JPH07248328A - Urine settlement tester - Google Patents

Abstract

PURPOSE:To aim at an improvement in the efficiency of urine settlement tests and the promotion of high accuracy by making the only specified settlement ingredient automatically sortable at the time of automating the urine settlement test. CONSTITUTION:A sample is made to flow into a flow cell 4 with both pickup and grain detection areas, detecting a pass of grains with a photodetector 22, and a flash lamp 1 is emitted, photographing a still picture of the grains, and then the grains are sorted by a technique of pattern recognitions. As for a judgment of whether automatic sorting should be done or manual sorting otherwise, it is judged by an operator, in regard to the case that urine settlement ingredients (items) selected so as to do the automatic sorting, a tester executes it automatically according to the specification.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a urine sediment inspection apparatus, and more particularly to a urine sediment inspection apparatus capable of automatically classifying urine sediment components.

[0002]

2. Description of the Related Art The following work is required when a laboratory technician attempts to visually determine a urinary sediment examination.

(1) Urine is placed in a test tube and centrifuged to separate the sediment.

(2) The precipitated sediment is dropped on a slide glass and dyed if necessary.

(3) A slide sample is prepared and observed under a microscope.

(4) The number of abnormal components constituting the urine sediment is examined by examining several ranges (regions).

As an apparatus for automating these inspections, there is an apparatus which allows particles to be suspended in a liquid and flowed in a flow cell for optical analysis. For example, there are devices that pass a fluid sample through a specially shaped channel, where an optical still image of the particles in the sample is created. According to this, the morphology of particles can be known by converting an optical still image into an electronic image and analyzing it, and the precipitate in urine can be classified based on this morphology. In addition, the morphological analysis of the particles allows detailed classification of the particles in accordance with a certain order, such as size and color, according to the class of visually discriminating characteristics by means of an optical still image. In the urinary sediment examination, urine is flown through a flow cell to create an optical still image, which is then converted into an electronic image for analysis, thereby partially automating the classification work by the technician.

[0008]

Since the urine of a healthy person has a very small amount of sediment components in the urine, classification / counting of sediments can be performed relatively easily by visual inspection by a technician. However, pathological urine contains high concentrations of amorphous salts, mucus threads, red blood cells, white blood cells, etc., and contains an extremely large number of sediments. The automation by the test device is effective for urine containing many sediments, but has the following problems.

(1) When the particles are classified in the order of size,
White blood cells and red blood cells in urine overlap in size because they overlap.

(2) In the case of color, since the glass column and the mucus thread have similar color tones, they are also mixed in the same layer.

As a solution to this problem, an inspection technician uses an optical still image created by an inspection apparatus to classify mixed particles by visual judgment.

Since a very large number of sediment components appear in the urine of a pathological sample, it takes time for the technician to judge all of these components, which is uncomfortable work and leads to inaccurate results. In addition, there is a risk of overlooking the sediment component (for example, atypical cells, cancer cells, etc.) that has a significant diagnostic significance and has a low appearance frequency.

As described above, in the conventional method, even if the work is automated by the inspection device, the classification determination is insufficient, so the efficiency is poor and the accuracy is inadequate. However, there is a problem that it is difficult to perform a rapid urine sediment test because it cannot be classified and counted.

An object of the present invention is to provide a urine sediment device capable of efficiently classifying urine sediment components with high accuracy.

[0015]

According to the present invention, electronic image data of a urinary sediment component is generated, and the characteristic amount of the urine sediment component is extracted based on the generated electronic image data. A urinary sediment inspection device that classifies the urinary sediment components based on the extracted characteristic amount, wherein the urine sediment components determined to require automatic classification are selectively selected from the urine sediment components. There is provided a urinary sediment examination apparatus characterized by comprising means for designating the designated urinary sediment component, and being configured to perform automatic classification based on the extracted feature amount of the designated urinary sediment component. .

