JPH09297098A - Method and device for judging particle reaction pattern - Google Patents

Abstract

PROBLEM TO BE SOLVED: To judge an error of a sample by representing an image characteristic, which is regarded as a judgment reference at visual inspection, with number row, and comparing a number of a line of the number row, which is obtained from each well of a control well group, with a number of a line corresponding to the line of the number row which is obtained from each well. SOLUTION: An error judging device 33 examines whether or not an image number same as the image number which is obtained from each well belonging in a ordinary well group is in a selected judging image number group, and when existing, it is judged that there is reaction for the well, and when not existing, it is judged that there is no reaction. In the device 33, a minuter analysis is performed for the well which is judged that there is no reaction. When a sample quantity which is injected into the ordinary well is locally different from the specified one, the fault appears as a change of the image number. From an output of a printer 35, an accurate reaction pattern judgment result is known for the normal well, and a number and an error content are known for the well in which an error is caused, in ordinary wells.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a particle reaction pattern judging method and a particle reaction pattern judging method which can contribute to automation of pattern judgment necessary for carrying out a so-called microtiter method for the purpose of conducting a bacterial susceptibility test and an immunoaggregation test. Regarding the device.

[0002]

2. Description of the Related Art In the medical field, when performing blood type determination, antigen / antibody detection, various protein detection, virus detection, bacterial susceptibility testing, etc., detection / determination is widely performed by the microtiter method. ing.

When carrying out this microtiter method, a specimen storage container called a microtiter plate (hereinafter abbreviated as MTP) 1 as shown in FIG. 8 is used to store a specimen. To be done.

The MTP 1 is usually formed of a translucent plastic such as acrylic resin, polystyrene, or polyvinyl chloride, and the well 2 for accommodating a sample is
For example, there are 96 8 × 12 arrays. Well 2
The mutual distance is kept constant, and the bottom wall of each well 2 is formed in a U shape or a V shape as shown in FIG. The peripheral wall 3 of the MTP 1 is formed so as to be higher than the depth of the well 2 by a predetermined amount, and a rib 4 having a thickness of about 3 mm protruding outwardly in the horizontal direction by about 2 mm is formed on the peripheral edge of the lower end thereof. A notch 5 having a width corresponding to the thickness of the peripheral wall 3 is formed on the inner peripheral edge of the lower surface of the rib 4, and by fitting the notch 5 and the upper end of another MTP, A plurality of MTP1 can be stacked. Further, as shown in FIG. 8, an area to which a barcode label 6 for identifying each MTP is attached is set on the outer surface of the peripheral wall 3.

When the susceptibility test of bacteria and fungi is performed by the microtiter method using the MTP1 thus constructed, as shown in FIG. 9, the culture solution 7 is contained in each well 2. A sample (bacteria) is stored and a reagent is added.

Regarding the reagents, generally, the concentration is diluted,
2 density (light, dark), 3 density (light, medium, dark),
Density settings such as 6 density and 12 density are performed. Further, in order to obtain a reference pattern, some of the plurality of wells, for example, three wells are used as control wells for growing a sample only in a culture solution without adding a reagent.

The degree of growth of the specimen appears as a pattern 8 as shown in FIG. 10 (a). The area of this pattern 8 often represents the degree of growth. That is, generally, when the concentration of the reagent is high, the growth is weak, and when the concentration of the reagent is low, the growth is strong.

Therefore, for example, the area of the pattern 8 is compared with the reference area, that is, the growth area in the control wells (well numbers H10, H11, H12 in the example of FIG. 10) to which no reagent is added, and the result is shown in FIG. 10 (b). As shown, negative (-), false positive (±), positive (+, ++, ++++,
...) is determined.

By the way, in many cases, the so-called particle reaction pattern is judged visually by a trained inspector. However, such visual judgment requires skill, and even a considerably skilled person requires a long time for judgment.

In order to reduce the time required for the determination, an automatic inspection device using a colorimeter or the like has been considered. This automatic inspection device adopts the turbidity measurement method. Therefore, for example, if the distribution of the reaction settled at the bottom of the well is non-uniform, or if there is no reaction at the center, it is determined to be "no reaction", and a diluted reaction is also determined to be "no reaction". Further, there is a problem that bubbles generated in the well are judged to be “reactive”. Therefore, in the automatic inspection device using the turbidity measuring method, visual re-determination is essential after the determination.

