KR102712830B1 - Method for quantitative analysis of target gene and device using the same - Google Patents

Abstract

A device for quantitative analysis of a target gene and a method for using the same according to an embodiment of the present invention are provided. A method for quantitative analysis of a target gene according to an embodiment of the present invention includes a step of obtaining a plurality of first sheet images of micro-droplets containing a target gene and a fluorescent substance using a confocal lens-based quantitative analysis device, a step of obtaining a second sheet image based on the plurality of first sheet images using an optical model based on an artificial intelligence algorithm, and a step of performing quantitative analysis of the target gene based on the fluorescence intensity of the micro-droplets in the second sheet images.

Description

METHOD FOR QUANTITATIVE ANALYSIS OF TARGET GENE AND DEVICE USING THE SAME

The present invention relates to a device for quantitative analysis of a target gene and a method of using the same.

DNA amplification technology is widely used for research and development and diagnostic purposes in the fields of life sciences, genetic engineering, and medicine, and among them, DNA amplification technology using polymerase chain reaction (PCR) is widely used.

Specifically, PCR is a molecular biology technique for replicating DNA from isolated biological samples, and is routinely used for a variety of tasks, such as diagnosing infectious diseases, detecting hereditary diseases, identifying genetic fingerprints, cloning genes, paternity testing, genotyping, gene sequencing, and DNA computing.

In particular, digital PCR (Digital Polymerase Chain Reaction; dPCR) technology is a genetic testing method that divides existing samples into nanoliter-volume microdroplets and observes the fluorescence changes of target genes contained therein. This digital PCR has very high sensitivity and enables absolute quantitative analysis, so it is widely used in genetic analysis, biomarker development, and genetic base sequence analysis.

Meanwhile, digital PCR technology can complement the shortcomings of the second-generation RT (real-time PCR) by enabling absolute quantitative analysis, but since it is based on a geometrical optical reader, there may still be limitations in the analysis stage due to the occurrence of optical interference, the use of a separate fluorescence reader, and low fluorescence intensity.

Therefore, there is a continuous demand for the development of a new quantitative analysis system for target genes that can overcome the limitations of conventional digital PCR and enable high-precision analysis of target genes in samples.

The background technology of the invention has been written to facilitate understanding of the present invention. It should not be understood that the matters described in the background technology of the invention are recognized as prior art.

Meanwhile, the inventors of the present invention have noted the limitations of conventional digital PCR technology, which is based on a geometrical optical reader and is performed by flowing microdroplets, at which the PCR step is completed, into a channel-shaped detection section for quantitative analysis of target genes.

More specifically, the inventors of the present invention were able to recognize that, in the case of conventional digital PCR technology, since the intensity of fluorescent material for each micro droplet flowing to the detection unit is analyzed, the analysis time is long, an expensive PMT (Photo Multiplier Tube) is required, droplet reproduction may be required, and errors in quantitative analysis may occur due to light interference.

Accordingly, the inventors of the present invention noted that image-based quantitative analysis of microdroplets can solve the above-mentioned problems.

In particular, the inventors of the present invention noted that the limitations of conventional digital PCR can be overcome by acquiring images of microdroplets containing target genes and fluorescent substances and performing quantitative analysis.

As a result, the inventors of the present invention have developed a new target gene quantification system based on microdroplet images that enables detection of target genes within microdroplets without the flow of microdroplets and enables faster quantitative analysis of target genes.

At this time, the inventors of the present invention were able to design a target gene quantification system to perform quantitative analysis based on micro-droplet images reconstructed in a 3D form.

As a result, it was expected that precise quantitative analysis of microdroplets and target genes within microdroplets would be possible.

In particular, the inventors of the present invention expected that since analysis of microdroplet units is possible, detection of viruses present in trace amounts would be possible, and differentiation of asymptomatic viral infection patients would be easy.

In addition, the inventors of the present invention sought to further apply artificial intelligence-based algorithms to a novel target gene quantification system.

More specifically, the inventors of the present invention sought to apply an optical model learned to restore a low-resolution image to a high-resolution image to a new target gene quantification system, and expected that this would enable advancement of diagnostic technology.

In this regard, the inventors of the present invention expected that high-precision quantitative analysis of target genes would be possible even when using low-performance lenses.