[0016]

According to such a urine sediment device, the technician or the operator excludes the urine sediment component which has a bad influence on the classification accuracy and the inefficient urine sediment component from the objects of the automatic classification, and the other urine sediment components. Therefore, the urinary sediment component thus selected as the target of the automatic classification can be automatically classified with high accuracy and efficiency.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS In FIG. 1, light emitted from a flash lamp 1 which is a microscope light source travels on a microscope optical axis 9 and is first collected by a field lens 2 and passes through a condenser lens 3 onto a sample flow 13 in a flow cell 4. Is focused on. In practice, the field stop 11 and the aperture stop 12 limit the light source luminous flux 10.

An image of the urine sediment component in the sample flow is formed at the image forming position 6 by the microscope objective lens 5. The image at this image formation position is projected onto the image pickup surface of the TV camera 8 by the projection lens 7, and is photoelectrically converted there to obtain electronic image data. As a TV camera, a CCD type camera having a small afterimage is generally used.

The emission timing of the flash lamp 1 is controlled according to the detection signal of the particle detection system. The light flux from the semiconductor laser 15 that is continuously oscillating is generated by the collimator lens 1
The light beam is converted into a parallel light beam by 7 and is focused by the cylindrical lens 18 in only one direction. The laser beam focusing position is on the sample flow in the flow cell, and the reflection mirror 19
And it is irradiated through the minute reflecting mirror 14. When the urine sediment component to be measured crosses the laser light flux, the laser light is scattered, and the scattered light is collected by the microscope objective lens 5 used for imaging and reflected by the beam splitter 20. The reflected scattered light is limited to the observation region on the sample flow by the diaphragm 21, and then converted into an electric signal by the photodetector 22 to obtain a urine sediment component detection signal.

The semiconductor laser 15 is constantly lit and always observes that the urine sediment component in the sample passes through the detection region. When the urine sediment component passes and the urine sediment component detection signal due to scattered light is equal to or higher than a predetermined signal, this urine sediment component is determined as the urine sediment component for image processing, and the flash lamp lighting control circuit 23 determines that the urine sediment component is TV. The lamp drive circuit 27 is controlled so that the flash lamp 1 is turned on when the predetermined position of the captured image area of the camera 8 is reached.

Since the light emission time of the flash lamp 1 is set so short that the movement amount of the urine sediment component in the sample flow can be ignored within the light emission time, when the flash lamp 1 emits light, the video output signal of the TV camera 8 is output. Can capture the urine sediment component in the sample stream as a still image. The image processing control circuit 26 controls writing to the image memory 25 and control of subsequent image processing. As the image processing, the feature extraction circuit 28 extracts primary parameters such as the shape, color and size of the sediment component. The image processing control circuit 26 discriminates these primary parameters and the secondary parameters generated by the combination calculation of these primary parameters.
To enter. The discrimination arithmetic circuit 29 classifies the urinary sediment components by using a Lamel heart type neural network. The neural network efficiently performs learning according to the presence / absence of automatic data classification based on the expert's judgment in advance, and the Since the coupling coefficient of is optimized, a neural network operation can be performed using the input feature parameters to automatically classify the target urinary sediment component. Based on the classification result, the central control unit 30 determines whether or not the image needs to be stored according to a preset condition, and if the image needs to be stored, causes the image storage device 31 to store the image. .

Further, by inputting from the keyboard 33,
When the urinary sediment component, that is, the presence or absence of automatic classification for each item is set, the central control unit 30 instructs the identification calculation circuit 29 that classification is necessary or not.

FIG. 2 is a layout of a screen for instructing an input, which is provided to implement the present invention. The item name input area 40 can be an area in which the item of the urine sediment component can be designated. For example, white blood cells, red blood cells, epithelium, casts,
The item names displayed as crystals, bacteria, etc. are updated unit by unit according to the input from the key. The number of units may be one, or may be two or three. The inspection engineer sets the necessity of automatic classification in the automatic classification designation area 41 according to the item name. For example, if necessary, it is displayed, and if not, it is displayed as nothing.