On the other hand, in order to alleviate the above-mentioned problems of the automatic inspection device using the turbidity measuring method, it is a guideline to stir the reaction in each well when using this device and then measure the turbidity. Has been issued as.

However, if the reaction is stirred in this way, the reaction pattern is forcibly destroyed. For example, even if the reaction pattern is visually different before stirring,
After stirring, the turbidity is about the same and it is indistinguishable. In general, the shape of the reaction pattern contains various valuable information. The combined use of agitation methods discards valuable information contained in the reaction pattern and is not a good idea.

Therefore, recently, an attempt has been made to completely automate the pattern determination by applying the image processing technique.
FIG. 11 shows a schematic configuration of a particle reaction pattern determination device which is a fully automated determination of particle reaction patterns.

This apparatus is provided with an MTP holding table 11 for horizontally holding the MTP 1 used for inspection. MTP
Below the holding table 11, the MTP1 held by the holding table 11 is provided.
An illuminating device 12 that illuminates the lower surface of the illuminating light is provided. The illumination device 12 passes the light emitted from the light source 13 through the diffusion plate 14, and irradiates the light diffused by the diffusion plate 14 on the lower surface of the MTP 1 as illumination light.

On the other hand, above the MTP holding table 11, the MTP is
CCDT for detecting the reaction pattern image in each well 2 of 1
An imaging device 15 represented by a V camera or the like is arranged. The pattern information detected by the imaging device 15 and the identification information read by a bar code reader (not shown) are introduced into the determination processing device 16. The judgment processing device 16 judges the pattern information obtained for each well by using a fuzzy computer, a neuro computer, artificial intelligence (AI), etc., and obtains a judgment result as shown in FIG. 10 (b).
Then, the determination result and the identification information are output to the printer 17. An apparatus that automatically conveys the MTP1 is also considered.

However, even the conventional particle reaction pattern determination device configured as described above has the following problems. In other words, when taking a susceptibility test for bacteria and fungi as an example, each well of the MTP to be tested should originally contain the same sample (bacteria), but sometimes a foreign species is mixed. There is. If the foreign bacterial species are mixed in this way, it is necessary to remove the data as an "alien bacterial species mixing error", but with the conventional particle reaction pattern determination device, it is determined as the same bacterial species, and the determination data There was a problem that would output. Similarly, "skip error" caused by poor reagent concentration setting or sample amount setting error,
There was a problem that various errors such as "fishing bacteria error" and "non-development error" could not be accurately recognized because the sample was not dispensed correctly.

In visual judgment by a trained inspector, the above-mentioned various errors are recognized with high accuracy based on experience. However, in the conventional particle reaction pattern determination device, the reality is that the accuracy of recognition by the inspector has not been reached.

[0018]

As described above, in the conventional particle reaction pattern determination apparatus which is designed to completely automate the particle reaction pattern, when an error element enters the sample, it can be quickly and accurately detected. It is not possible to deal with this, and improvement in determination accuracy is desired. Therefore, an object of the present invention is to provide a particle reaction pattern determination method and an apparatus therefor that can satisfy the above-mentioned demands.

[0019]

In order to achieve the above object, in the invention according to claim 1, the reaction pattern of the specimen contained in each of the normal well group and the control well group of the microtiter plate is imaged by an imaging device. The image information obtained by imaging the normal well group and the control well group with the imaging device is processed prior to determining the reaction pattern of each sample by processing the obtained image information. An image number represented by a numerical string consisting of a numerical value indicating the reaction pattern of each well and a numerical value indicating the reaction characteristics based on a preset judgment standard is created in association with each well, One judgment image number group from the judgment image number group table created in advance based on the image numbers obtained from the wells to which it belongs. To check whether the image numbers obtained from each well belonging to the normal well group belong to the selected judgment image number group, and when those belonging to different judgment image number groups are detected. The feature is that it is determined to be an error.

