KR20190127247A - Method and program for provideing diagnostic image by early dynamic image data - Google Patents

Abstract

The present invention relates to a method and a program for generating a diagnostic image based on initial dynamic image data. According to an embodiment of the present invention, the method for generating a diagnostic image based on initial dynamic image data comprises: a step of allowing a computing system to acquire initial dynamic image data for learning and delay image data for learning for at least one patient; and a step of allowing the computing system to generate a diagnostic image prediction model by learning the initial dynamic image data for learning and the delay image data for learning. The diagnostic image prediction model calculates new delay image data based on initial dynamic image data for diagnosis of a new patient. The initial dynamic image data is data obtained during a specific first time in an initial range after drug injection and including one or more image frames. The delay image data is obtained after a reference time and is a diagnostic image used for patient diagnosis. The present invention can reduce time of patients spent in hospitals for PET inspection.

Description

Diagnosis image generation method and program based on initial dynamic image data {METHOD AND PROGRAM FOR PROVIDEING DIAGNOSTIC IMAGE BY EARLY DYNAMIC IMAGE DATA}

The present invention relates to a method and program for generating an image for diagnosis based on initial dynamic image data, and more specifically, to predict an image for diagnosis after a lapse of a reference time used for patient diagnosis based on initial dynamic image data acquired in an initial time range. It relates to a method and a program for generating.

The Positron Emission Computed Tomography (PET) test is a state-of-the-art nuclear imaging technique that administers positron emitting radioisotopes and obtains radiation emitted outside the body to obtain useful diagnostic information regarding metabolic changes and receptor distribution in the body. . In recent years, the company has been gradually evolving into a fused hybrid scanner, from acquiring only PET images to computing CT (Computed Tomography) or MRI (Magnetic Resonance Image) equipment. In the current PET test, the PET / CT scanner, which combines the PET device and the CT device in one fuselage, is used as a basis, and additionally, anatomical information of the CT image is obtained, so that the exact location and depth of the lesion identified in the PET image can be obtained. It was possible to provide.

PET testing typically involves injecting a radiolabeled tracer into a human body, followed by a specific time (eg, 2 hours to 2 hours) when the tracer specific and nonspecific bonds reach a stable state or the difference is maximized. Patient diagnosis is performed based on the PET image acquired after 3 hours). That is, a PET image without the influence of blood flow is obtained (for example, when FP-CIT is used as a tracer) or a PET image with a large difference in ratio between regions is obtained (for example, a FDG is tracer). In order to obtain a PET image after a specific time has elapsed. Specifically, the radiopharmaceuticals (Radiopharmaceuticals, Radiotracer or Radioligand) used in the clinically widely used tests of nuclear medicine images are different from each other in the body, but most of them are taken more than 1 hour after the injection and delayed Acquire image data. Bone scan takes about 3 hours, FDG PET takes about 1 hour, dopamine carrier PET takes about 3 hours, amyloid PET takes about 90 minutes, and then the image is taken for delay image acquisition. To this end, the patient will come to the hospital to receive the diagnostic medicine (ie, the tracer) and wait until the reference point to take the PET image. Accordingly, for a long time before reaching Steady State, the patient should stay in the waiting room so that other patients and caregivers are not exposed to radiation by the administered medication. In particular, when the FDG is used as a tracer, when the brain is stimulated or moves voluntarily or involuntarily during the waiting time, the glucose carrier and hexokinase activity are increased to increase glucose metabolism to generate distortion in the image. Wait in a dark room for a certain amount of time without moving.

The present invention uses a diagnostic image prediction model that learns initial dynamic image data and delay image data for a plurality of patients, and does not wait until the time when the medical staff photographs the delay image data for diagnosis. An object of the present invention is to provide a diagnostic image generation method and program based on initial dynamic image data, which predict and provide diagnostic delay image data using data.

Problems to be solved by the present invention are not limited to the above-mentioned problems, and other problems not mentioned will be clearly understood by those skilled in the art from the following description.

According to an embodiment of the present invention, a method for generating diagnostic image based on initial dynamic image data may include: acquiring, by a computing system, initial dynamic image data for learning and delay image data for one or more patients; And generating, by a computing system, a diagnostic image prediction model by learning the training initial dynamic image data and the training delay image data, wherein the diagnostic image prediction model is based on diagnostic initial dynamic image data for a new patient. New delay image data is calculated. The initial dynamic image data is data obtained during a first specific time in an initial range after drug injection, and includes data including one or more image frames. The delay image data is acquired after a reference time. It is a diagnostic image used for diagnosing a patient.

