KR102715921B1 - Apparatus and method for estimation concentration of blood compound - Google Patents

Abstract

A method for predicting the concentration of a blood component according to an aspect may include a step of removing drift from NIR spectral data to obtain drift-free spectral features, a step of obtaining a global feature set based on the obtained drift-free spectral features, and a step of predicting the concentration of a blood component by regression analysis using the obtained global feature set.

Description

{Apparatus and method for estimation concentration of blood compound}

The present invention relates to a device and method for noninvasively estimating the concentration of blood components using near-infrared spectroscopy data.

Monitoring the concentration of blood components is always a topic of great interest. Monitoring the concentration of blood components is typically done by invasive methods, which involve obtaining a blood sample from the skin of a subject (human or animal). However, non-invasive methods do not require the collection of blood samples. Therefore, non-invasive methods allow people who need to check the concentration of a particular component in their blood several times a day to monitor the concentration of a particular component in their blood without any pain. Common methods used to non-invasively monitor the concentration of a blood component are mid-infrared, near-infrared, and Raman spectroscopy.

Among the above methods, near-infrared spectroscopy is widely used to monitor the concentration of blood components. However, the prediction of blood component concentration based on near-infrared spectroscopy data is very difficult when the concentration of a specific component is calculated in the presence of other components that are not of interest. For example, when monitoring the concentration of blood glucose, other components in the blood, such as water, collagen, keratin, and cholesterol, act as interference components. Another major challenge is to remove drift components from near-infrared spectroscopy data that have a negative effect on the features used for prediction. This affects the accuracy of prediction of blood components based on near-infrared spectroscopy data.

One of the prior art techniques uses an optimal filter to remove drift. However, this method requires an error covariance matrix that cannot be accurately calculated. Another prior art technique uses a baseline scatter removal algorithm that can simultaneously calculate drift associated with multiple spectra. However, this method requires continuous measurements of the same blood component concentration. This is impossible because blood component concentrations typically change over time.

Therefore, a method to remove drift from near-infrared spectroscopy data is needed to monitor blood component concentrations. In addition, a global feature set that can be used to predict blood component concentrations using regression analysis is needed.

Republic of Korea Patent Publication No. 10-2004-0020878 (2004.03.09)

The purpose is to provide a device and method for predicting the concentration of blood components.

A method for predicting the concentration of a blood component according to an aspect may include a step of removing drift from NIR spectral data to obtain drift-free spectral features, a step of obtaining a global feature set based on the obtained drift-free spectral features, and a step of predicting the concentration of a blood component by regression analysis using the obtained global feature set.

The step of obtaining the above drift-free spectral features can remove the drift using principal component analysis.

The step of obtaining the above drift-free spectral features may include the steps of calculating a plurality of principal components of the NIR spectral data, obtaining a drift approximation from the plurality of principal components, obtaining a spectral drift approximation from the drift approximation for each spectral feature according to a size of the spectral feature, and obtaining drift-free spectral features by removing each spectral drift approximation from each spectral feature.

The step of obtaining a drift approximation from the above plurality of principal components may include the step of selecting a principal component including a drift from the above plurality of principal components, and the step of obtaining a polynomial approximation of a predefined degree of the selected principal component as the drift approximation.

The step of selecting a principal component including drift from the above plurality of principal components may select the first principal component from the above plurality of principal components as the principal component including drift.

The step of obtaining a predefined polynomial approximation of the above-mentioned selected principal component by means of a drift approximation can obtain a polynomial approximation that minimizes the least square error by means of a drift approximation.

The step of obtaining the spectral drift approximation from the drift approximation may include the steps of normalizing the drift approximation and scaling the normalized drift approximation by an amplitude-span of the spectral feature to obtain the spectral drift approximation.

The step of normalizing the above drift approximation can normalize the drift approximation by dividing the drift approximation by the amplitude-span of the drift approximation.

The step of obtaining the above global feature set may include the steps of obtaining similarity values of each drift-free spectral feature for the compound vector, the step of obtaining a similarity metric for each drift-free spectral feature using the obtained similarity values, the step of ranking the drift-free spectral features according to the similarity metric, and the step of selecting a predefined number of optimal drift-free spectral features as the global feature set.

A device for predicting the concentration of a blood component according to another aspect may include a drift removal unit for removing drift from NIR spectral data to obtain drift-free spectral features, a global feature extraction unit for obtaining a global feature set based on the obtained drift-free spectral features, and a prediction unit for predicting the concentration of a blood component by regression analysis using the obtained global feature set.