In order to achieve the above object, in the invention according to claim 2, the illumination light is directed toward the translucent microtiter plate in which the specimens are housed in the normal well group and the control well group, respectively. Illumination means for irradiating,
In a particle reaction pattern determination device comprising means for capturing an image of a reaction pattern of a sample contained in each well of the microtiter plate with an imaging device, and determination means for determining image information obtained by this means, The judging means uses the image information obtained by imaging the normal well group and the control well group with the image pickup device, and based on a preset judgment criterion, the numerical value and the reaction of the reaction pattern of each well are expressed. Image number creating means for creating an image number represented by a numerical sequence consisting of numerical values representing characteristics in association with each well, and previously created based on the image numbers obtained from the wells belonging to the control well group. Image number group selection means for selecting one image number group for determination from the image number group table for determination, It is determined whether or not the image number obtained from each well belonging to the normal well group belongs to the selected determination image group, and when an image number belonging to a different determination image group is detected, an error is determined. It is characterized by including a determining means.

According to the first and second aspects of the present invention, the reaction pattern of each well is used on the basis of the preset judgment standard by using the image information obtained by imaging the normal well group and the control well group with the imaging device. The image number represented by a numerical sequence consisting of a numerical value indicating the characteristic of the above and a numerical value indicating the characteristic of the reaction is made to correspond to each well.

Numerical values representing the characteristics of the reaction pattern are, for example, "0" for a pattern that spreads very thinly, "1" for a star-shaped dispersed pattern such as sparklers, and "2" for a pattern with irregular center. ", The pattern type that can be easily identified by humans is expressed by a numerical value. On the other hand, the numerical value that represents the characteristic of the reaction corresponds to, for example, the level at which a person can easily identify the area of the pattern, the maximum value (or average value) of the grayscale brightness, the gradient of the grayscale brightness, the circularity of the pattern, and the like. It is expressed as a numerical value. For example,
Taking the maximum value of the grayscale brightness as an example, a human can easily discriminate the grayscale brightness into three levels of low, medium, and high. Corresponding to these three stages, for example, "0" is expressed in the case of low, "1" in the case of medium, and "2" in the case of high.

By adopting such an expression system, it is possible to express the image-like feature, which is a criterion for a visual inspection by a trained inspector, in a relatively accurate manner. For example,
When the characteristic of an image is expressed using the characteristic of the reaction pattern and the maximum value of the light and shade brightness of the reaction by using the numerical example described above, when it is expressed by “12” and the numerical sequence, the meaning of the numerical sequence is It can be seen that the pattern that is being displayed is a "star-shaped dispersed pattern such as sparkler and a pattern with high brightness". In the present invention, the number string expressing the image-like characteristics as described above is called "image number".

In the present invention, with respect to the images obtained from each well of the normal well group and the control well group, an image number is created in association with each well. here,
For example, taking a bacterial susceptibility test as an example, each well in the control well group contains only the specimen (bacteria) and the culture solution, so the image numbers obtained from these wells, that is, the rows with the numerical sequence, Always contains a number that represents the characteristics of the sample.

Therefore, in the present invention, the number of the row showing the characteristics of the specimen in the number series obtained from the wells of the control well group corresponds to the above row of the number series obtained from each well of the normal wells. If the numbers are different, it is determined that there is an error in the sample or the like except for the specific condition.

Therefore, it is possible to detect an error with high accuracy and at the same speed as when a skilled inspector makes a visual inspection. Then, the final pattern determination is performed on the wells excluding the error wells.

[0027]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a particle reaction pattern determination device according to an embodiment of the present invention.

In the figure, reference numeral 21 is an MT used for inspection.
P is shown. This MTP is made of a translucent plastic material and has 96 wells in an 8 × 12 array as shown in FIG. In this MTP21, well numbers H10, H11, and H12 are used as control wells, and the rest are used as normal wells, as shown in FIG. Then, the MTP 21 is held on an MTP holding table (not shown).

An illuminating device 22 is provided below the MTP 21, and the illuminating device 22 allows the light emitted from the light source to pass through a diffusion plate 23, and the light diffused by the diffusion plate 23 is used as illumination light on the lower surface of the MTP 21. To irradiate.

An image pickup device 24 typified by a CCD TV camera or the like is arranged above the MTP 21 to detect a reaction pattern image in each well of the MTP 21. A bar code reader (not shown) is arranged on the side. In addition, in order to shorten the imaging time, a plurality of imaging devices 24 may be provided, and these may share the respective areas of the MTP 21 to perform imaging.