Accordingly, the problem to be solved by the present invention is to provide a method for quantitative analysis of a target gene and a device for quantitative analysis of a target gene using the same, which is configured to receive a sheet image of a micro droplet based on a confocal lens, obtain a high-resolution sheet image using an optical model based on an artificial intelligence algorithm, and perform quantitative analysis on a target gene based on the sheet image.

The tasks of the present invention are not limited to the tasks mentioned above, and other tasks not mentioned will be clearly understood by those skilled in the art from the description below.

In order to solve the above-described task, a method for quantitative analysis of a target gene according to one embodiment of the present invention is provided.

A method for quantitative analysis of a target gene according to one embodiment of the present invention includes the steps of: obtaining a plurality of first sheet images of micro-droplets including a target gene and a fluorescent substance using a confocal lens-based quantitative analysis device; obtaining a plurality of second sheet images based on the plurality of first sheet images using an optical model based on an artificial intelligence algorithm; and performing quantitative analysis of the target gene based on fluorescence intensities of micro-droplets in the plurality of second sheet images.

According to a feature of the present invention, the step of acquiring a plurality of first sheet images may include a step of acquiring sheet images for microdroplets containing target genes and fluorescent substances for which PCR has been completed within the tube.

According to another feature of the present invention, the tubes are plural, and the step of obtaining a plurality of first sheet images may include the steps of arranging the plurality of tubes in a row, the step of controlling each of the plurality of tubes arranged in a row to move in one direction, and the step of obtaining a first sheet image for a tube including a microdroplet using a quantitative analysis device while moving in one direction.

According to another feature of the present invention, the method may further include, after the step of acquiring a plurality of second sheet images, a step of generating a three-dimensional image based on the plurality of second sheet images. In this case, the step of performing quantitative analysis may include a step of performing quantitative analysis of a target gene based on the fluorescence intensity of microdroplets within the three-dimensional image.

According to another feature of the present invention, the step of performing quantitative analysis may include the step of counting microdroplets expressing fluorescence within the three-dimensional image.

According to another feature of the present invention, the step of performing quantitative analysis may include the step of performing quantitative analysis of a target gene for each microdroplet based on fluorescence intensity within a three-dimensional image.

According to another feature of the present invention, the method may further include, after the step of generating a three-dimensional image, a step of determining the size of each microdroplet.

According to another feature of the present invention, the step of performing quantitative analysis may include the step of predicting the number of micro-droplets in the plurality of second sheet images using a prediction model configured to detect micro-droplets by taking as input images of micro-droplets.

According to another feature of the present invention, the optical model may be a model learned through a step of learning to reconstruct a low-resolution image for learning into a high-resolution image.

According to another feature of the present invention, the method may further include, after the learning step, a step of converting the reconstructed high-resolution image into a medium-resolution image for learning, and a step of re-learning the medium-resolution image for learning to reconstruct it into a high-resolution image.

According to another feature of the present invention, the step of acquiring a plurality of second sheet images may include a step of comparing the plurality of first sheet images with a previously stored reconstructed high-resolution image using an optical model to convert the same into a high-resolution second sheet image.

In order to solve the problem described above, a device for quantitative analysis of a target gene according to another embodiment of the present invention is provided. The device includes: a tube in which microdroplets containing a target gene for which polymerase chain reaction (PCR) has been completed and a fluorescent material are arranged; a light source for irradiating light to at least one surface of the tube; a confocal lens corresponding to at least one surface of the tube; an image sensor arranged in a different direction from the light irradiated through the light source and configured to provide a plurality of first sheet images of the microdroplets in the tube; and a processor operably connected to the image sensor. At this time, the processor is configured to acquire a plurality of second sheet images based on the plurality of first sheet images, and perform quantitative analysis of the target gene based on the fluorescence intensities of the microdroplets in the plurality of second sheet images, by using an optical model based on an artificial intelligence algorithm.

According to a feature of the present invention, the device may further include a tube control unit that arranges a tube and controls the arranged tube to move in one direction.

According to another feature of the present invention, the processor may be further configured to generate a three-dimensional image based on a plurality of second sheet images, and perform quantitative analysis of a target gene based on fluorescence intensity of micro-droplets within the three-dimensional image.

According to another feature of the present invention, the processor may be further configured to count microdroplets that exhibit fluorescence within the three-dimensional image.

According to another feature of the present invention, the processor may be further configured to perform quantitative analysis of a target gene for each microdroplet based on fluorescence intensity within the three-dimensional image.

