshold value dh will be can get. Thereby, the background and the object can be distinguished with 8A accuracy.

FIG. 3 shows the distribution curve for density values for pixels below the threshold dh, and [density fv(=30
) is the stable boundary level between the background and the object. If this determined boundary level, that is, the density fv, is used as a new threshold and the density of the image already stored in the first memory is binarized (0-!tag), the background and object Things are separated. An example of a case where an image is divided into three types of density (for example, 'i!!i', protoplasm, and nucleus) is shown in Figure 4, and the density histogram obtained from this is shown in Figure 5. . Note that the sample is a binuclear transitional epithelial cell. In Figure 4, each number and A, B,
C and D indicate how many pixels corresponding to each density and difference value are included in the image; 1 to 9 and A6
-iIOIB shows 11 pieces, C shows 12 pieces, and D shows 13 pieces. Note that the threshold value dh is set in advance so as to include the stable bright region of the background and the object. In other words, if the L threshold dh is set to a too large value, noise (information due to dust, etc.) will be picked up, and if it is set to a small value, even the main background and object information will be discarded. , fv becomes difficult to set. Also 9J3 to obtain density histogram! As a matter of fact, averaging is a key element.

In other words, if the number of particles below the threshold dh in Figure 4 is accumulated for each concentration, the number -1 written in the bottom g of Figure 4 is
As shown in the number 2 (sum of threshold values), the unevenness force is severe; however, by repeating the addition of adjacent cumulative values twice (averaging), a relatively smooth figure can be obtained. A density histogram as shown is obtained. In FIG. 5, the left side is the background density group, and the right side is the object density group. The valley of the right curve is the boundary value of the concentration difference.

The median value of the valley obtained in the above manner is used as the threshold for the next image analysis, and an example of the result of processing one screen already stored in the first memory is shown in the first example. It is shown in Figure 6. In FIG. 6, the background is not printed, the cytoplasm is indicated by a cross, and the nucleus is indicated by an X. If the [t4@ area is divided in this way, the classification process is easy. For example, in classification based on area, even if the image quality varies somewhat depending on the staining situation, a constant threshold value can always be obtained through the above processing, so accurate area can be obtained. Furthermore, when extracting features based on the presence or absence of nuclei, information from high-density scars other than nuclei is
- In determining the threshold value described in (2) above, n (because the difference value exceeds the threshold value dh), two consecutive pixels or more.
Since only continuous information of 1 m is used as information for determining the threshold value, it is possible to accurately determine the nucleus and divide the area, so that the area of the nucleus can always be accurately obtained. It is almost translucent or transparent, like a glass cylinder in urinary sediment, and there is no noticeable difference depending on the staining, and it is difficult to set a threshold value depending on the density of the background. According to the above method, the threshold value is determined only based on valid information from which unstable boundary components and noise components are removed, so that the region can be divided reliably.

Subsequent processing is performed using conventional pattern recognition or other feature extraction methods.
Classification and counting.

Next, an example of applying the above method to a real circuit is shown in the 7th section.
As shown in the figure. ++ shown in Figure 7! The image analysis device includes an imaging device 1 such as a television camera, a setting correction circuit 2 connected to the imaging device 1, and an imaging device 1.
A sampling pulse generator 3 generates a sampling pulse corresponding to each pixel in response to a horizontal synchronizing pulse signal from the 3D converter, and converts the density of each pixel into an AD signal using this sampling pulse and outputs a logical value representing the density. A.D.
A converter 4, a read access memory 6 connected to the AD converter 4 via a memory address register 5, and a
An image memory 8 connected to the D converter 4 via a memory cheater register, two registers 10.11 connected to the AD converter 4, and two registers connected to these registers 10.11. A subtracter 12 compares the data and performs subtraction to obtain a density difference value, and compares this density difference value with a preset threshold value and passes only those below the threshold value. a digital comparator 13 which removes the following two; a digital adder 15 connected to the digital comparator 13 and the read access memory 6 via a memory data register 14; and a digital adder 15 connected to the memory data register 14 and which removes the A computer 16 connected to the image memory 8 via the memory data register 7, and an X provided between the imaging device 1 and the image memory 8.
It consists of a counter 17, a Y counter 18, and a memory address register 20.