In another embodiment, the image data is Positron Emission Tomography image data, the drug is characterized in that the tracer.

In another embodiment, the initial dynamic image data is characterized in that the image quality of the image frame within the first time is set according to the required number of frames.

In another embodiment, when the drug used to capture image data is a tracer coupled to a specific target region, the delay image data is obtained when the influence of blood flow is excluded, and the initial dynamic image data is After the point at which the influence of blood flow in the reference area begins to decrease, or around the time point when the dose ratio difference between the target region and the reference region becomes more than a specific value or the ratio difference value becomes the maximum.

In another embodiment, the method may further comprise: obtaining initial dynamic image data for initial setting and delay image data for initial setting during an initial time range; And calculating the optimal first time included in the initial time range by learning the initial dynamic image data for the initial setting and the delay image data for the initial setting in the initial time range.

In another embodiment, the generating of the diagnostic image predictive model may include learning each pixel in the delay image frame by matching a change of each pixel in the initial dynamic image data over time, or initializing over time. Delay image data may be learned by using each image frame in the dynamic image data as a multi-channel input.

In another embodiment, when the amount of the medicine used when acquiring the training data and the diagnosis is different, the new delay image data may be different in brightness or color between the initial dynamic image data for learning and the initial dynamic image data for diagnosis. Characterized in that it is generated by reflecting.

In another embodiment, the new delay image data may be generated by normalizing the brightness of the training initial dynamic image data and the diagnostic initial dynamic image data based on a maximum value or an average value.

According to another aspect of the present invention, there is provided a method of generating an image for diagnosis based on initial dynamic image data, the method comprising: acquiring, by a computing system, initial dynamic image data and learning delay image data for one or more patients; Generating, by a computing system, a diagnostic image prediction model by learning the training initial dynamic image data and the training delay image data; And calculating, by the computing system, new delay image data by inputting initial diagnostic dynamic image data of a new patient to the diagnostic image predictive model, wherein the initial dynamic image data is specified in an initial range after drug injection. The data obtained for one hour is data including one or more image frames, and the delay image data is obtained after a reference time and is a diagnostic image used for patient diagnosis.

In accordance with another aspect of the present invention, there is provided a method for generating a diagnostic image based on initial dynamic image data, the method comprising: receiving, by a computing system, initial dynamic image data for a new patient in a diagnostic image prediction model; And calculating new delay image data based on the initial dynamic image data, wherein the diagnostic image prediction model acquires initial dynamic image data for training and learning delay image data for one or more patients, and then learns. The initial dynamic image data is data obtained during a specific first time in an initial range after drug injection, and includes data including one or more image frames, and the delay image data is obtained after a reference time. This is a diagnostic image used for.

In another embodiment, when the amount of drugs used for acquiring and diagnosing the training data is different, the step of calculating the new delay image data may include the brightness of the initial dynamic image data for training and the initial dynamic image data for diagnosis, or the like. The new delay image data may be generated by reflecting color differences.

The initial dynamic image data-based diagnostic image generation program according to another embodiment of the present invention is combined with hardware to execute the above-mentioned initial dynamic image data-based diagnostic image generation method and stored in a medium.

According to the present invention as described above, has the following various effects.

First, the patient does not have to wait until the baseline time after administering the medication (eg, a tracer), thereby reducing the time patients spend in the hospital for PET testing.

Second, since a small amount of drug can be used to acquire the initial dynamic image data for diagnosis as compared to when the drug is injected to directly obtain the delay image data, the amount of radiation affecting the patient's body can be reduced.

1 is a method for generating an image for diagnosis based on initial dynamic image data according to an embodiment of the present invention.
2 is a graph showing a change in image data and image data obtained by inputting the FP-CIT according to an embodiment of the present invention.
3 is a graph illustrating a change in image data and image data obtained by inputting an FDG according to an embodiment of the present invention.
4 is an initial dynamic image data-based diagnostic image generation method further comprising a first time range calculation process according to an embodiment of the present invention.
5 is a method for generating an image for diagnosis based on initial dynamic image data according to another embodiment of the present invention.
6 is a flow chart of a method for generating an image for diagnosis based on initial dynamic image data according to another embodiment of the present invention.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Advantages and features of the present invention and methods for achieving them will be apparent with reference to the embodiments described below in detail with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various forms, and only the embodiments are intended to complete the disclosure of the present invention, and the general knowledge in the art to which the present invention pertains. It is provided to fully inform the person having the scope of the invention, which is defined only by the scope of the claims. Like reference numerals refer to like elements throughout.