The reaction pattern information of each well obtained by the image pickup device 24 and the identification information read by a bar code reader (not shown) are temporarily stored in the image memory device 25.

When the collection of the reaction pattern information for all the wells of the MTP 21 is completed, the reaction pattern information of each well is sequentially read from the image memory device 25 and given to the characteristic value detecting device 26.

The characteristic value detecting device 26 includes a pattern judging device 27 for judging a reaction pattern, a pattern area detecting device 28 for detecting a pattern area which is one index representing the reaction characteristic, and a pattern area detecting device 28 for distinguishing the reaction characteristic. And a brightness detecting device 29 for detecting the brightness which is an index of The index indicating the reaction characteristics is not limited to the pattern area and the brightness, and other indices such as circularity and brightness gradient may be used.

The pattern determination device 27 determines which of a plurality of reaction patterns the input reaction pattern information is classified in advance so that a person can easily identify the AI, fuzzy computer or neuro computer. And the like, and the number X attached to the corresponding reaction pattern is output as the identification result.

In this example, the reaction patterns are classified into, for example, five types in advance, and the five types of classified reaction patterns and the number X have the following relationship. Reaction pattern X = 0: Very thin spreading pattern Reaction pattern X = 1: Sparkler-like star-like dispersion pattern Reaction pattern X = 2: Irregular pattern with central concentration Reaction pattern X = 3: Central concentration Rounded pattern Reaction pattern X = 4: Pattern spread over the whole surface of the well Reaction pattern X = 5: No reaction pattern The pattern area detection device 28 can easily identify the input reaction pattern information by a person in advance. Which category of a plurality of areas which are sensuously classified is classified using the above-described method or the like, and the number Y attached to the corresponding area is output as the identification result.

In this example, the types are classified into, for example, three types in advance, and the three types of areas and the number Y have the following relationship. Area Y = 0: slightly widened area Y = 1: slightly widened area Y = 2: well widened The luminance detection device 29 is such that humans can easily identify the input reaction pattern information in advance. Which of the plurality of sensuously classified luminances is to be classified is identified using the above-described method, and the number Z attached to the corresponding luminance is output as an identification result.

In this example, the luminance is classified into, for example, three types in advance, and the three types of luminance and the number Z have the following relationship. Luminance Z = 0: Low luminance Luminance Z = 1: Medium luminance Luminance Z = 2: High luminance The characteristic information X, Y, Z output from the characteristic value detecting device 26 as numerical values is the image number generating device 30. Will be introduced to. The image number generating device 30 uses the feature information X, Y, and
A numerical sequence in which Z is arranged in the order of XYZ is used as an image number, and these image numbers are associated with each well and temporarily stored. Therefore, the image numbers of one MTP once stored in the image number generating device 30 are as shown in Table 1.

[0038]

[Table 1]

In Table 1, the well numbers H10, H11, H
Image numbers in 12 are for control wells, other image numbers are for normal wells. This Table 1
In, the image number 500 represents a non-reaction pattern. Image number 000 is very thin and has low brightness.
Moreover, it represents a slightly widespread reaction, which is treated the same as no reaction. When the reaction pattern is 0 (X = 0, a pattern that spreads very thinly), Y and Z are 0,
In the case of 1 and 2, the form shown in Table 2 is shown.

[0040]

[Table 2]

Further, the image number 111 is a star-like dispersed pattern such as sparklers, which represents a pattern having medium brightness and a little area. Image number 211 represents a pattern in which there is an irregular reaction in the center of the well, the brightness is medium, and the area is a little.

Image number 221 represents a pattern in which there is an irregular reaction in the center of the well, the brightness is medium, and the area is large. The image number 421 is a pattern that spreads over the entire surface of the well and has a medium brightness and a large area.

Image number 422 is a pattern that spreads over the entire well and has high brightness and a large area. When the image numbers for one MTP are stored in the image number generating device 30, the image numbers in the well numbers H10, H11, H12 from this image number generating device 30, that is, the image numbers of the control wells are determined by the determination image number group selecting device 31. Is output to.

When the image number of the control well is given to the judgment image number group selection device 31, the judgment image number designated by the image number of the control well is selected from the judgment image number group table stored in advance in the memory device 32. Select and output a number group.