According to another feature of the present invention, the processor may be further configured to predict the number of microdroplets within the plurality of second sheet images using a predictive model configured to detect microdroplets by taking as input images of microdroplets.

According to another feature of the present invention, the optical model may be a model learned through a step of learning to reconstruct a low-resolution image for learning into a high-resolution image.

Specific details of other embodiments are included in the detailed description and drawings.

The present invention can provide a novel target gene quantification system based on microdroplet images, which can detect target genes within microdroplets without the flow of microdroplets.

More specifically, the present invention can provide a target gene quantification system configured to perform quantitative analysis based on a high-resolution micro-droplet image reconstructed based on an artificial intelligence algorithm.

In particular, the present invention provides a target gene quantification system configured to perform quantitative analysis based on a micro-droplet image reconstructed in a 3D form, thereby enabling precise quantitative analysis of micro-droplets and, further, target genes within the micro-droplets.

Accordingly, the present invention has the effect of overcoming the limitations of conventional digital PCR technology, which requires a long analysis time by analyzing the intensity of fluorescent material for each microdroplet flowing into a detection channel, requires an expensive PMT (Photo Multiplier Tube), may require droplet reproduction, and may cause errors in quantitative analysis due to light interference.

In particular, the present invention can shorten the quantitative analysis time of a target gene compared to conventional digital PCR, thereby providing fast and precise analysis results for a large number of samples.

Furthermore, since the present invention enables analysis of microdroplet units, it enables detection of viruses present in trace amounts, and can provide ease in distinguishing asymptomatic viral infection patients.

The effects according to the present invention are not limited to those exemplified above, and more diverse effects are included in this specification.

FIGS. 1A and 1B illustrate the structure and configuration of a device for quantitative analysis of a target gene used in various embodiments of the present invention.
FIG. 2a illustrates an exemplary configuration of an integrated device based on a device for quantitative analysis of a target gene according to various embodiments of the present invention.
Figure 2b illustrates an example of microdroplets after a PCR reaction depending on the presence or absence of a target gene.
FIGS. 3A to 3D illustrate an example of a procedure for quantitative analysis of a target gene using a device for quantitative analysis of a target gene according to various embodiments of the present invention.
FIGS. 4A and 4B illustrate an example of a learning procedure of an optical model of a device for quantitative analysis of a target gene according to various embodiments of the present invention.

The advantages of the invention and the method for achieving them will become apparent by referring to the embodiments described in detail below together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and these embodiments are provided only to make the disclosure of the present invention complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined only by the scope of the claims.

The shapes, sizes, ratios, angles, numbers, etc. disclosed in the drawings for explaining embodiments of the present invention are exemplary, and therefore the present invention is not limited to the matters illustrated. In addition, in describing the present invention, if it is determined that a detailed description of a related known technology may unnecessarily obscure the gist of the present invention, the detailed description will be omitted. When the terms “includes,” “has,” “consists of,” etc. are used in this specification, other parts may be added unless “only” is used. When a component is expressed in the singular, it includes a case where the plural is included unless there is a specifically explicit description.

When interpreting components, it is interpreted as including the error range even if there is no separate explicit description.

The individual features of the various embodiments of the present invention may be partially or wholly combined or combined with each other, and as can be fully understood by those skilled in the art, various technical connections and operations are possible, and each embodiment may be implemented independently of each other or may be implemented together in a related relationship.

For clarity in the interpretation of this specification, the terms used in this specification are defined below.

The term "sample" as used herein may mean a fluid sample to be analyzed. For example, the fluid sample may be, but is not limited to, microdroplets containing a detection target, and further, cell lysates, whole blood, plasma, serum, saliva, ocular fluid, cerebrospinal fluid, sweat, urine, milk, ascites fluid, synovial fluid, and peritoneal fluid.

Meanwhile, the target gene in the sample may be a specific DNA or RNA. Preferably, the target gene may be RNA for a specific virus, but is not limited thereto.

As used herein, the term "microdroplet" refers to a microdroplet for digital PCR, which may contain a target gene (or non-target gene) to be amplified, a fluorescent substance, and a sample for PCR. Furthermore, the microdroplet may contain a target gene for which PCR has been completed and a fluorescent substance.

At this time, microdroplets can be generated by contact between the sample and oil having immiscible properties.

As used herein, the term "confocal lens" refers to a lens that selectively provides an optical image matching the focal length by using a special pinhole, and can provide a 3D image depending on the adjustment of the focal length.