Unless otherwise defined, all terms (including technical and scientific terms) used in the present specification may be used in a sense that can be commonly understood by those skilled in the art. In addition, the terms defined in the commonly used dictionaries are not ideally or excessively interpreted unless they are specifically defined clearly.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. In this specification, the singular also includes the plural unless specifically stated otherwise in the phrase. As used herein, "comprises" and / or "comprising" does not exclude the presence or addition of one or more other components in addition to the mentioned components.

In the present specification, the 'computing system' includes all the various devices for performing arithmetic processing. A computing system may include one or more computers. For example, a computer can be a desktop PC, a notebook, as well as a smartphone, a tablet PC, a cellular phone, a PCS phone (Personal Communication Service phone), synchronous / asynchronous The mobile terminal of the International Mobile Telecommunication-2000 (IMT-2000), a Palm Personal Computer (PC), a Personal Digital Assistant (PDA), and the like may also be applicable. The computer may also correspond to medical equipment for acquiring or observing medical images. In addition, the computer may correspond to a server computer connected to various client computers.

In the present specification, 'image data' means an image obtained by the medical imaging apparatus.

As used herein, the term "medical imaging apparatus" refers to a device used to acquire a medical image. For example, the medical imaging apparatus may include a Positron Emission Tomography imaging apparatus, a Magnetic Resonance Imaging (MRI) imaging apparatus, or the like.

In the present specification, 'delay image data' is obtained after a reference time, and means a diagnostic image used for patient diagnosis.

In the present specification, the 'reference time' means a time from an initial time of injecting a medicine (eg, contrast agent or tracer) to a time point (ie, reference time point) at which image data capable of diagnosing a patient's condition can be obtained.

In the present specification, "drug" means injected into the body at the time of taking medical image data. For example, 'Drug' is a tracer used for magnetic resonance imaging (MRI) imaging or computed tomography (CT) imaging, and Positron Emission Tomography imaging. (Tracer) may be applicable.

In the present specification, 'initial dynamic image data' means image data including a plurality of consecutive image frames. 'Initial dynamic image data' is obtained before the reference time point at which the delay image data is acquired, and includes an initial time range (for example, a time range shortly after inserting a contrast agent or tracer administered during imaging). Is to be obtained.

In the present specification, 'learning initial dynamic image data' means initial dynamic image data included in learning data used for learning a diagnostic image prediction model.

In the present specification, 'learning delay image data' means delay image data included in learning data used for learning a diagnostic image prediction model.

In the present specification, 'diagnostic initial dynamic image data' means initial dynamic image data acquired for calculating delay image data of a specific patient.

In the present specification, 'new delay image data' refers to delay image data calculated through a diagnostic image prediction model for a specific patient diagnosis.

In the present specification, the 'first time' refers to a time range in which initial dynamic image data is obtained. That is, the 'first time' refers to a time range from the time when the initial dynamic image data is acquired to the time when the last image frame is acquired.

In the present specification, the 'initial time range' refers to a time range for extracting a first time for acquiring learning initial dynamic image data.

In the present specification, 'initial dynamic image data for initial setting' means image data included in a plurality of image frames consecutive in the initial time range.

In the present specification, 'initial setting delay image data' refers to delay image data used to calculate a first time suitable for learning a diagnostic image prediction model. 'Delay image data for initial setting' may be the same as the delay image data for learning.

Hereinafter, a detailed description of an initial dynamic image data-based diagnostic image generation method and program according to embodiments of the present invention will be described with reference to the accompanying drawings.

1 is a method for generating an image for diagnosis based on initial dynamic image data according to an embodiment of the present invention.

Referring to FIG. 1, in the initial dynamic image data-based diagnostic image generation method according to an embodiment of the present invention, the computing system acquires initial dynamic image data and learning delay image data for one or more patients (S400). ; And generating, by the computing system, a diagnostic image prediction model by learning the training initial dynamic image data and the training delay image data (S600).