The judgment image number group table is used for judging the reaction pattern of a normal well based on the image number of the control well, and is prepared in advance by utilizing the input and learning function by a trained examiner. Table 3
As shown in FIG. 5, it is composed of a judgment number and a judgment image number group.

[0046]

[Table 3]

Here, the additional image numbers in the determination image number group table shown in Table 3 have the following meanings. In other words, taking a susceptibility test for bacteria as an example, even in the case of a specimen (bacteria) that shows a reaction pattern of 4 when it grows well originally, it spreads very thinly when the growth does not proceed so much under conditions of high reagent concentration. It becomes a reaction. When represented by an image number, it is 012 or 022. 012 and 022 indicate that the brightness is very thin and the brightness is high. Therefore, when the value of the reaction pattern of the control well (positive control, only sample and culture solution, no reagent) is 4,
Not only the image number (400 to 422), but also 012 and 0
22 must be treated as the same group.

Further, even in the case of a specimen (bacillus) whose reaction pattern has a numerical value of 2 when it grows well originally, when the growth does not proceed so much under the condition that the reagent concentration is high, the reaction becomes very thin and slightly spread. When expressed by image number, it is 0
In the case of 01. 001 means that the brightness is very thin and the brightness is slightly widened in the middle. Therefore,
When the value of the reaction pattern of the control well (positive control) is 2, the image number (200 to 222)
Not only that, 001 must be treated as the same group.

The same thing as above occurs when the numerical value of the reaction pattern of the control well is 3. Therefore, when the numerical value of the reaction pattern of the control well is 3, not only the image number (300 to 322) but also 0
01 must be treated as the same group.

Table 3 shows an example of a positive control, but it is not limited to the positive control and may be a negative control. In the case of the negative control, the contents of the judgment image number group table are different, but the judgment image number group table can be created by the method described above.

The judgment image number group selection device 31 selects one judgment image number group from the judgment image number group table shown in Table 3 by the following method. That is, assuming that the number of control wells is N, in this example, wells with well numbers H10, H11, and H12 are used as control wells, so N = 3.
Therefore, it is calculated how many image numbers obtained from these three control wells are included in each group from the image number group having the determination number 0 to the image number group having the final determination number. This value is M (control value)
Then, the determination image number group with the largest M is
It is adopted as a group for judging the normal well in the MTP. If there are a plurality of judgment image number groups having the same M value, for example, the judgment image number group having the smallest row number in the table shown in Table 3 is preferentially adopted as the “judgment image number group”.

For example, when N = 3 and the image numbers of the control wells are 422, 422, and 422, the M value of the determination image number group of determination number 4 on the table shown in Table 3 is M = 3. It will be the maximum. Therefore, in this case, the determination image number group of determination number 4 is adopted for determination.

In this way, the selected judgment image number group information of judgment number 4, for example, is given to the error judgment device 33. The error determination device 33 checks whether or not the same image number as the image number obtained from each well belonging to the normal well group is present in the selected determination image number group, and if there is, the well is "reacted". If not, the well is judged as "no reaction". In the judgment image number group of judgment number 4, 012,
Since 022 is registered as the additional image number, the normal wells with these image numbers are also determined to be “reactive”.

The error judging device 33 performs a more detailed analysis on the normal wells judged as "no reaction". That is, when the judgment image number group of the judgment number 4 is selected based on the image number of the control well, the reaction pattern radially spreading from the growth state of the sample should not be detected. Therefore, the image number is 2x
The reaction in the normal well marked with x is not the same sample,
It can be considered that another sample (bacteria) is mixed.
When it is considered that another sample (bacteria) is mixed in this way, the error determination device 33 determines that the normal well is a "faulty species mixed error".

The concentration of the reagent dispensed to the wells should be a concentration at a predetermined stage. However, if a part of the concentration is not as determined for some reason, this collapse may occur. Appears as a change in image number. That is,
In some cases, the reaction does not correspond to the concentration. When it is considered that the reaction does not correspond to the concentration, the error determination device 33 determines that the normal well is "skip error".

Similarly, when the amount of the sample to be injected into the well is not a locally determined amount, this collapse appears as a change in the image number. In this way, when it is considered that the reaction does not correspond to the sample amount, the error determination device 3
In No. 3, the normal well is judged to be "undeveloped error".