That is, it is possible to acquire images of sheet-shaped microdroplets using a confocal lens and an image sensor.

The term "sheet image" as used herein may mean a plurality of slice images forming a 3D microdroplet. In this case, the sheet image may correspond to an optical sheet image obtained from a confocal lens-based device by a microdroplet that fluoresces when a target gene is amplified.

Meanwhile, according to a feature of the present invention, the sheet image can be obtained from a tube including microdroplets. That is, the sheet image can be an image of a tube in which microdroplets exist.

Additionally, the sheet image may be a still cut image of a tube moving in one direction.

As used herein, the term "first sheet image" may mean an original sheet image obtained from a confocal lens-based device, which may refer to a sheet image of a microdroplet (or a tube including a microdroplet) containing a target gene and a fluorescent material. In this case, the microdroplet may contain a target gene amplified according to a PCR reaction as described above, but is not limited thereto.

The term "second sheet image" as used herein may mean an image with improved resolution (or image quality) from the first sheet image. That is, the second sheet image may have a higher resolution than the first sheet image. Accordingly, the second sheet image may facilitate quantitative analysis of fine droplets than the first sheet image.

As used herein, the term “optical model” may be an artificial intelligence algorithm-based model configured to convert a low-resolution (or, low-quality) image into a high-resolution (or, high-quality) image.

More specifically, the optical model may be a model trained to take as input a training image of a low-resolution microdroplet and restore it to an original high-resolution microdroplet image.

That is, the optical model can be trained to reconstruct the first sheet image and output a high-resolution second sheet image.

At this time, the low-resolution image used for learning may be an image converted from a high-resolution image of a microdroplet (or an image of a tube including a microdroplet) to a low-resolution image.

According to a feature of the present invention, the optical model may be a CNN (Convolutional Neural Network)-based model using a convolutional layer. For example, the optical model may be based on a resolution restoration algorithm based on a GAN (Generative Adversarial Network), a VDSR (Very Deep Super Resolution), a SRUGAN (Super-Resolution Using a Generative Adversarial Network), a DCGAN (Deep Convolution GAN), an iGAN (Interactive GAN), or a StackGAN (Stacked GAN).

Accordingly, the optical model can provide precise quantitative analysis results for target genes within microdroplets under various shooting conditions, such as using low-performance lenses that provide low-quality images.

As used herein, the term "predictive model" may be a model trained to predict the number of microdroplets that express fluorescence within the image, given a sheet image of microdroplets as input.

That is, the prediction model can be configured to detect fluorescently expressed microdroplets within an image and provide quantitative values of target genes.

At this time, the prediction model may be based on a deep learning algorithm for image segmentation, such as, but not limited to, Segnet, Unet, faster rcnn, FCN or Voxnet.

Hereinafter, with reference to 1a and 1b, a device for quantitative analysis of a target gene and its configuration according to various embodiments of the present invention will be specifically described.

FIGS. 1A and 1B illustrate the structure and configuration of a device for quantitative analysis of a target gene used in various embodiments of the present invention.

First, referring to FIG. 1A, a device (1000) for quantitative analysis of a target gene according to one embodiment of the present invention may be composed of a tube (110) including microdroplets, particularly microdroplets in which PCR has been completed, and a fluorescent substance, a tube control unit (120) in which the tube (110) is placed and which controls the placed tube to move in one direction, a confocal lens (210), a light source (310), an image sensor (400) for providing a sheet image, and a processor (500) configured to communicate with the same.

At this time, the device (1000) for quantitative analysis of a target gene according to one embodiment of the present invention may further include an objective lens (220a, 220b) positioned in the vicinity of a tube (110) containing microdroplets, and a reflector (320) for changing the direction of a light source.

More specifically, sheet light is irradiated onto microdroplets on a tube (110) through a light source (310) and a reflector (320), and at the same time, the tube (110) can be moved in one direction or stopped by a tube control unit (120) for scanning. At this time, a plurality of first sheet images for the tube (110) moving at a constant speed or stopped can be acquired by a confocal lens (210), an objective lens (220), and an image sensor (400).

Meanwhile, the acquired plurality of first sheet images, as images of a tube including microdroplets, can correspond to optical sheet images of microdroplets that express fluorescence according to amplification of the target gene.

A plurality of first sheet images acquired from the image sensor (400) are transmitted to a processor (500) configured to communicate with the image sensor (400), and quantitative analysis of target genes can be performed based on the fluorescence intensity of microdroplets in the images by the processor (500).