Whether or not the automatic classification is necessary depends on the judgment of the inspection technician, and FIG. 3 shows an example of the conditions for judging whether or not the automatic classification is necessary. FIG. 3- (a) schematically shows an image 42 of red blood cells. Further, FIG. 3- (b) shows a cylindrical image 43. As shown in FIG. 3- (a), the red blood cell image 42 usually has a donut shape or a circular shape. In such a case, since the relationship between the size and area of the urinary sediment component particles and the perimeter is almost constant, the found particles can be judged to be red blood cells.
By the way, in the case of the columnar image 43 in FIG. 3B, the composition of the column itself is different, such as being formed from blood cells, epithelium or crystals, and the size and shape are not constant.
In order to make a clinical distinction, it is necessary to rely on the image itself, and the technologist determines that automatic classification is necessary or not. Although the determination condition of the necessity of the automatic classification in FIG. 3 is considered to be an example, it is a target that the necessity of the automatic classification is not satisfied in addition to the other conditions necessary for the selection.

A classification flow chart is shown in FIG. 4 in order to explain in detail the method of implementing the present invention. In the embodiment of the present invention, after detecting the particles of the sediment component in step 50 of FIG. 4, the particle image is captured in step 51 and the particle image is binarized. From the binarized image data, a feature amount for identification is calculated (step 52). In step 52, for example, the diameter, perimeter, and area are calculated, and the numerical values are used to calculate the specific diameter-to-area ratio. The determination as to whether automatic classification is necessary (step 53) follows the information designated for each item shown in FIG. That is, it is treated as other in the classification of the neural network (step 55). Items for which automatic classification has been designated are classified into corresponding items in step 54. Here, there are several possible methods for realizing the necessity of automatic classification by the procedure from step 53 to step 55. For example, (1) The feature amount of the item corresponding to the necessity of the automatic classification necessity is considered at the time of determination (53 in FIG. 4). (Method based on conditional branch) (2) Classify, but do not count as applicable items.

(3) It is not reflected in the weight coefficient. Several methods are conceivable, such as (Neural network method). If the detected particles are classified and confirmed in step 56, the classification is completed, and if the classification is not completed, the process is repeated.

A method for realizing the necessity of automatic classification for the procedure from step 53 to step 55 will be described in detail below.

FIG. 5 shows an example in which the necessity of automatic classification is realized by a method that does not consider the feature amount. In the particle images (a) and (b) of FIG. 5, blood cells can be determined in the particle image 60 and cylinders can be determined in the particle image 61 with the length of the diagonal line as a feature amount. That is, in the particle image 61, the blood cells are determined when the diagonal lengths a1 and a2 are substantially equal. In the particle image 62, when the diagonal lengths b1 and b2 are n times (n is greater than 1), it is determined to be a cylinder. In this way, when the difference in the feature amount (here, the length of the diagonal line) can determine the classification, when the feature amount is n times, other classification is performed to realize the necessity of automatic classification. be able to.
(Reflecting Method of Automatic Classification Necessity (1)) In addition, the results of automatic classification using these feature amounts are temporarily saved, and the results are displayed according to the automatic classification necessity specified for each item. You can also save. FIG. 6 shows an implementation example of necessity of automatic classification by a method of not counting as applicable items. After the feature amount is calculated in step 71, the item is determined in step 72. Steps 73, 75, 7
Reference numeral 7 is an automatic classification necessity determination for each item, and the information designated in the input area 41 of FIG. 2 is used. This step 7
If it is judged No in 3, 75, 77, step 7
Since 4, 76, and 78 are not executed, they are not stored as the corresponding items of automatic classification. (Method of Reflecting Necessity of Automatic Classification (2)) An example of implementation of necessity of automatic classification by changing weighting factors will be described with reference to FIG. 7. FIG. 7- (a) shows the improvement of the classification accuracy by changing the weighting coefficient, and the example of the drawing is a three-layer neural network. For example, the feature amount is input to the input layer 81. The feature quantity may be a diameter, a perimeter, an area, or the like. The intermediate layer 82 is provided to calculate the value given to each element d, e, f. For example, the value of d is the sum of products of the feature quantities a, b, c and the weighting factors w1, w2, w3.
The output layer 83 shows the classification results g, h, and q, and may be a number corresponding to the item name. For example, the result g is the sum of the values of the elements d, e, f of the intermediate layer multiplied by the weighting factors w4, w5, w6. The classification result is output to the output layer 83 as a value of the result. Here, changing the weighting coefficient changes the classification result, and the accuracy is improved by an appropriate change. In the embodiment of the present invention, the weighting factor is given as a parameter and can be rewritten. When the weighting coefficient is updated, the result is extracted from the neural network in step 84 as shown in FIG. 7B. If the classification results do not match, compared with the results obtained by the manual method in step 85, the weighting factor is changed. Repeating this procedure for each item improves the accuracy of the classification result. Furthermore, an item that has been treated as being unnecessary for automatic classification will not be classified unless the weighting factor that affects that item is updated, and therefore it can be removed from the automatic classification items. Further, the updating of the weighting factor corresponds to the learning of the neuron, but it is efficient because only a limited number of items are updated in a short time.