The computing system acquires learning initial dynamic image data and learning delay image data for one or more patients (S400). The computing system constructs the learning data by acquiring the learning initial dynamic image data and the learning delay image data from the client device. For example, the computing system receives initial dynamic image data and delay image data for a specific patient from the medical imaging apparatus or a client device connected to the medical imaging apparatus when the initial dynamic image data and the delay image data are captured for the specific patient. do.

In one embodiment, the image data is Positron Emission Tomography (PET) image data. The medical staff has a certain time after the input of a radiotracer (radioactive material used in the human body observation) containing a ligand that attaches or reacts to a specific target region or a specific substance (for example, dopamine carriers, glucose carriers) Afterwards, the image is generally taken to obtain delay image data as PET image data, and diagnostics are performed therethrough.

Depending on the type of tracer used in Positron emission tomography, the type of initial dynamic image data acquired by the computing system may vary.

In one embodiment, when using a tracer (Tracer) that decreases the radiation concentration of the area other than the target area over time, the initial dynamic image data for learning is the first time (ie, initial drug injection) when blood flow effects begin to decrease The learning delay image data is acquired at a reference time point at which the amount of radiation is confirmed or determined by ligand characteristics without the influence of blood flow.

For example, as shown in FIG. 2, when FP-CIT is used as a drug (ie, a tracer), in the initial time range in which the drug is administered, the radiation dose is high in the whole region of the brain due to blood flow effects, and thus diagnosis is difficult. However, over time, the drug is metabolized into the urine or liver and excreted, and then the concentration in the plasma drops, so the drug is lost in the blood and the drug is released from the brain tissue to balance to balance. The computer system acquires initial dynamic image data over a first time range at which the radiation dose in the brain tissue changes. Since FP-CIT is effective in differentiating Parkinson's disease due to its strong binding ability with dopamine, the effects of blood flow on other areas are reduced as time passes as FP-CIT is bound to the dopamine neurotransmitter. Even if the concentration is lowered, the target portion maintains a high concentration. Therefore, the computing system acquires delay image data at a time point at which the influence of blood flow decreases after a reference time passes after drug administration. The delay image data may be image data used by a medical staff to diagnose the size, shape, and the like of a target portion.

In another embodiment, when the medicine used to capture image data is a tracer coupled to a specific target region, the first time range for acquiring the initial dynamic image data may be determined based on various criteria. For example, the first time range may be set to a time range after a time point at which the influence of blood flow in the reference region begins to decrease. The reference region is a specific region of the brain that generally has no or little substance to which the tracer binds and thus does not differ between diseases. The reference area depends on the type and characteristics of the tracer. For example, when the tracer is FP-CIT, the reference region is a site having few dopamine neurotransmitters such as cerebellum and occipital cortex. In another example, the first time range may be set around a time point at which the dose ratio difference value between the target region and the reference region becomes equal to or greater than a specific value or the ratio difference value reaches a maximum value.

In another embodiment, in the case of a tracer (eg, Fluorodeoxyglucose (hereinafter referred to as FDG) PET) in which the amount of chemicals used to capture image data is continuously increased to bind to a target area over time, there is no peak point. Since the dose of each region of the brain is increased, the computing system acquires the initial dynamic image data in the initial first time range where the difference in radiation dose is not large, and delays at the reference time point where the ratio of radiation dose difference between regions is increased. Acquire image data. In other words, in order to clearly identify the difference between the regions of the brain tissue, the medical staff previously diagnosed the patient with the image data obtained at the baseline point at which the radiation dose rate difference is large, so that the computing system acquired the image data at the baseline point. Is used as learning delay image data.

For example, Fluorodeoxyglucose (hereinafter referred to as FDG) has the same chemical properties in vivo as deoxyglucose (Deoxyglucose). Deoxyglucose, like glucose, is introduced into cells through the glucose transporter of vascular endothelial cells of the brain bloodstream barrier and the glucose transporter of neurons and astrocytes, and then phosphate groups from ATP to hexokinase (glucose or other hexose). Transfers to a deoxyglucose-6-P by the action of a corresponding hexose-6-phosphate-catalyzed enzyme). FDG PET reflects both glucose transporter and hexokinase activity during glucose metabolism. That is, as shown in FIG. 3, as the time passes, the amount of FDG introduced into the brain cells increases, and the dose provided by each cell increases with time. Accordingly, the computing system acquires initial dynamic image data for learning in a first time range close to the initial time point, and obtains a delay image at a reference time point when a dose difference of each region increases after sufficient time passes.