At this time, the processor (500) may be based on an optical model based on an artificial intelligence algorithm.

More specifically, the received plurality of first sheet images can be converted into second sheet images by an optical model learned to reconstruct low-resolution images into high-resolution images. Accordingly, even if a low-resolution image is acquired from the image sensor (400), acquisition of a high-resolution sheet image can be possible by an optical model based on an artificial intelligence algorithm.

Optionally, the processor (500) can be configured to count the fluorescence-emitting microdroplets after converting the reconstructed image into 3D voxels, i.e., a 3D image.

According to a feature of the present invention, the processor (500) may be further configured to perform quantitative analysis of target genes for each microdroplet based on fluorescence intensity within the three-dimensional image.

According to another feature of the present invention, the processor (500) may be further configured to predict the number of microdroplets in the second sheet image using a prediction model learned to detect microdroplets by inputting an image of the microdroplets.

Meanwhile, the configuration and position of the device (1000) for quantitative analysis of a target gene according to one embodiment of the present invention are not limited to those described above.

For example, the image sensor (400) is not limited to being located at the rear of the objective lens (220b), but may be located in a different direction (e.g., perpendicular to) from the light source to obtain a sheet image of a microdroplet irradiated by the light source.

Furthermore, a plurality of confocal lenses (210) may be provided on the device (1000) for quantitative analysis of target genes.

Meanwhile, referring to FIG. 1b, according to another feature of the present invention, a device (1000') for quantitative analysis of a target gene may be provided with a plurality of tubes (110).

At this time, a plurality of tubes (110) can be arranged in a row in a tube control unit (not shown) for scanning microdroplets and then moved in the direction in which the sheet light is irradiated.

For example, while PCR is being performed, a plurality of tubes (110) that were grouped together for temperature control can be arranged in a row on a tube control unit for analysis of each tube (110) at the detection step of the target gene.

At this time, a plurality of tubes (110) can be moved between image sensors (400) positioned perpendicular to the sheet-light being investigated, and may be temporarily stopped during scanning to obtain a sheet image of a still cut of each tube (110).

However, it is not limited thereto, and a sheet image for each tube (110) may be acquired while a plurality of tubes (110) are arranged in a row and moving.

Furthermore, the direction of movement of the tube (110) by the tube control unit is not limited to the direction parallel to the image sensor (400), and can be set in various directions.

According to the structural features above, since analysis is possible only with the image of the tube containing the microdroplets where PCR is completed without the procedure of moving to the channel-type detection section of the microdroplets, precise and fast quantitative analysis of the target gene can be possible. Furthermore, since an artificial intelligence-based optical model and/or a prediction model can be applied, the reliability of the target gene test can be improved.

Hereinafter, with reference to FIGS. 2A and 2B, an integrated analysis device equipped with a device for quantitative analysis of a target gene according to various embodiments of the present invention will be described. FIG. 2A illustrates an exemplary configuration of an integrated device based on a device for quantitative analysis of a target gene according to various embodiments of the present invention. FIG. 2B illustrates an exemplary microdroplet after a PCR reaction depending on the presence or absence of a target gene.

The integrated analysis equipment (10000) can be composed of a micro-droplet generation unit (2000) and a PCR unit (3000) together with a device (1000) for quantitative analysis of a target gene according to various embodiments of the present invention.

More specifically, contact between a sample containing a target gene and oil can be induced in the microdroplet generation unit (2000). When the sample and the oil come into contact, a plurality of microdroplets encapsulated in the oil can be formed. At this time, the microdroplets containing the target gene can be generated. At this time, the microdroplets can contain the target gene, a fluorescent substance, a primer, and a polymerase.

The microdroplets generated from the microdroplet generation unit (2000) are moved to the PCR unit (3000). At this time, the microdroplets can be accommodated in a pre-equipped tube. Then, the microdroplets are provided with a temperature condition cycle for amplification of a target gene in the microdroplet, and as a result, the microdroplets in the tube can include the amplified target gene and a fluorescent substance.

More specifically, referring to FIG. 2b together, microdroplets containing a target gene, such as a viral gene, can express fluorescence along with amplification of the gene.

In contrast, for empty microdroplets and microdroplets containing normal genes, fluorescence expression may not occur even if PCR is performed by the PCR unit (3000).

Next, the tube containing the microdroplets in which PCR has been completed can be moved by the tube control unit to a device (1000) for quantitative analysis of a target gene according to various embodiments of the present invention.