The above drift removal unit can remove the drift using principal component analysis.

The above drift removal unit can calculate a plurality of principal components of the NIR spectral data, obtain a drift approximation from the plurality of principal components, obtain a spectral drift approximation from the drift approximation for each spectral feature according to the size of the spectral feature, and remove each spectral drift approximation from each spectral feature to obtain drift-free spectral features.

The above drift removal unit can select a principal component including a drift from the plurality of principal components, and obtain a polynomial approximation of a predefined degree of the selected principal component as the drift approximation.

The above drift removal unit can select the first principal component from the plurality of principal components as the principal component including the drift.

The above drift removal unit can obtain a polynomial approximation that minimizes the least square error as a drift approximation.

The above drift removal unit can obtain the spectral drift approximation by normalizing the drift approximation and scaling the normalized drift approximation by the amplitude-span of the spectral feature.

The above drift removal unit can normalize the drift approximation by dividing the drift approximation by the amplitude-span of the drift approximation.

The above global feature extraction unit can obtain similarity values of each drift-free spectral feature for a compound vector, obtain a similarity metric for each drift-free spectral feature using the obtained similarity values, rank the drift-free spectral features according to the similarity metric, and select a predefined number of optimal drift-free spectral features as a global feature set.

The accuracy of blood component concentration prediction can be improved by removing drift from near-infrared spectroscopy data and obtaining a global feature set to use for predicting blood component concentrations.

FIG. 1A is a flowchart illustrating a method for noninvasively predicting the concentration of a blood component of interest using near-infrared spectroscopy data, according to one embodiment.
FIG. 1B is a block diagram illustrating a blood component concentration prediction device according to one embodiment.
FIG. 2A is an exemplary diagram illustrating a graphical representation of a first principal component of NIR spectral data of a subject (S11) and a corresponding linear approximation thereof, according to one embodiment.
FIG. 2B is an exemplary diagram illustrating a graphical representation of a first principal component of NIR spectral data of a subject (S61) and a corresponding linear approximation thereof, according to one embodiment.
FIG. 2C is an exemplary diagram illustrating a graphical representation of a first principal component of NIR spectral data of a subject (S71) and a corresponding linear approximation thereof, according to one embodiment.
FIG. 3 is an example diagram illustrating the first principal component of 4-decimation of NIR spectral data corresponding to a subject (S11) according to one embodiment.
FIG. 4 is a flowchart illustrating a method for removing drift using principal component analysis, according to one embodiment.
FIG. 5 is a flowchart illustrating a method for extracting global features for predicting the concentration of blood components according to one embodiment.
FIG. 6 is a block diagram illustrating a blood component concentration prediction device according to another embodiment.

Hereinafter, the embodiments will be described in detail with reference to the attached drawings. When adding reference numerals to components in each drawing, it should be noted that the same components are given the same numerals as much as possible even if they are shown in different drawings. In addition, if it is determined that the description of related known functions or configurations may unnecessarily obscure the understanding of the embodiments, the detailed description thereof will be omitted.

Meanwhile, for each step, each step may occur in a different order than stated, unless the context clearly states a specific order. That is, each step may be performed in the same order as stated, may be performed substantially simultaneously, or may be performed in the opposite order.

Although the terms first, second, etc. may be used to describe various components, the components should not be limited by the terms. The terms are used only to distinguish one component from another. The singular expression includes the plural expression unless the context clearly indicates otherwise, and it should be understood that the terms "comprises" or "has" are intended to specify the presence of a feature, number, step, operation, component, part, or combination thereof described in the specification, but do not exclude the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

In addition, the division of components in this specification is only a division by the main function of each component. In other words, two or more components may be combined into one component, or one component may be divided into two or more by more detailed functions. In addition to its own main function, each component may additionally perform some or all of the functions of other components, and some of the main functions of each component may be performed exclusively by other components. Each component may be implemented by hardware or software, or by a combination of hardware and software.

Below, we describe a method for noninvasively predicting the concentration of blood components using NIR spectroscopy. In addition, we describe a drift removal algorithm for the drift removal process using information from principal components of NIR spectral data. In addition, we describe the extraction of a global feature set for predicting the concentration of blood mixtures using regression analysis.

FIG. 1a is a flowchart illustrating a method for noninvasively predicting the concentration of a blood component using near-infrared spectroscopy data according to one embodiment.