The computing system may intentionally acquire the initial dynamic image data for learning and the delay image data for learning by performing two shots in the first time range and the reference time point, and the initial phase and the reference time point (ie In addition, the initial dynamic image data for learning and the delay image data for learning may be acquired while acquiring images at a delay phase. For example, medical staff may compare early FPCIT images (ie, early phase FPCIT PET images) with Delay FPCIT images (ie, FPCIT PET images obtained during the Delay phase) to determine the cause of Parkinson's disease. Since the FPCIT image and the Delay FPCIT image may be acquired, when the Early FPCIT image is acquired as the initial dynamic image data in the first time range, the computing system may acquire the initial dynamic image data for learning and the delay image data for learning.

In another embodiment, the initial dynamic image data is set according to the required number of frames, the image quality of the image frame within the first time. In other words, if the same amount of drug is used, the maximum amount of radiation emitted to the outside is limited, and thus the amount of signal used to generate the image frame is limited. Therefore, the computing system may adjust the image quality of each image frame according to the number of image frames based on the same signal amount. Can be set differently. For example, if the number of video frames is increased, the time length used for generating each video frame is shortened, and thus the amount of signals available for generating one video frame is reduced, and the computing system generates low quality of each video frame.

A computing system learns the training initial dynamic image data and the training delay image data to generate a diagnostic image prediction model (S600; generating a diagnostic image prediction model). The diagnostic image predictive model calculates new delay image data based on initial dynamic image data for diagnosis of a new patient.

In one embodiment, the computing system learns by matching each pixel in the delay image frame with the change of each pixel in the initial dynamic image data over time. That is, in a combination of image data for a particular patient (i.e., a combination of initial dynamic image data and delay image data), the computing system is initially dynamic for each point (i.e., pixels) of body tissue (e.g., brain tissue). The data set for each image data combination is constructed by matching the characteristic change information in the image data with the pixel characteristic information in the delay image data, and the data sets for a plurality of patients are learned for each point (that is, pixels) for diagnosis. Construct an image prediction model.

In one embodiment, the diagnostic image prediction model is constructed of a deep neural network (DNN). That is, the diagnostic image prediction model learns initial learning dynamic image data and learning delay image data for one or more patients by applying a deep learning algorithm.

The deep neural network (DNN) refers to a system or a network in which one or more layers are built in one or more computers to perform determination based on a plurality of data. For example, the deep neural network may be implemented as a set of layers including a convolutional pooling layer, a locally-connected layer, and a fully-connected layer. The convolution pooling layer or the local access layer can be configured to extract features in the image. The fully connected layer can determine the correlation between the features of the image. In some embodiments, the overall structure of the deep neural network may be in the form of a local connection layer followed by a convolution pooling layer and a complete connection layer in the local connection layer. The deep neural network may include various criteria (ie, parameters), and may add new criteria (ie, parameters) through an input image analysis.

In one embodiment, the deep neural network, a structure called a convolutional neural network suitable for image analysis, and feature extraction layer (Feature Extraction Layer) for self-learning the features of the largest discriminant power (Discriminative Power) from the given image data Based on the extracted feature, a prediction layer learning a prediction model to have the highest prediction performance may be configured as an integrated structure.

The feature extraction layer spatially integrates a convolution layer and a feature map to create a feature map by applying a plurality of filters to each region of an image, thereby allowing a feature to be changed to change in position or rotation. It can be formed in a structure that repeats the alternating layer (Pooling Layer) to be extracted several times. Through this, it is possible to extract various levels of features, from low level features such as points, lines and planes to complex high level features.

The convolutional layer obtains a feature map by taking a non-linear activation function for the inner product of the filter and the local receptive field for each patch of the input image. In comparison, CNN is characterized by the use of filters with sparse connectivity and shared weights. This connection structure reduces the number of parameters to be learned and makes the learning through the backpropagation algorithm more efficient, resulting in better prediction performance.

The pooling layer or sub-sampling layer creates a new feature map using the local information of the feature map obtained from the previous convolutional layer. In general, the feature map newly created by the integration layer is reduced to a smaller size than the original feature map. Typical integration methods include the maximum pooling to select the maximum value of the corresponding area in the feature map and the corresponding feature map in the feature map. Average pooling, which finds the mean of a domain. The feature map of the unified layer is generally less affected by the location of any structure or pattern present in the input image than the feature map of the previous layer. That is, the integrated layer can extract features that are more robust to regional changes such as noise or distortion in the input image or previous feature maps, and these features can play an important role in classification performance. Another role of the integration layer is to allow for the reflection of a wider range of features as one goes up to the upper learning layer in the deeper structure, and as feature extraction layers accumulate, the lower layer reflects local features and rises to the upper layer. Increasingly, it is possible to generate a feature that reflects a feature of the entire abstract image.

