JPH0731155B2 - Method and apparatus for automatic recognition of electrophoretic image - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a method for automatically recognizing an electrophoretic image and an apparatus therefor, and particularly to automatically analyzing and acquiring data from a DNA sequence film image of an electrophoretic image. The present invention relates to a method and an apparatus for automatically recognizing an electrophoretic image for surely identifying boundaries of migration lanes and performing automatic recognition in the apparatus.

[Conventional technology]

Genetic material is a central issue in biotechnology
Research on DNA (Deoxyribo Nucleic Acid) has been actively conducted. In DNA research, determining and analyzing the sequences of bases that are the constituent elements of DNA is an important issue. As a method for determining the base sequence, a method called sequencing by the dideoxy method is widely used.
This is a method in which DNA labeled with radiation is developed by electrophoresis and photographed on an X-ray film so that it can be read by the naked eye. The film thus produced is called a DNA sequence film, and its base sequence can be determined by tracing the arrangement of black spots on the film.
Since the determination of the nucleotide sequence is widely used as a basic technique from basic research fields of biology to applications such as medicine, pharmacy and agricultural chemistry, the number of DNA sequence films produced is increasing rapidly. About 2,000 to 3,000 bases are projected on one sheet on the DNA sequence film, and reading by a skilled person takes several hours. Therefore, there is a demand for a DNA sequence film automatic recognition apparatus that automatically reads and determines the base sequence of DNA from the DNA sequence film.

By the way, in an apparatus for automatically recognizing an electrophoretic image, such as a DNA sequence film, for automatic recognition, boundary discrimination of an electrophoretic image lane is the most basic technique for automatic recognition. The boundaries of the electrophoretic image lanes are identified by, for example, obtaining a characteristic amount representing a change in a direction perpendicular to the electrophoretic direction from the image pattern of the electrophoretic image for the entire electrophoretic electrophoresis. Also, when the boundary identification method of the electrophoretic image lane is applied to each small fragment of the electrophoretic image, the boundary is determined by changing the individual characteristic amounts.

[Problems to be Solved by the Invention]

By the way, in the electrophoretic image, as is clear from the process of creating the electrophoretic image, the boundaries of the electrophoretic image lanes are not necessarily linear, and various curvatures and deformations occur due to the concentration difference of the electrophoretic substrate, temperature conditions, and the like. Further, the density of the electrophoretic image is not always uniform, and the density gradually decreases as the migration distance becomes longer. Therefore, it becomes difficult to identify the boundaries of the electrophoretic image lanes.

In the conventional technology, no consideration is given to this point, and if the boundary of the electrophoretic image lane is curved or deformed or if the density of the electrophoretic image is markedly reduced, erroneous recognition occurs in the boundary recognition. However, there was a problem that the boundary could not be recognized.

The present invention has been made to solve the above problems.

An object of the present invention is to provide an apparatus for automatically analyzing and acquiring data from a DNA sequence film image of an electrophoretic image, by which the boundaries of electrophoretic lanes are reliably identified, and an automatic electrophoretic image recognition method for performing automatic recognition and To provide a device.

The above and other objects and novel features of the present invention will be apparent from the description of this specification and the accompanying drawings.

[Means for Solving the Problems]

In order to achieve the above object, in the present invention, an electrophoretic image is acquired, and in the method for automatically recognizing the electrophoretic image, the electrophoretic image is divided into small sections in the electrophoretic direction. Separately, for each image data of each small section, the boundary characteristic value of each small section is obtained from the sum of the absolute values of the changes between the adjacent pixels, and the boundary characteristic value is matched for each small section, A curve coordinate system showing the correspondence relationship of the axis orthogonal to the electrophoretic direction for each section is obtained, the integral amount of the boundary characteristic value is obtained along the curve coordinate system, and waveform synthesis is performed, and there is a maximum point at each boundary position. Calculate the one-dimensional characteristic value, perform filtering to obtain a moving average with a different period width from the one-dimensional characteristic value, separate the peak from noise,
It is characterized in that the peak is developed along the curved coordinate system to determine the boundary shape of the migration lane.