The image obtained by the imaging device 1 usually has shading characteristics, and since the brightness of the image is not uniform, a phenomenon occurs in which the obtained image is dark at the periphery and dark at the center. Therefore, the shifting correction circuit 2 is a circuit that sends a correction signal for correcting the output value according to the position of the image to the imaging device 1, and corrects it so that it outputs the desired signal in the future. Is it possible to record and record the output value for one screen in the state without the image, correct the obtained image d8 later with a computer user, and output it? There was a problem, so I synchronized it with the scanning time of the imaging device 1 and
The power of doing E is speed amplified. The imaging device 1 outputs an analog signal corresponding to the density of a pixel of an image, a horizontal synchronizing pulse signal and a vertical 191 pulse signal necessary for indicating the position of the pixel. In response to the horizontal synchronizing pulse signal, the sampling pulse generator 3
A sampling pulse corresponding to each pixel as shown in the figure is generated. The density of each pixel is AD converted by this sampling pulse, and a logical value f representing the density is outputted from the oil exchanger 4. The output f of the AD converter 4 is sent to the image memory 8, and is also sent to the read access memory 6 and register 1o for setting the threshold value for area division as described ifJ. In the above-mentioned method, in order to make the explanation easier to understand, histograms including the difference values as shown in FIGS. 2 and 4 are shown, and the number of rLVC groups with respect to the concentration as shown in FIGS. 3 and 5 is shown. The method for obtaining a density histogram that displays the difference value has been described, but as a result, it is sufficient to cut out the part where the difference value is large and create a density histogram based only on the L-shaped part, so that part can be omitted in the actual circuit.

Note that the memory address register 5.20 is a temporary storage element for designating an address in the memory; the former designates an address corresponding to the density, and the latter designates the address in the memory corresponding to the pixel position. On the other hand, memory data register 14.71-1 is a temporary storage element for information to be written into vcB in memory, and writing is performed when the signals to the address register and data register are aligned. Registers 10 and 11 are so-called registers where data is transferred.
The data is transferred from the register 10 to the register 11, and the stored contents change one after another. This timing is achieved by pulses from the Sanoprisig pulse generator.

The register 11 contains the data of 1 life 7 times or 8 times (meaning), therefore, by comparing the data in registers 1.0 and 11 and subtracting with the subtracter 12, the difference value Δf is calculated.
This difference value Δf is compared with a preset threshold value dh by a digital comparator 13, and only those below the threshold b value are passed through, and those below the threshold value dh are removed. That is, the value F below the threshold dh is set to 1, and the value below the threshold dh -1 is set to 0D. Next digital adder 151j? This is to read out the number that has already been allocated to the read access memory 6 according to the logical value of the address specification of the memory address register 5, add the (WVcl), and write it to the read access memory 6 again. Difference value Δf
Since the threshold value dh J: is zero when fi is also large, the content Ii of the one-node access memory 6 does not change.

That is, it is the original storage content. - Force, (The surface image memory 8 is sent via the memory address register 20 data of the specified address counted by the X kakunku 17 and the Y counter 18, that is, the x, y coordinates corresponding to the position of the pixel of the image. On the other hand, the output f of the AD converter 4 is
It is sent to the data register 7 and stored as density information in the image memory.

As described above, the image memory 8 stores image information, while the read access memory 6 stores the density and stogram information of pixels having a difference value greater than or equal to the threshold value dh.
Both memories are processed for each image at a high speed of several tenths of a second. When the memory writing for the l image is completed, the density Hist/Durham averaging is performed as described above, and the valley (device detection) is performed according to the main program built in the computer, and then the valley position 111 is detected. A new threshold,
The density of the pixel stored in the image memory 8 is processed by a new threshold value, and the density of the pixel stored in the image memory 8 is processed according to the new threshold value, and the concentration of the pixel is
The area is determined by performing multi-value conversion such as IRFT conversion. An example of the output obtained by this process (as shown in FIG. 6), and subsequent processes such as classification are also performed according to the program stored in the computer 16. The former method was too time-consuming because the area was determined by a computer software program, or everything was processed using calls or fixed thresholds, without using the method or device described in 1f1. However, with the method and apparatus of the present invention, the accuracy is improved and the time is significantly shortened. The processing can be carried out in a short time, and the processing capacity of leukocyte classification devices, urine sediment classification and counting devices, etc. can be greatly improved.