As such, the final feature extracted through the iteration of the convolutional layer and the integrated layer is that the classification model such as Multi-layer Perception (MLP) or Support Vector Machine (SVM) is fully connected. Combined in the form of -connected layer, it can be used for classification model training and prediction.

However, the structure of the deep neural network according to the embodiments of the present invention is not limited thereto, and may be formed of a neural network having various structures.

In one embodiment, in the generation of the diagnostic image prediction model (S600), the computing system generates the diagnostic image prediction model by inputting the initial dynamic image data and the delay image data to the deep neural network in various ways. For example, the computing system learns by matching each pixel in the delay image frame with the change of each pixel in the initial dynamic image data over time (for example, using a Recurrent Neural Network (RNN)). In another example, the computing system learns delay image data using each image frame in the initial dynamic image data over time as a multi-channel input (for example, a multi-channel convolutional neural network (CNN) method). In addition, as shown in FIG. 4, in another embodiment, an initial setting initial dynamic image data and an initial setting delay image data for an initial time range are obtained (S100); And calculating the optimal first time included in the initial time range by learning the initial dynamic image data for the initial setting and the delay image data for the initial setting in the initial time range (S200). In order for the computing system to acquire an image similar to the delay image data actually photographed using the diagnostic image prediction model, the first time range having the best learning effect should be determined. Therefore, the computing system may first perform a process for calculating the first time range before acquiring the learning image data combination.

The computing system obtains initial dynamic image data for initial setting using a time range longer than the first time range as an initial time range, and acquires delay image data for initial setting at a reference time point (S100). Thereafter, the computing system extracts a first time range that is optimal for predicting delay image data by learning initial dynamic image data for initial setting and delay image data for initial setting (S200). Thereafter, the computing system acquires initial dynamic image data for learning of a new patient to be included in the learning data in the extracted first time range (S400). In addition, the computing system extracts the image frame of the first time range from the initial dynamic image data for initial setting to generate the initial dynamic image data for learning, and sets the delay image data for initial setting as the delay video data for learning to add the learning data. can do.

In another embodiment, when the amount of tracer used when acquiring learning data and when diagnosing the data is different from each other, the new delay image data may include a difference in brightness or color of the initial dynamic image data for learning and the initial dynamic image data for diagnosis. It is generated by reflecting. For example, the new delay image data is generated by normalizing brightness of the training initial dynamic image data and the diagnostic initial dynamic image data based on a maximum value or an average value. In the case of a tracer (eg, Fluorodeoxyglucose (hereinafter referred to as FDG) PET) in which the amount of drug used to capture image data is continuously increased in the target area with time, PET consumed in proportion to the dose of the drug Because of the nature of the tracer, the ratio of the brightness of the reference and target areas is not dependent on the dose of the drug, so the computing system normalizes the brightness of the initial dynamic image data for training and the initial dynamic image data for diagnosis based on the maximum value or the average value. (normalization) to generate new delay image data. In the case of the tracer that is used to capture the image data is a tracer that binds to a specific target area (FP-CIT is used as a tracer), the linear tracer dynamic model that inputs the reference area and outputs the target area is used. Since the parameters of the tracker model are the same even though the input and output are linearly increased or decreased simultaneously, the computing system normalizes the brightness of the initial dynamic image data for training and the initial dynamic image data for diagnosis based on the maximum value or the average value. New delay image data can be generated.

In general, when a doctor makes a diagnosis based on delayed image data, the doctor takes a sufficient amount of radiation even after the radioisotope decreases due to a decrease in physical exhalation and extracorporeal discharge (ie, physiological reduction) such as urination and defecation. In order to perform the diagnosis with the delayed image data, a large amount of medicine is administered at an initial point. At this time, there is a problem that the amount of radiation provided to the patient's body becomes high.