[Action]

According to the above means, in the step of acquiring and processing the electrophoretic image, the absolute value of the change between the adjacent pixels is divided into the small sections developed in the electrophoretic direction and the image data of each small section. Then, the boundary characteristic value is calculated for each small section, and the boundary characteristic values are matched by each small section to show the correspondence between the axes orthogonal to the electrophoretic direction for each small section. Ask for. Then, the waveform synthesis for obtaining the integral amount of the boundary characteristic value is performed along the curved coordinate system, the one-dimensional characteristic value having the maximum point is calculated for each boundary position, and the one-dimensional characteristic value is calculated for the moving average having different cycle widths. Then, the peak is separated from the noise, and the peak is expanded along the curved coordinate system to obtain the boundary curve.

That is, the boundary characteristic value for each small section of the electrophoretic image is once convoluted into one-dimensional information and restored again to two-dimensional boundary information. Therefore, the influence of local noise is less likely to occur, and the information of the entire image can be accurately determined. In addition, since the boundary characteristic values can be matched regardless of the magnitude of the absolute value, the density is not affected by the change.

As a result, since the boundary is obtained along the curved coordinate system, it is possible to accurately identify the curve and deformation of the boundary.

〔Example〕

An embodiment of the present invention will be specifically described below with reference to the drawings.

FIG. 1 is a block diagram showing a main configuration of an electrophoretic image automatic recognition apparatus according to an embodiment of the present invention. Also, 2a
FIG. 2 and FIG. 2b are diagrams showing an example of processing for determining migration lane boundaries from an electrophoretic image of a DNA sequence film.

The hard wafer structure of the apparatus for automatically recognizing electrophoretic images according to the present embodiment is the same as the Japanese Patent Application No. 62-
Since it is the same as that described in No. 325778, it is omitted here.

In FIGS. 2a and 2b, 10 is the DNA sequence film of the input electrophoretic image, and 20 is the D of the electrophoretic image.
It is an output image example in which a migration lane boundary line is added to an NA sequence film. Further, 21 is a migration lane boundary line.
In the DNA sequence film 10, a target substance labeled with a radioactive substance or a fluorescent substance is electrophoretically developed in a substrate by applying a voltage, and a radiation detection film (X-ray film) is exposed to light to record an electrophoretic image. It was obtained as a film.

For example, in the example of the DNA sequence film obtained by the dideoxy method, as shown in FIGS. 2a and 2b, the boundaries of the migration lanes are the boundaries of the pattern in which the high-concentration portions are alternately seen, Boundary lines cannot be detected. The shape of the boundary line cannot be clearly detected due to fluctuations in electrophoresis, changes in width, and changes in shading.

Therefore, according to the method for automatically recognizing the electrophoretic image according to the embodiment of the present invention, the process of identifying the migration lane boundary line
By performing the one-dimensionalization with less information loss on the boundary extraction characteristic value, the lane boundary shape that is less susceptible to the influence of image deformation, noise, and density change and has high accuracy of identification is identified.

FIG. 1 is a block diagram showing a main configuration of an electrophoretic image automatic recognition apparatus. Predetermined arithmetic processing is performed by the circuit block having the configuration shown in FIG. 1 to detect the migration lane boundary. First
The process of detecting the migration lane boundary will be described with reference to the drawings.

The electrophoretic original image of the DNA sequence film 10 is read by an original image input device 11 composed of a CCD scanner, and in the characteristic extraction circuit 12, divided into a predetermined number of small sections in a direction perpendicular to the electrophoretic direction, The characteristic value for each small section is calculated. The characteristic values for each small section calculated by the characteristic extraction circuit 12 are sequentially stored in the small section characteristic value storage unit 13. Small section characteristic value data 13a stored here
Is calculated by obtaining the absolute value of the change between adjacent pixels for the image data of each small section. That is, the absolute value of the difference is calculated for the grayscale data of the original image for each small section.
For example, a method of integrating the absolute value of the difference between the adjacent pixels, a method of integrating the absolute value of the difference between the average values of several pixels of the adjacent minute areas, and the like are used to calculate the variation between the minute area pixels. It is calculated by obtaining the absolute value of the difference between. The small section characteristic value data 13a thus calculated is waveform data having a peak at the boundary position of the migration lane of the electrophoretic original image. However, this waveform data still contains a lot of noise, and since it is divided, it is not possible to directly determine the entire migration lane boundary of the original electrophoretic image. Therefore, the small section characteristic value data 13a stored in the small section characteristic value storage unit 13 is
Given to 4. The matching circuit 14 performs DP matching processing on the given small section characteristic value data 13a. This DP matching processing is a method that is widely used for matching waveforms having many noise components such as voice waveforms and image data, and is a known method, so a detailed description thereof will be omitted. Circuit 14
For example, the hardware or software performs a process for obtaining a deviation that maximizes the sum of products of corresponding points of the small section characteristic value 13a. By this matching processing, the curved coordinate system data 15a corresponding to the deviation of each point indicating the entire connection of the electrophoretic original image is obtained and stored in the curved coordinate system data storage unit 15. Thereby, the curve coordinate system data 15a representing the optimum correspondence of each small section characteristic value data 13a
Can be asked.