Referring to Fig. 1a, in step 102, a plurality of principal components are calculated from a near-infrared spectral data set. In step 104, a drift approximation is obtained from the plurality of principal components. In addition, in step 106, a spectral drift approximation is obtained from the drift approximation for each spectral feature according to the magnitude of the spectral feature. Then, in step 108, the drift is removed from the spectral features by subtracting each spectral drift approximation from each spectral feature. In step 110, when the drift is removed from the spectral features, a global feature set for a plurality of subjects is obtained. Finally, in step 112, the concentration of a blood component is predicted through regression analysis using the obtained global feature set.

FIG. 1b is a block diagram illustrating a blood component concentration prediction device according to one embodiment.

Referring to FIG. 1b, the blood component concentration prediction device (150) may include a drift removal unit (152), a global feature extraction unit (154), and a prediction unit (156). Here, the drift removal unit (152), the global feature extraction unit (154), and the prediction unit (156) may be implemented by one or more processors. One or more functions of each module are described in detail below.

The drift removal unit (152) can calculate and remove drift from NIR spectral data. The NIR spectral data can be obtained as follows.

First, blood component values are acquired using a standard invasive procedure. Next, a non-invasive spectrum scan is performed on the subject using a near-infrared spectrometer to acquire a raw NIR spectrum. The raw NIR spectrum is labeled with the values of blood components acquired by the invasive procedure. The acquired raw NIR spectrum is preprocessed to acquire a compound spectrum. The compound spectrum and the related component values can be arranged in the form of a matrix X using data acquired from continuous measurements, and may be referred to as a data matrix hereinafter.

Here, is a compound vector. Matrix is NIR spectral data. NIR spectral data can be affected by drift, which affects the prediction accuracy of the component of interest. Matrix Each column of the wavelength is the absorption spectrum associated with the vector can be expressed as an absorption spectrum in some embodiments. may be referred to as “spectral features” or “features”.

Absorption spectrum can be expressed as follows.

Here, is the actual absorption spectrum, may be a drift that affects the actual absorption spectrum.

The drift removal unit (152) estimates the drift component. Obtain the hop count spectrum Estimation of the drift component in By subtracting the drift-free spectrum as follows: can be obtained.

According to one embodiment, the drift removal unit (152) can remove the drift using principal component analysis (PCA). The drift removal unit (152) can remove the i-th principal component as follows: PCA operation can be performed to obtain .

Here, the variables follow the standard notation. If the drift component of the data set is significant enough to affect the prediction based on the data set, it can appear in the first principal component of the data set. Otherwise, the drift can appear in the i-th principal component. In addition, since all the principal components are uncorrelated, it can be reasonably assumed that if the drift component is captured in the i-th principal component, the drift component is unlikely to appear significantly in other principal components. The i-th principal component that the drift represents Let's write it as .

Figures 2a to 2c are graphical representations of the first principal component and the corresponding linear approximation of NIR spectral data of different subjects labeled S11, S61 and S71, according to one embodiment. As illustrated in Figures 2a to 2c, the drift component is dominant, and the slope of the linear approximation can provide the rate of change of drift with respect to time.

FIG. 3 illustrates the first principal component of 4-decimation of NIR spectral data of subject S11 according to one embodiment. In this case, the following NIR spectral data may be considered.

Spectral data The d-decimation of can be defined as follows:

set is a matrix can be obtained by including every d-th row of . As shown in Fig. 3, The first principal component of is the slope of the original matrix in Fig. 2a. can be characterized by a linear approximation with a slope of about four times that of the first principal component. This may indicate that the drift component is essentially captured in the first principal component.

Figure 4 is a flow chart illustrating a method for removing drift using principal component analysis according to one embodiment. According to one embodiment, the drift removal unit (152) can remove drift using principal component analysis. In step 402, the principal component that characterizes the drift is selected from multiple principal components. In step 404, the selected principal components Polynomial approximation of is obtained by drift approximation. In one embodiment, the drift approximation is the least squares error to minimize can be obtained. In addition, in step 406, each hop spectrum Regarding the drift approximation, Approximation of spectral drift by scaling according to the size of can be obtained by spectral drift approximation. can be obtained as follows.

Here, Is Given by is the amplitude span, Is Given by can be an amplitude span.

Finally, at step 408, all wavelengths About each Spectral drift approximation in Drift removal is performed by subtracting . This can be expressed as follows.

In some embodiments, may be referred to as drift-free spectral feature or drift-free feature.