At this time, multiple first sheet images and high-resolution second sheet images of the tube are sequentially acquired by a device (1000) for quantitative analysis of the target gene, and a three-dimensional image is reconstructed. Then, the number of micro droplets containing the target gene is determined from the reconstructed three-dimensional image.

That is, the integrated analysis equipment (10000) can be configured to perform not only generation of microdroplets containing a target gene and amplification of the target gene within the microdroplets, but also quantitative analysis of the target gene.

Accordingly, molecular biological analysis of target genes is possible without user handling, and highly reliable quantitative analysis results can be obtained.

Hereinafter, with reference to FIGS. 3a to 3d, a procedure for a method for quantitative analysis of a target gene using a device for quantitative analysis of a target gene according to various embodiments of the present invention will be described.

FIGS. 3A to 3D illustrate an example of a procedure for a quantitative analysis method of a target gene using a device for quantitative analysis of a target gene according to various embodiments of the present invention.

First, referring to FIG. 3a, first, a plurality of first sheet images for micro droplets are acquired from a quantitative analysis device based on a confocal lens (S210), and a plurality of second sheet images for the plurality of first sheet images are acquired by an optical model (S220). Finally, quantitative analysis of a target gene is performed based on the fluorescence intensity of the micro droplets in the second sheet images (S230).

More specifically, referring to FIG. 3b together, in step (S210) where a plurality of first sheet images are acquired, a plurality of first sheet images (612) for microdroplets are acquired by a quantitative analysis device (100) according to various embodiments of the present invention based on a confocal lens.

At this time, a plurality of first sheet images (612) can correspond to images for the tube.

Furthermore, the plurality of first sheet images (612) may be low-resolution images.

Next, in the step (S220) where the second sheet image is acquired, a plurality of first sheet images (612) are input into the artificial intelligence-based optical model (510), and a plurality of high-resolution second sheet images (614) are acquired. Thereafter, the high-resolution second sheet images (614) are reconstructed into 3D voxels.

That is, in the step (S220) where the second sheet image is acquired, a low-resolution image can be restored to a high-resolution image by the optical model (510).

At this time, the optical model (510) may be an artificial intelligence algorithm-based model configured to convert a low-resolution (or, low-quality) image into a high-resolution (or, high-quality) image.

More specifically, referring together with FIG. 3c, in the step (S220) where the second sheet image is acquired, a plurality of low-quality first sheet images (612) are input to the optical model (510) by sheet-light irradiated on a tube including microdroplets. Then, the optical model (510) compares the input plurality of first sheet images (612) with the second sheet images stored in the learning step to output a plurality of high-resolution second sheet images (614). The plurality of high-resolution second sheet images (614) thus output are converted into 3D voxels, and as a result, a 3D image (616) of the microdroplet can be acquired.

Referring again to FIG. 3a, in step (S230) where quantitative analysis of a target gene is performed, quantitative analysis of the target gene can be performed based on the fluorescence intensity of micro-droplets within the 3D image.

According to a feature of the present invention, in a step (S230) in which quantitative analysis of a target gene is performed, detection of micro-droplets in a second sheet image can be performed using a prediction model configured to predict the number of micro-droplets by inputting an image of the micro-droplets.

More specifically, referring to FIG. 3d, in the step (S230) where quantitative analysis of a target gene is performed, a reconstructed 3D image (616) is input into an artificial intelligence-based prediction model (520), and a quantitative analysis result (618) for a microdroplet containing a target gene and a fluorescent substance can be output from the prediction model (520).

At this time, the prediction model may be a model trained to input a sheet image of microdroplets, segment microdroplets that express fluorescence within the image, and predict the number of microdroplets from this.

According to a feature of the present invention, the prediction model (520) may be based on a deep learning algorithm for image segmentation based on Segnet, Unet, faster rcnn, FCN or Voxnet, but is not limited thereto.

Accordingly, the quantitative analysis results (618) for the microdroplets may include the results of segmenting the regions of the microdroplets in which fluorescence is expressed within the 3D image (616), and may further include the results of counting the microdroplets in which fluorescence is expressed, i.e., the microdroplets containing the target gene and the fluorescent substance.

Meanwhile, according to a feature of the present invention, in the step (S230) where quantitative analysis of a target gene is performed, quantitative analysis of the target gene for each microdroplet can be performed based on the fluorescence intensity within the three-dimensional image (616).