From these data, the waveform synthesizing circuit 16 performs waveform synthesization by integrating each of the plurality of small section characteristic value data 13a along the curved coordinate system of the curved coordinate system data 15a, and synthesizes the entire small section characteristic value data 13a. A synthetic characteristic value 16a is obtained by one-dimensionalizing the value of. The composite characteristic value 16a becomes data having a strong peak at a position corresponding to the migration lane boundary of the original image of the electrophoretic image. The combined characteristic value 16a has a peak, but noise is still present, and moving average filter processing is performed to distinguish the peak from fine noise. This is a composite characteristic value 16a
Is given to, for example, a first moving average filter 17A that outputs a moving average of a half cycle width and a second moving average filter 17B that outputs a moving average of a one cycle width, and the difference between these filter outputs is the difference. This is done by getting from the circuit 18. As a result, the low-noise synthesis characteristic value waveform 18a can be obtained. The period width for obtaining the moving average in this case is appropriately changed according to the characteristics of the original image so that the lowest noise waveform can be obtained. The low-noise synthesis characteristic value waveform 18a thus obtained has only a component corresponding to the migration lane boundary, and the peak position can be easily obtained.

The peak re-expansion circuit 19 performs the peak re-expansion of the low-noise composite characteristic value waveform 18a along the curved coordinate system from the curved coordinate system data storage unit 15, and the migration lane boundary data 21 to be obtained.
To get By this migration lane boundary data 21, by a processing device not shown, or by manual operation manually, each lane of the electrophoretic image of the original image is determined,
Perform automatic recognition of DNA sequences.

As described above, according to the present embodiment, the composite characteristic value is a one-dimensional representation of the characteristics of the entire original image of the electrophoretic image, along the curved coordinate system obtained by performing the matching,
The deformation of the migration lane boundary can be correctly identified, and the moving average filter efficiently removes noise that does not match the periodic component. Therefore, the boundary of the migration lane can be determined extremely accurately, and the automatic recognition of the DNA sequence can be easily performed. The electrophoretic original image has been described here as an electrophoretic image on a DNA sequence film, but the electrophoretic image to be processed is for the purpose of DNA sequence analysis, analysis, and identification, and a label is a radioisotope or a fluorescent substance. , Which uses a dye, a film, a photosensitive paper, a gel, an adsorption paper, a magnetic recording material, a chargeable recording material, an IC memory, etc. as a medium, and processes the electrophoretic image obtained with the migration axis as the distance and time. good. Also,
The image data of the electrophoretic image may be analog image data or digital image data.

The original electrophoretic image can be read as dot data by the CCD scanner of the original image input device and placed on the image display memory of the computer of the processing device. Thereby, the image display and the calculation can be performed in parallel, and the memory efficiency is not reduced.

The main points of this embodiment described above can be summarized as follows. That is, (1) in an apparatus for automatically recognizing a pattern image of an electrophoretic image, the electrophoretic image original image data is divided into small sections developed in the electrophoresis direction, and a sum of absolute differences between the small sections is obtained. The boundary characteristic value is obtained, the boundary characteristic value is matched for each small section, and the matching processing of the characteristic value of the divided small section is performed.

(2) By a matching process, a curved coordinate system along the shape direction of the migration lane boundary line is obtained, and a combined waveform of boundary characteristic values along the curved coordinate system is obtained.

(3) Noise is removed by using a filter that obtains a moving average having a different cycle width with respect to the boundary characteristic value to obtain a difference between the moving averages having different widths with respect to the boundary characteristic value.