Therefore, by using only one embodiment of the present invention, the diagnostic dynamic image data is obtained after inserting only a tracer of sufficient radiation dose to obtain the diagnostic dynamic image data, and the diagnostic dynamic image data is added to the prediction model. By inserting the final diagnostic delay image data by inserting, it is possible to reduce the amount of radiation provided to the patient body by reducing the amount of radioactive material injected into the patient. Specifically, when the new delay image data is calculated based on the initial dynamic image data of the patient through the diagnostic image predictive model, the computing system obtains the image without the loss of brightness due to the physiological reduction in the initial dynamic image data for the diagnosis. Since the signal-to-noise ratio (SNR) is determined (approximately inversely proportional to the square root of the brightness) depending on the brightness of the image, the medical staff can use fewer drugs than when the diagnosis is made by directly acquiring delay image data. have. At this time, the computing system corrects the new delay image data calculated based on the initial dynamic image data for diagnosis of the patient like the learning delay image data, so that the medical staff can apply the same method of diagnosing the patient through the PET image. 5 is a flowchart of a method for generating an initial diagnostic image-based diagnostic image according to another embodiment of the present invention.

Referring to FIG. 5, in the method for generating initial image for diagnosis based on dynamic image data according to another embodiment of the present invention, the computing system acquires initial dynamic image data and learning delay image data for at least one patient (S400). ); Generating, by a computing system, a diagnostic image prediction model by learning the training initial dynamic image data and the training delay image data (S600); And calculating, by the computing system, new delay image data by inputting initial diagnostic dynamic image data for a new patient to the diagnostic image prediction model (S800). Hereinafter, a detailed description of each step will be described. In addition, detailed description of the previously described steps will be omitted.

In another embodiment, when the amount of tracer used when acquiring learning data and when diagnosing the learning data is different, the step of calculating the new delay image data (S800) may include the initial dynamic image data for training and the initial dynamic image data for diagnosis. The new delay image data may be generated by reflecting brightness or color difference. FIG. 6 is a flowchart illustrating a method for generating an image for diagnosis based on initial dynamic image data according to another embodiment of the present invention.

Referring to FIG. 6, in the initial dynamic image data-based diagnostic image generation method according to another embodiment of the present invention, the computing system receives initial diagnostic dynamic image data for a new patient in a diagnostic image prediction model (S810). ); And calculating new delay image data based on the initial dynamic image data (S820).

In the present embodiment, the computing system is based on the initial dynamic image data for a new patient using a diagnostic image prediction model constructed by learning image data combinations (ie, initial dynamic image data for training and delay image data for learning). New delay image data can be generated.

The computing system receives initial diagnostic dynamic image data for the new patient in the diagnostic image prediction model (S810). The initial diagnostic dynamic image data is data obtained for a first specific time in an initial range after tracer injection, and includes data including one or more image frames. For example, if the computing system is a medical imaging apparatus itself or a computer connected to the medical imaging apparatus, the computing system may acquire initial dynamic image data for diagnosis obtained in real time according to photographing a new patient.

Thereafter, the computing system calculates new delay image data by applying the initial diagnostic dynamic image data to the diagnostic image prediction model (S820). The new delay image data is image data predicted to be acquired after a reference time, and is a diagnostic image used for patient diagnosis. The diagnostic image predictive model is generated by acquiring and learning initial dynamic image data and learning delay image data for one or more patients.

In another embodiment, when the amount of tracer used when acquiring learning data and when diagnosing the data is different, the step of calculating the new delay image data (S820) may include brightness or brightness of the initial dynamic image data for training and the initial dynamic image data for diagnosis. New delay image data is generated by reflecting color difference.

The diagnostic image generating method for initial dynamic image data based diagnosis according to an embodiment of the present invention described above may be implemented as a program (or an application) to be executed in combination with a computer which is hardware and stored in a medium.

The above-described program includes C, C ++, JAVA, machine language, etc. which can be read by the computer's processor (CPU) through the computer's device interface so that the computer reads the program and executes the methods implemented as the program. Code may be coded in the computer language of. Such code may include functional code associated with a function or the like that defines the necessary functions for executing the methods, and includes control procedures related to execution procedures necessary for the computer's processor to execute the functions according to a predetermined procedure. can do. In addition, the code may further include memory reference code for additional information or media required for the computer's processor to execute the functions at which location (address address) of the computer's internal or external memory should be referenced. have. Also, if the processor of the computer needs to communicate with any other computer or server remotely in order to execute the functions, the code may be used to communicate with any other computer or server remotely using the communication module of the computer. It may further include a communication related code for whether to communicate, what information or media should be transmitted and received during communication.