FIG. 8 is a screen for registering detailed item names for classifying the automatically classified items by the optical image created in the embodiment of the present invention. Input area 91,
Numerals 92, 93 and 94 correspond to the key of the keyboard and the item, and the item is indicated by pressing the keyboard when classifying the images. The input areas 95 and 96 are areas for inputting item names, and can input item names required for manual classification. By using this screen, manual classification can also be used in combination with automatic classification.

Since automatic classification can be set for each item in this manner, it is possible to realize a urine sediment examination apparatus capable of efficient learning, highly accurate classification, and efficient classification.

An embodiment of the present invention based on the above description will be described below.

1. The electronic image data of the urine sediment component is generated, the characteristic amount of the urine sediment component is extracted based on the generated electronic image data, and the characteristic amount of the urine sediment component is extracted based on the extracted characteristic amount. A urinary sediment inspection device for classifying, comprising a means for selectively designating a urine sediment component determined to require automatic classification among the urine sediment components, and the designated urine sediment component A urinary sediment examination apparatus, characterized in that it is configured to perform automatic classification on the basis of the extracted feature quantities of the components.

2. Flow cell, means for flowing a sample solution containing a urine sediment component to the flow cell, means for optically generating a still image of the urine sediment component flowing through the flow cell, and the generated still image To electronic image data, means for selectively designating urinary sediment components determined to require automatic classification among the urinary sediment components, and the converted electronic image Means for extracting a feature amount of the urinary sediment component based on the data, and automatically classifying the designated urine sediment component based on the extracted feature amount. Urine sediment inspection device.

3. Flow cell, means for flowing a sample solution containing a urine sediment component to the flow cell, means for optically generating a still image of the urine sediment component flowing through the flow cell, and the generated still image To electronic image data, means for specifying the necessity of automatic classification of the urine sediment components for each urine sediment component, and the urine based on the converted electronic image data. A means for extracting a feature amount of a sediment component and performing automatic classification on the basis of the extracted feature amount of the urine sediment component designated as requiring the automatic classification. Sediment inspection device.

4. 4. The urinary sediment inspection apparatus according to aspect 1 or 2, further comprising means for displaying the urinary sediment component name and the designated content indicating that the automatic classification is necessary. The urinary sediment inspection apparatus according to aspect 3, further comprising means for displaying the urinary sediment component name and the content for specifying whether or not the urinary sediment component is automatically classified.

6. 6. The urinary sediment examination apparatus according to aspect 4 or 5, wherein the display means is configured to update and display the urine sediment component name unit by unit.

7. The urinary sediment examination apparatus according to aspect 6, wherein the automatic classification result is compared with the classification result by the manual method, and learning is performed so as to match the result.

8. A urinary sediment test according to aspect 5, characterized in that the urinary sediment component designated as not to be automatically classified is classified to be distinguished from the urinary sediment component designated to be required to be automatically classified. apparatus.

9. The urinary sediment examination apparatus according to aspect 8, wherein the previous term display means is configured to update and display the urine sediment name for each unit.

10. Aspect 3, 5, 8 or 9 characterized in that the urine sediment component name required to be automatically classified and the urine sediment component name rejected are registered
The urinary sediment inspection device described in 1.

[0041]

According to the present invention, there is provided a urine sediment device capable of efficiently classifying urine sediment components with high accuracy.

[Brief description of drawings]

FIG. 1 is an overall schematic configuration diagram of a urinary sediment examination apparatus showing one embodiment according to the present invention.

FIG. 2 is a diagram showing an example of a screen when performing automatic classification setting based on the present invention.

FIG. 3 is a determination condition 1 of necessity of automatic classification based on the present invention.
It is explanatory drawing of an example.