(4) The original electrophoretic image is the CCD of the original image input device.
By scanning as dot data with a scanner and placing it on the image display memory of the computer of the processing device, by sharing the image output memory and the image processing memory,
The image display and the calculation are performed in parallel, and the processing is performed without reducing the memory efficiency. As described above, the present invention has been specifically described based on the embodiments.
It is needless to say that the present invention is not limited to the above-mentioned embodiments and can be variously modified without departing from the scope of the invention.

〔The invention's effect〕

As described above, according to the present invention, the image boundary of the migration lane boundary of the electrophoretic image can be accurately determined, so that the electrophoretic image is automatically measured and automatically recognized for various migration patterns in the apparatus. A stable identification rate can be secured.

[Brief description of drawings]

FIG. 1 is a block diagram showing a main configuration of an electrophoretic image automatic recognition apparatus according to an embodiment of the present invention, and FIGS. 2a and 2b show electrophoretic lane boundary determination from a DNA sequence film electrophoretic image. It is a figure which shows the example of a process to perform. In the figure, 10 ... DNA sequence film, 20 ... Output image example with migration lane boundaries, 21 ... migration lane boundaries. 11 ... Original image input device, 12 ... Characteristic extraction circuit, 13 ...
… Small section characteristic value memory, 14 …… Matching circuit, 15 ……
Curved coordinate system data storage unit, 16 ... Waveform synthesis circuit, 17A, 17
B: Moving average filter, 18 ... Difference circuit, 19 ... Peak re-expansion circuit.

Claims (3)

[Claims] 1. A method for automatically recognizing an electrophoretic image by obtaining an electrophoretic image and automatically recognizing the electrophoretic image, the electrophoretic image is divided into small sections which are developed in an electrophoresis direction, and each of the small sections is divided. For the image data of, the boundary characteristic value of each small section is obtained from the sum of the absolute values of the changes between the adjacent pixels,
The boundary characteristic values are matched for each small section, a curved coordinate system showing the correspondence relationship of the axes orthogonal to the electrophoretic direction for each small section is obtained, and the integral amount of the boundary characteristic values is obtained along the curved coordinate system. Waveform synthesis is performed to calculate a one-dimensional characteristic value having a maximum point for each boundary position, and filtering is performed to obtain a moving average with a different period width from the one-dimensional characteristic value, the peak is separated from noise, and the peak is separated. A method for automatically recognizing an electrophoretic image, which comprises developing along a curved coordinate system to obtain a boundary shape of a migration lane. 2. From the image pattern of the electrophoretic image film to DN
In the method for automatically recognizing the A sequence, the original image of the electrophoretic image is divided into small sections in the direction orthogonal to the electrophoretic direction, and the absolute value of the change between adjacent pixels in the divided small sections is calculated. The characteristic value is extracted, the characteristic value for each small section is obtained, each characteristic value is matched, the curved coordinate system along the shape direction is obtained, and the waveform of the characteristic value is synthesized along the curved coordinate system. As a one-dimensional boundary characteristic value, the difference between the moving averages of different cycle widths is taken for the one-dimensional boundary characteristic value,
The feature is that the noise is removed from the peak value, the peak is developed along the curved coordinate system to obtain the boundary shape, the migration lane boundary of the original image of the electrophoretic image is determined, and the DNA sequence is automatically recognized. A method for automatically recognizing electrophoretic images. 3. An original image for inputting an electrophoretic image original image in an apparatus for automatically recognizing a pattern image of an electrophoretic image film, determining boundaries of electrophoretic image original image lanes, and automatically recognizing a DNA sequence. The input device and the input electrophoretic image original image are divided into small sections in a direction orthogonal to the electrophoretic direction, and the characteristic value of the absolute value of the change between adjacent pixels in the divided small sections is extracted, A characteristic extraction circuit that obtains small section characteristic values, a matching circuit that performs matching of each of the small section characteristic values, obtains curved coordinate system data along the shape direction, and synthesizes characteristic values along the curved coordinate system. A waveform synthesis circuit, a plurality of moving average filters that take moving averages of different cycle widths with respect to boundary characteristic values, a difference circuit that takes the difference between the outputs of the plurality of moving average filters, a difference circuit output and curvilinear coordinate system data Automatic recognition apparatus of the electrophoretic image, characterized in that a peak redeployment circuit to perform more peaks redeployment.