The stored medium is not a medium for storing data for a short time such as a register, a cache, a memory, but semi-permanently, and means a medium that can be read by the device. Specifically, examples of the storage medium include, but are not limited to, a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like. That is, the program may be stored in various recording media on various servers to which the computer can access or various recording media on the computer of the user. The media may also be distributed over network coupled computer systems so that the computer readable code is stored in a distributed fashion.

In the above, embodiments of the present invention have been described with reference to the accompanying drawings, but those skilled in the art to which the present invention pertains may realize the present invention in other specific forms without changing the technical spirit or essential features thereof. I can understand that. Therefore, it should be understood that the embodiments described above are exemplary in all respects and not restrictive.

Claims (12)

Acquiring, by the computing system, the initial dynamic image data and the learning delay image data for the one or more patients; And
Computing system learning the initial dynamic image data and the learning delay image data for training to generate a diagnostic image prediction model, including;
The diagnostic image predictive model calculates new delay image data based on initial dynamic image data for diagnosis of a new patient.
The initial dynamic image data is data obtained during a specific first time in an initial range after drug injection, and includes data including one or more image frames.
The delay image data is obtained after a reference time and is a diagnostic image used for diagnosing a patient. The method of claim 1,
The image data is Positron emission tomography (Positron Emission Tomography) image data, characterized in that the drug is a tracer, initial dynamic image data-based diagnostic image generation method. The method of claim 1,
The initial dynamic image data,
And an image quality of the image frame within the first time is set according to the required number of frames. The method of claim 1,
When the medicine used for image data capture is a tracer that binds to a specific target area,
The delay image data is obtained when the influence of blood flow is excluded.
The initial dynamic image data may be acquired after a time point at which the influence of blood flow in the reference area begins to decrease, or around a time when the dose rate difference between the target area and the reference area becomes more than a specific value or the rate difference is maximum. Characterized in that the initial dynamic image data based diagnostic image generation method. The method of claim 1,
Acquiring initial dynamic image data for initial setting and delay image data for initial setting during an initial time range; And
Calculating an optimal first time included in an initial time range by learning the initial dynamic image data for initial setting and the delay image data for initial setting of an initial time range; and further including the initial dynamic image data. Diagnostic image generation method. The method of claim 1,
The diagnostic image prediction model generation step,
The delay image data may be obtained by matching each pixel in the delay image frame with the change of each pixel in the initial dynamic image data over time, or by inputting each image frame in the initial dynamic image data over time as a multi-channel input. A diagnostic image generation method based on initial dynamic image data, characterized in that learning. The method of claim 1,
If the amount of drug used when acquiring learning data and when diagnosing is different,
The new delay image data,
And generating brightness or color difference between the initial dynamic image data for learning and the initial dynamic image data for diagnosis. The method of claim 7, wherein
And generating new delay image data by normalizing the brightness of the initial initial dynamic image data for training and the initial initial dynamic image data for diagnosis based on a maximum value or an average value. Acquiring, by the computing system, the initial dynamic image data and the learning delay image data for the one or more patients;
Generating, by a computing system, a diagnostic image prediction model by learning the training initial dynamic image data and the training delay image data; And
And calculating, by the computing system, new delay image data by inputting initial diagnostic dynamic image data of a new patient to the diagnostic image prediction model.
The initial dynamic image data is data obtained during a specific first time in an initial range after drug injection, and includes data including one or more image frames.
The delay image data is obtained after a reference time and is a diagnostic image used for diagnosing a patient. Receiving, by the computing system, initial diagnostic dynamic image data of the new patient in the diagnostic image prediction model; And
Calculating new delay image data based on the initial dynamic image data;
The diagnostic image prediction model,
After learning the initial dynamic image data and the learning delay image data for at least one patient is generated by learning,
The initial dynamic image data is data obtained during a specific first time in an initial range after drug injection, and includes data including one or more image frames.
The delay image data is obtained after a reference time and is a diagnostic image used for diagnosing a patient. The method of claim 9 or 10,
If the amount of drug used when acquiring learning data and when diagnosing is different,
The new delay image data calculating step,
And generating the new delay image data by reflecting a difference in brightness or color of the initial dynamic image data for learning and the initial dynamic image data for diagnosis. An initial dynamic image data-based diagnostic image generation program, coupled to a computer which is hardware and stored on a medium for executing the method of any one of claims 1 to 10.

Images