When N = 3 and the image number of each control well is 500, 500, 500 (no reaction), the M value of the determination image number group of determination number 5 on the table shown in Table 3 is It becomes the maximum when M = 3. Therefore, in this case, the determination image number group of determination number 5 is adopted for determination. However, in the control well, there should normally be some reaction, and in this case there is no reaction, indicating that this is a mistake in fishing. When it is considered that the fishing bacterium is wrong, the error determining device 33 determines that the MTP is "fishing error".

As shown in FIG. 10B, the image numbers of the normal wells judged as "reacted" by the error judgment device 33 are finally negative (-), false positive (±), positive (+). , ++, +++, ...) is introduced into the final determination device 34. The final determination information obtained by the final determination device 34 is given as an input signal to the printer 35. Further, the error information obtained by the error determination device 33 is introduced into the printer 35 via the error notification device 36.

Therefore, from the output of the printer 35, it is possible to know the accurate reaction pattern determination result for a normal well among normal wells, the number of the well in which an error has occurred and the error content. Therefore, it is possible to obtain a highly accurate determination result.

2 to 4, the flow of the process from the reading of the reaction pattern information from the image memory device 25 to the final pattern determination and error notification is shown in step S1.
~ S19. Of course, such processing is not limited to hardware-based processing, but software can be mainly processed.

The above description of the operation is based on the premise that the reaction in the control well in which the reagent is not dispensed reacts better than the reaction in the normal well. However, in reality, the reaction in the normal well may react better than that in the control well. This may be because the amount of the sample dispensed into the control well is smaller than the amount of the sample dispensed into the normal well. As described above, when the reaction in the normal well is ahead of the reaction in the control well, the judgment image number group is directly determined from the judgment image number group table of Table 3 based on the image number of the control well. If selected, the accuracy of error determination may be reduced.

In such a case, first, it is determined which group of the control well image number groups the image number of the control well belongs to, and based on this control image number group value, the determination image is determined from the determination image number group table. The accuracy of error determination can be improved by selecting a number group.

That is, the control device image number group table shown in Table 4 and the judgment image number group table shown in Table 3 are stored in the memory device 32 in advance.

[0064]

[Table 4]

The control well image number group table is shown in Table 3 with an identification number, an image number group frequently appearing in the control well, and an image number group including an image number group added to improve the judgment accuracy. Determination image number group pointer corresponding to the determined determination number.

Then, the judgment image number group selection device 3
First, using the input control well image number, the control image number group of the control well image number group table in Table 4 is first selected, and the judgment image number group pointer of the selected control image number group is used. The determination number in the determination image number group table shown in Table 3 is selected, and the determination image number group is selected from this determination number.

For example, assume that the number of control wells is N = 3. When the image numbers obtained from these three control wells are 322, 322, and 422, this image number is the last from the image number group of the identification number 0 in the control well image number group table shown in Table 4. The number of images included in each group up to the image number group of the identification number is calculated.

When this value is set as Mc, the judgment image number group pointer assigned to the control well image number group having the largest Mc is read. In this case, the identification number 10, that is, the Mc value of the determination image number group pointer 4 (additional group) becomes maximum at Mc = 3. Therefore, in this case, the judgment image number group of the judgment number 4 in the judgment image number group table shown in Table 3 is adopted for the judgment for judging the normal well in the MTP.

Here, if the control image number group with the identification number 10 shown in Table 4 is not registered, the Mc value of the control image number group with the identification number 3 becomes the maximum 2, and this identification number The determination image number group pointer 3 of 3 finally selects the determination image number group of the determination number 3. That is, the wrong determination image number group will be selected. However, the accuracy of determination can be improved by adding an image number group such as the identification number 10 to the control well image number group table.

Providing the control well image number group table described above is effective when the reaction in the normal well is more advanced than in the control well. However, when the reaction in the control well is more advanced than that in the normal well, sufficient judgment accuracy can be ensured only by using the judgment image number group table. Therefore, it is possible to provide a switching means for determining the progress of the reaction between the control well and the normal well and switching which is selected according to the determination result.

FIGS. 5 to 7 show reaction pattern information from the image memory device 25 when both the control well image number group table shown in Table 4 and the judgment image number group table shown in Table 3 are used. The flow of processing from reading to finally performing pattern determination and error notification is shown as steps S31 to S50.