That is, by the quantitative analysis method according to various embodiments of the present invention, the fluorescence intensity for each microdroplet can be determined, so that not only quantitative analysis of the target gene for the entire sample but also quantitative analysis of the target gene for each microdroplet can be possible. Accordingly, precise analysis results for the target gene can be provided.

According to another feature of the present invention, in the step (S230) where quantitative analysis of a target gene is performed, the size of each microdroplet may be determined based on the fluorescence intensity.

Therefore, according to various embodiments of the present invention, a device for quantitative analysis of a target gene based on a tube image and a quantitative analysis method using the same can perform quantitative analysis based on a micro-droplet image reconstructed in a 3D form, thereby providing precise quantitative analysis results for micro-droplets and, further, target genes within the micro-droplets.

In particular, the present invention can overcome the limitations of conventional digital PCR technology, which requires a long analysis time, requires an expensive PMT (Photo Multiplier Tube), may require droplet reproduction, and may cause errors in quantitative analysis due to light interference as a factor, by analyzing the intensity of a fluorescent material for each microdroplet flowing into a detection channel.

Hereinafter, a learning method of an optical model used in various embodiments of the present invention is described with reference to FIGS. 4a and 4b.

FIGS. 4A and 4B illustrate an example of a learning procedure of an optical model of a device for quantitative analysis of a target gene according to various embodiments of the present invention.

First, referring to Fig. 4a, a second sheet image for a microdroplet is acquired for learning an optical model (S310), then the second sheet image is converted into a low-resolution image for learning (S320). Then, the optical model is trained to reconstruct the low-resolution image for learning into a second sheet image (S330).

More specifically, referring together with FIG. 4b, in the step (S310) of acquiring the second sheet image, a high-resolution sheet image (712) with a low FOV (field of view) is acquired. Then, in order to acquire training data of the optical model (510), a step (S320) of converting into a low-resolution image for training is performed. As a result, a low-resolution training sheet image (714) is acquired. At this time, various conversions such as a ratio and a size may be performed on the low-resolution training sheet image (714) in order to diversify the training data. Then, in the learning step (S330), the converted training sheet image (716) is input to the optical model (510), and the optical model (510) is trained to restore the converted training sheet image (716) into an image corresponding to the high-resolution sheet image (712).

That is, the optical model (510) is trained to output a high-resolution sheet image (718) reconstructed from a low-resolution training sheet image (714). Optionally, the reconstructed high-resolution sheet image (718) can be converted into a medium-quality sheet image (719) and reused for training the optical model (510).

The optical model (510) learned by the above procedure can provide precise quantitative analysis results for target genes in microdroplets under various shooting conditions equipped with low-performance lenses or image sensors that provide low-quality images.

At this time, the optical model (510) may be a CNN (Convolutional Neural Network)-based model using a convolutional layer. For example, the optical model may be based on a resolution restoration algorithm based on a GAN (Generative Adversarial Network), a VDSR (Very Deep Super Resolution), a SRUGAN (Super-Resolution Using a Generative Adversarial Network), a DCGAN (Deep Convolution GAN), an iGAN (Interactive GAN), or a StackGAN (Stacked GAN).

However, the learning procedure and algorithm of the optical model (510) are not limited to those described above.

Although the embodiments of the present invention have been described in more detail with reference to the attached drawings, the present invention is not necessarily limited to these embodiments, and various modifications may be made without departing from the technical idea of the present invention. Accordingly, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention, but to explain it, and the scope of the technical idea of the present invention is not limited by these embodiments. Therefore, it should be understood that the embodiments described above are exemplary in all aspects and not restrictive. The protection scope of the present invention should be interpreted by the following claims, and all technical ideas within a scope equivalent thereto should be interpreted as being included in the scope of the rights of the present invention.