FIG. 4 is a diagram showing an example of a classification flow chart according to the present invention.

FIG. 5 is a diagram for explaining an example (depending on a feature amount) of realizing automatic classification necessity according to the present invention.

FIG. 6 is a flow chart of another example (not counted as a corresponding item) of realization of necessity of automatic classification according to the present invention.

FIG. 7 is a neuro-network work and flow chart of still another example (by weighting factor) of the realization of the necessity of automatic classification according to the present invention.

FIG. 8 is a diagram showing an example of a manual classification item setting screen according to the present invention.

[Explanation of symbols]

1. Flash lamp, 2. Field lens, 3. Condenser lens, 4. Flow cell, 5. Objective lens,
6. Imaging position, 7. Projection lens, 8. TV camera, 1
1. Field stop, 12. Aperture stop, 13. Sample flow,
14. Micro-reflector, 15. Semiconductor laser, 17. Collimator lens, 18. Cylindrical lens, 19. Reflector, 20. Beam splitter, 21. Diaphragm, 22. Photodetector, 24. AD converter, 25. Image memory, 26. Image processing unit, 27. Lamp drive circuit, 28. Feature extraction circuit, 29. Identification circuit, 30. Central control unit, 31. Image storage device, 32. Display, 33. Keyboard

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Hakuo Owada 882 Ichimo, Katsuta, Ibaraki Prefecture 882, Ichige, Hitachi Ltd. (72) Inventor Hatsue Katahira, 882, Ichige, Ita, Katsuta, Ibaraki Hitachi Measuring Instruments Division

Claims (10)

[Claims] 1. Electronic image data of a urinary sediment component is generated, a characteristic amount of the urinary sediment component is extracted based on the generated electronic image data, and based on the extracted characteristic amount. A urinary sediment inspection device for classifying the urinary sediment components, comprising means for selectively designating urinary sediment components determined to require automatic classification among the urine sediment components, A urinary sediment examination apparatus, characterized in that it is configured to automatically classify a designated urinary sediment component based on the extracted feature amount. 2. A flow cell, means for flowing a sample solution containing a urine sediment component to the flow cell, and means for optically generating a still image of the urine sediment component flowing through the flow cell.
Means for converting the generated still image into electronic image data; means for selectively designating the urinary sediment component determined to require automatic classification among the urine sediment components; A means for extracting a characteristic amount of the urinary sediment component based on the converted electronic image data, and performing automatic classification based on the extracted characteristic amount for the designated urinary sediment component. A urinary sediment inspection device characterized in that 3. A flow cell, means for flowing a sample solution containing a urine sediment component to the flow cell, and means for optically generating a static image of the urine sediment component flowing through the flow cell.
Means for converting the generated still image into electronic image data, means for designating whether or not the automatic classification of the urinary sediment components is necessary for each urinary sediment component, and the converted electronic image data Means for extracting a feature amount of the urinary sediment component based on the data, and performing automatic classification based on the extracted feature amount for the urine sediment component designated as requiring the automatic classification. A urine sediment inspection device characterized by the above. 4. The urinary sediment examination apparatus according to claim 1, further comprising means for displaying the name of the urinary sediment component and the designation content indicating that the automatic classification is necessary. 5. The urinary sediment examination apparatus according to claim 3, further comprising means for displaying the urinary sediment component name and the contents of designation of the necessity / unnecessity of automatic urinary sediment component classification. 6. The urinary sediment examination apparatus according to claim 4, wherein the display means is configured to update and display the urinary sediment component name unit by unit. 7. The urinary sediment examination apparatus according to claim 6, wherein the result of the automatic classification is compared with the classification result by the manual method, and learning is performed so as to match the result. . 8. The urinary sediment component designated as not to be automatically classified is configured to be distinguished from the urine sediment component designated to be necessary to be automatically classified. The described urinary sediment inspection device. 9. The urinary sediment examination apparatus according to claim 8, wherein the previous term display means is configured to update and display the urine sediment name by a unit number. 10. The urine sediment component name for which the automatic classification is required and the urine sediment component name for which the automatic classification is not registered are registered. The urinary sediment inspection device described in 1.