FIG. 5 is a flowchart illustrating one or more steps involved in extracting global features for predicting concentrations of blood components according to one embodiment. In this embodiment, one or more steps performed by a global feature extraction unit (154) to extract global features are described below. In step 502, a similarity value of each drift-free spectral feature for each compound vector is obtained. There are P subjects, denoted by , and the subjects The corresponding drift-free spectral features are , and the concentration of the mixture is In one embodiment, the similarity value is a drift-free spectral feature and compound vector It can be obtained as a correlation. This can be expressed mathematically as follows.

In step 504, a similarity metric for each drift-free spectral feature can be obtained using the similarity values obtained across all subjects. According to one embodiment, the similarity metric can be calculated as follows.

In step 506, the drift-free spectral features are ranked according to a similarity metric. In step 508, K optimal drift-free spectral features are selected for prediction of blood component concentrations using regression analysis. The value of K may be determined depending on the performance of a particular regression method used for prediction. These K optimal drift-free spectral features may be referred to as "global features".

Based on the acquired global features, the prediction unit (156) can predict the concentration of the blood mixture using regression analysis from the drift-free spectral data.

FIG. 6 is a block diagram illustrating a blood component concentration prediction device according to another embodiment.

Referring to FIG. 6, it may include a processor (610), an input unit (620), a storage unit (630), a communication unit (640), and an output unit (650). Here, the processor (610) may perform the functions of the drift removal unit (152), the global feature extraction unit (154), and the prediction unit (156) of FIG. 2, so a detailed description thereof will be omitted.

The input unit (620) can receive NIR spectral data. In addition, the input unit (620) can receive various manipulation signals from the user. According to one embodiment, the input unit (620) can include a key pad, a dome switch, a touch pad (positive/negative), a jog wheel, a jog switch, a H/W button, etc. In particular, when the touch pad forms a mutual layer structure with the display, it can be called a touch screen.

The storage unit (630) can store programs or commands for the operation of the blood component concentration prediction device (600), and can store data input to the blood component concentration prediction device (600), data processed by the blood component concentration prediction device (600), and data output from the blood component concentration prediction device (600).

The storage (630) may include at least one type of storage medium, such as a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (e.g., an SD or XD memory, etc.), a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a Read Only Memory (ROM), an Electrically Erasable Programmable Read Only Memory (EEPROM), a Programmable Read Only Memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc. In addition, the blood component concentration prediction device (600) may operate an external storage medium, such as a web storage that performs the storage function of the storage (630) on the Internet.

The communication unit (640) can perform communication with an external device. For example, the communication unit (640) can transmit data input to the blood component concentration prediction device (600), stored data, processed data, etc. to an external device, or receive various data helpful for estimating blood component concentration from an external device.

At this time, the external device may be a medical device that uses data input to the blood component concentration prediction device (600), stored data, processed data, etc., a printer or display device for outputting the results. In addition, the external device may be, but is not limited to, a digital TV, a desktop computer, a mobile phone, a smart phone, a tablet, a laptop, a PDA (Personal Digital Assistants), a PMP (Portable Multimedia Player), a navigation device, an MP3 player, a digital camera, a wearable device, etc.

The communication unit (640) can communicate with an external device using Bluetooth communication, BLE (Bluetooth Low Energy) communication, Near Field Communication (NFC), WLAN communication, Zigbee communication, Infrared Data Association (IrDA) communication, WFD (Wi-Fi Direct) communication, UWB (ultra-wideband) communication, Ant+ communication, WIFI communication, RFID (Radio Frequency Identification) communication, 3G communication, 4G communication, and 5G communication. However, this is only an example and is not limited thereto.

The output unit (650) can output data input to the blood component concentration prediction device (600), stored data, processed data, etc. According to one embodiment, the output unit (650) can output data input to the blood component concentration prediction device (600), stored data, processed data, etc. using at least one of an auditory method, a visual method, and a tactile method. To this end, the output unit (650) can include a display, a speaker, a vibrator, etc.

The above-described embodiments can be implemented as computer-readable codes on a computer-readable recording medium. The computer-readable recording medium can include all kinds of recording devices that store data that can be read by a computer system. Examples of the computer-readable recording medium can include ROMs, RAMs, CD-ROMs, magnetic tapes, floppy disks, optical disks, etc. In addition, the computer-readable recording medium can be distributed over network-connected computer systems, and can be written and executed as computer-readable codes in a distributed manner.

We have examined preferred embodiments so far. Those skilled in the art will understand that the invention can be implemented in modified forms without departing from the essential characteristics of the invention. Accordingly, the scope of the invention should be interpreted so as to include various embodiments within the scope equivalent to the contents described in the patent claims, and not be limited to the embodiments described above.