As explained above, according to the present invention, threshold settings for dividing the measured object into regions, which were conventionally done using software (computer programs), can now be done using hardware (circuits). Therefore, what conventionally took several minutes to several minutes can now be done in about 1/60 to 1/30 seconds, greatly reducing the time required. In addition, the conventional method using a fixed threshold, which was fast, did not have accuracy when processing samples that changed or translucent objects, but using the method and apparatus of the present invention, It has the advantage that high-precision processing is performed in accordance with the concentration of the sample without changing the speed2, and it is not easily affected by noise components such as dust and scratches.

[Brief explanation of the drawing]

Fig. 1 is an explanatory diagram showing pixels in an image taken by an imaging device such as a television camera, Fig. 2 is an explanatory diagram showing the number of pixels corresponding to predetermined density values and density difference values, and Fig. 3 is an explanatory diagram showing the number of pixels corresponding to predetermined density values and density difference values. The following graph showing the distribution curve for density values for pixels, Figure 4, shows two examples of the case where an image cannot be divided into three types of density (for example, dorsal fin, protoplasm, and nucleus) Theory 11 5 is a density histogram obtained from FIG. 4, FIG. 6 is an explanatory diagram showing an example of the result of processing one screen, and FIG. 7 is an embodiment of the apparatus of the present invention. A systematic explanatory diagram, FIG. 8, is a waveform diagram of numbering pulses corresponding to each pixel in the method of the present invention. ■...Imaging device? 't, 2...Shading correction circuit, 3...Sampling pulse generator, 4...A
D converter, 5...Memory address register, 6...
Read access memory, 7...Memory data register, 8...Image memory, 10.11...Register, 1
2... Subtractor, 13... Digital comparator, 14...
...Memory data register, l'5...Ticital adder, 16...Calculator, 17...X kakunta, 18
... Y Kakunta, 20... Memory address register patent application Toa Medical Electronics Co., Ltd. Figure 1 Figure 2 'ft, 7-fL45-ri

Claims (1)

[Claims] 1. A density difference value between each pixel of an image obtained by an imaging device and a pixel measured 077 times is obtained to obtain a density histogram of pixels having density difference values below a predetermined value, and this density histogram is obtained. An image analysis method characterized by detecting the -6 position of a histogram, analyzing the image using the density value at this valley position as a new threshold, and dividing the vA region. 2. An imaging device that generates an analog signal corresponding to the density of a pixel in an image and a horizontal cyclic pulse signal and a vertical synchronous pulse signal necessary to indicate the position of the pixel, and an imaging device that is connected to this imaging device and outputs an analog signal corresponding to the position of the image. A shifting correction circuit that sends a correction signal to the imaging device to correct the signal so that it outputs the correct signal, and a shifting correction circuit that receives the horizontal synchronization pulse signal from the imaging device and responds to each pixel. A dunzoring pulse generator generates a sampling pulse, and this sampling pulse generates 6+iW.
An AD converter that performs AD conversion of the elementary concentration and outputs a logical value representing the concentration, a read access memory connected to this AD converter via a memory address register, and a read access memory connected to the AD converter via a memory data register. The image memory connected to the AD converter, the two registers connected to the AD converter, the tl connected to these registers, and the two registers are compared and subtracted to calculate the density difference value. A subtracter, a digital comparator that compares this density difference value with a preset threshold value, passes only those below the threshold value, and removes those above the threshold value, and this digital comparator and lead. It is connected to the access memory via the memory data register, reads out the number already allocated to the read access memory by the logical value of the address specification of the memory address register, adds 1 to that value, and writes it to the read access memory again. a digital adder, a computer connected to the memorandum disk and connected to the image memory via a memory data register, an X kakunta, a Y kakunta, and a An image analysis device comprising a memory address register.