Although the above-described operation has been described by taking a bacterial susceptibility test as an example, the method and apparatus of the present invention can also be used in a fungal susceptibility test, a bacteria identification test, a virus immune agglutination test, and the like. Is possible.

[0073]

As described above, according to the judging method and apparatus of the present invention, the image-like feature which is a judgment criterion when a skilled inspector makes a visual inspection is expressed by a numerical sequence and controlled. Compare the number in the row representing the characteristics of the sample in the number sequence obtained from each well of the well group with the number in the row corresponding to the above row in the number sequence obtained from each well in the normal well. If the numbers are different, it is determined that the sample or the like has an error except for a specific condition.

Therefore, it is possible to detect an error at a high speed and with the same accuracy as when a skilled inspector makes a visual inspection, and it is possible to dramatically improve the determination accuracy.

[Brief description of drawings]

FIG. 1 is a block configuration diagram of a particle reaction pattern determination device according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a processing procedure when error detection is performed using the same apparatus.

FIG. 3 is a diagram showing an example of a processing procedure when error detection is performed using the same apparatus.

FIG. 4 is a diagram showing an example of a processing procedure when error detection is performed using the apparatus.

FIG. 5 is a diagram showing a processing procedure of another example when error detection is performed using the same apparatus.

FIG. 6 is a diagram showing a processing procedure of another example when error detection is performed using the same apparatus.

FIG. 7 is a diagram showing a processing procedure of another example when error detection is performed using the same apparatus.

FIG. 8 is a perspective view of a microtiter plate used in the microtiter method.

FIG. 9 is a cross-sectional view showing the plate by locally removing it.

FIG. 10 is a diagram showing an example of a pattern form of a particle reaction pattern by a microtiter method and an example of a determination result of the pattern.

FIG. 11 is a schematic configuration diagram of a conventional particle reaction pattern determination device.

[Explanation of symbols]

 1, 21 ... Microtiter plate (MTP) 2 ... Well 7 ... Culture solution 22 ... Illumination device 23 ... Diffusion plate 24 ... Imaging device 25 ... Image memory device 26 ... Feature value detection device 27 ... Pattern determination device 28 ... Area detection device 29 ... Luminance detection device 30 ... Image number generation device 31 ... Judgment image number group selection device 32 ... Memory device 33 ... Error judgment device 34 ... Final judgment device 35 ... Printer 36 ... Error notification device

Claims (2)

[Claims] 1. A reaction pattern of a sample contained in a normal well group and a control well group of a microtiter plate is imaged by an imaging device, and the obtained image information is processed to determine the reaction pattern of each sample. Prior to, using the image information obtained by imaging the normal well group and the control well group with the imaging device, and a numerical value representing the characteristics of the reaction pattern of each well based on a preset criteria An image number represented by a numerical sequence consisting of numerical values representing the characteristics of the reaction was created in association with each well, and the determination was made in advance based on the image number obtained from the wells belonging to the control well group. One image number group for judgment is selected from the image number group table for evaluation, and the image data is obtained from each well belonging to the normal well group. A particle reaction pattern characterized by checking whether or not an image number belongs to the selected judgment image number group, and judging that an error belongs to a different judgment image number group in this check, it is judged as an error. Judgment method. 2. An illuminating means for irradiating a translucent microtiter plate with an illuminating light, the specimen being housed in each of a normal well group and a control well group, and housed in each well of the microtiter plate. In the particle reaction pattern determination device comprising means for capturing an image of the reaction pattern of the sample with an imaging device, and determination means for determining the image information obtained by this means, the determination means is the normal well group and Using the image information obtained by imaging the control well group with the imaging device, a number consisting of a numerical value representing the characteristic of the reaction pattern of each well and a numerical value representing the characteristic of the reaction based on a preset criterion. Image number creating means for creating an image number represented by a column in association with each well, and the wells belonging to the control well group. The determination image group selection means for selecting one determination image number group from the determination image number group table created in advance based on the obtained image numbers, and the determination image group selection means are obtained from each well belonging to the normal well group. Particles characterized by including a determination means for determining whether or not the image number belongs to the selected determination image group, and determining that there is an error when a member belonging to a different determination image group is detected. Reaction pattern determination device.