150: Blood component concentration prediction device
152: Drift removal section
154: Global feature extraction section
156: Prediction Department

Claims (18)

A device for predicting the concentration of blood components,
A step of removing drift from NIR spectral data to obtain drift-free spectral features;
A step of obtaining a global feature set based on the above-obtained drift-free spectral features; and
A step of predicting the concentration of a blood component by regression analysis using the above-mentioned acquired global feature set; including;
The step of obtaining the above drift-free spectral features is:
A step of producing multiple principal components of the above NIR spectral data;
A step of obtaining a drift approximation from the above plurality of principal components;
A step of normalizing the drift approximation for each spectral feature according to the size of the spectral feature, and scaling the normalized drift approximation by the amplitude-span of the spectral feature to obtain a spectral drift approximation; and
A step of obtaining drift-free spectral features by removing each spectral drift approximation from each spectral feature; comprising;
Method for predicting the concentration of blood components. In the first paragraph,
The step of obtaining the above drift-free spectral features is:
Removing the above drift using principal component analysis,
Method for predicting the concentration of blood components. delete In the first paragraph,
The step of obtaining a drift approximation from the above plurality of principal components is:
a step of selecting a principal component including a drift from the above plurality of principal components; and
A step of obtaining a polynomial approximation of a predefined degree of the selected principal component by the drift approximation; comprising;
Method for predicting the concentration of blood components. In paragraph 4,
The step of selecting a principal component including drift from the above plurality of principal components comprises selecting the first principal component from the above plurality of principal components as the principal component including drift.
Method for predicting the concentration of blood components. In paragraph 4,
The step of obtaining a predefined polynomial approximation of the above-mentioned selected principal component as a drift approximation is obtaining a polynomial approximation that minimizes the least square error as a drift approximation.
Method for predicting the concentration of blood components. delete In the first paragraph,
The step of normalizing the drift approximation comprises normalizing the drift approximation by dividing the drift approximation by the amplitude-span of the drift approximation.
Method for predicting the concentration of blood components. In the first paragraph,
The step of obtaining the above global feature set is:
A step of obtaining similarity values of each drift-free spectral feature for the compound vector;
A step of obtaining a similarity metric for each drift-free spectral feature using the obtained similarity values;
a step of ranking the drift-free spectral features according to the above similarity metric; and
comprising a step of selecting a predefined number of optimal drift-free spectral features as a global feature set;
Method for predicting the concentration of blood components. A drift removal unit for removing drift from NIR spectral data to obtain drift-free spectral features;
A global feature extraction unit for obtaining a global feature set based on the above-mentioned acquired drift-free spectral features; and
A prediction unit for predicting the concentration of blood components by regression analysis using the above-mentioned acquired global feature set;
The above drift removal unit,
Generate multiple principal components of the above NIR spectral data,
Obtain a drift approximation from the above multiple principal components,
For each spectral feature, the drift approximation is normalized according to the magnitude of the spectral feature, and the spectral drift approximation is obtained by scaling the normalized drift approximation by the amplitude-span of the spectral feature.
Obtain drift-free spectral features by removing each spectral drift approximation from each spectral feature.
A device for predicting the concentration of blood components. In Article 10,
The above drift removal unit removes the drift by using principal component analysis.
A device for predicting the concentration of blood components. delete In Article 10,
The above drift removal unit,
Selecting a principal component including a drift from the above plurality of principal components, and obtaining a polynomial approximation of a predefined degree of the selected principal component as the drift approximation.
A device for predicting the concentration of blood components. In Article 13,
The above drift removal unit,
Selecting the first principal component from the above plurality of principal components as the principal component containing the drift.
A device for predicting the concentration of blood components. In Article 13,
The above drift removal unit,
Obtain a polynomial approximation that minimizes the least squares error by drift approximation.
A device for predicting the concentration of blood components. delete In Article 10,
The above drift removal unit,
Normalizing the drift approximation by dividing the drift approximation by the amplitude-span of the drift approximation.
A device for predicting the concentration of blood components. In Article 10,
The above global feature extraction unit,
Obtain the similarity values of each drift-free spectral feature for the compound vector,
Using the obtained similarity values, a similarity metric for each drift-free spectral feature is obtained,
Rank the drift-free spectral features according to the above similarity metric,
Selecting a predefined number of optimal drift-free spectral features as a global feature set,
A device for predicting the concentration of blood components.

Images