10000: Integrated Analysis Equipment
1000, 1000': Device for quantitative analysis of target genes
110: Tube
120: Tube Control Unit
210: Confocal lens
220a, 220b: Objective lens
310: Light source
320: Reflector
400: Image sensor
500: Processor
510: Optical Model
520: Prediction Model
612: Multiple First Sheet Images
614: Multiple high resolution second sheet images
616: Reconstructed 3D image
618: Quantitative analysis results
712: High resolution sheet image
714: Low-resolution image of learning sheet
716: Converted learning sheet image
718: Reconstructed high-resolution sheet image
719: Medium-quality sheet image
2000: Microdroplet Generation Unit
3000: PCR Unit

Claims (18)

A step of acquiring a plurality of first sheet images of microdroplets containing a target gene and a fluorescent substance using a confocal lens-based quantitative analysis device;
A step of obtaining a plurality of high-resolution second sheet images based on the plurality of first sheet images using an optical model based on an artificial intelligence algorithm, and
A method for quantitative analysis of a target gene, comprising the step of predicting the number of micro-droplets in the plurality of second sheet images using a prediction model configured to detect micro-droplets by inputting images of micro-droplets. In the first paragraph,
The step of obtaining the plurality of first sheet images is:
A method for quantitative analysis of a target gene, comprising the step of obtaining a sheet image for a tube containing microdroplets containing a target gene and a fluorescent material for which gene amplification has been completed. In the second paragraph,
The above tubes are plural,
The step of obtaining the plurality of first sheet images is:
A step of arranging the above plurality of tubes in a row;
A step of controlling each of a plurality of tubes arranged in a row to move in one direction, and
A method for quantitative analysis of a target gene, comprising the step of obtaining a first sheet image for a tube containing the microdroplet using the quantitative analysis device while moving in one direction. In the first paragraph,
After the step of obtaining the plurality of second sheet images,
Further comprising a step of generating a three-dimensional image based on the plurality of second sheet images,
The steps for performing the above quantitative analysis are:
A method for quantitative analysis of a target gene, comprising the step of performing quantitative analysis of the target gene based on the fluorescence intensity of micro-droplets within the three-dimensional image. In paragraph 4,
The steps for performing the above quantitative analysis are:
A method for quantitative analysis of a target gene, comprising a step of counting microdroplets expressing fluorescence within the three-dimensional image. In paragraph 4,
The steps for performing the above quantitative analysis are:
A method for quantitative analysis of a target gene, comprising the step of performing quantitative analysis of a target gene for each microdroplet based on the fluorescence intensity within the three-dimensional image. In paragraph 4,
After the step of generating the above 3D image,
A method for quantitative analysis of a target gene, further comprising the step of determining the size of each microdroplet. delete In the first paragraph,
The above optical model is,
A method for quantitative analysis of target genes, wherein the model is trained through a step of learning to reconstruct low-resolution images for training into high-resolution images. In Article 9,
After the above learning steps,
A step of converting the reconstructed high-resolution image into a medium-resolution image for training, and
A method for quantitative analysis of a target gene, further comprising a step of retraining a medium-resolution image for training to be reconstructed into a high-resolution image. In Article 9,
The step of obtaining the plurality of second sheet images is:
A method for quantitative analysis of a target gene, comprising the step of comparing the plurality of first sheet images with the previously stored reconstructed second sheet images using the optical model to convert them into a high-resolution second sheet image. A light source that irradiates light to at least one side of a tube in which microdroplets containing target genes and fluorescent substances in which PCR (polymerase chain reaction) has been completed are placed;
A confocal lens corresponding to at least one surface of the tube;
An image sensor arranged in a different direction from the light irradiated through the light source and configured to provide a plurality of first sheet images of microdroplets within the tube, and
A processor operatively connected to the image sensor,
The above processor,
Using an optical model based on an artificial intelligence algorithm, a plurality of high-resolution second sheet images are obtained based on the plurality of first sheet images,
A device for quantitative analysis of a target gene, configured to predict the number of micro-droplets in the plurality of second sheet images using a prediction model configured to detect micro-droplets by inputting images of micro-droplets. In Article 12,
A device for quantitative analysis of a target gene, further comprising a tube control unit that controls the tubes to be placed and move in one direction. In Article 12,
The above processor,
A device for quantitative analysis of a target gene, further configured to generate a three-dimensional image based on the plurality of second sheet images and perform quantitative analysis of the target gene based on the fluorescence intensity of micro-droplets within the three-dimensional image. In Article 14,
The above processor,
A device for quantitative analysis of a target gene, further configured to count micro-droplets expressing fluorescence within the three-dimensional image. In Article 14,
The above processor,
A device for quantitative analysis of a target gene, further configured to perform quantitative analysis of the target gene for each microdroplet based on the fluorescence intensity within the three-dimensional image. delete In Article 12,
The above optical model is,
A device for quantitative analysis of target genes, wherein the model is trained through a step of learning to reconstruct low-resolution images for training into high-resolution images.

Images