CN115937103A - Alkaline phosphatase concentration detection method based on deep learning and application software - Google Patents
Alkaline phosphatase concentration detection method based on deep learning and application software Download PDFInfo
- Publication number
- CN115937103A CN115937103A CN202211450123.6A CN202211450123A CN115937103A CN 115937103 A CN115937103 A CN 115937103A CN 202211450123 A CN202211450123 A CN 202211450123A CN 115937103 A CN115937103 A CN 115937103A
- Authority
- CN
- China
- Prior art keywords
- deep learning
- alkaline phosphatase
- picture
- follows
- concentration detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 102000002260 Alkaline Phosphatase Human genes 0.000 title claims abstract description 71
- 108020004774 Alkaline Phosphatase Proteins 0.000 title claims abstract description 71
- 238000013135 deep learning Methods 0.000 title claims abstract description 55
- 238000001514 detection method Methods 0.000 title claims abstract description 52
- 238000000034 method Methods 0.000 claims abstract description 45
- 238000003062 neural network model Methods 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000013461 design Methods 0.000 claims abstract description 8
- 238000012545 processing Methods 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 45
- 230000008569 process Effects 0.000 claims description 16
- 230000004913 activation Effects 0.000 claims description 8
- 238000009499 grossing Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000004737 colorimetric analysis Methods 0.000 abstract description 6
- 239000000463 material Substances 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 5
- 238000006243 chemical reaction Methods 0.000 description 4
- 238000004416 surface enhanced Raman spectroscopy Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000001069 Raman spectroscopy Methods 0.000 description 2
- 238000002835 absorbance Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000005518 electrochemistry Effects 0.000 description 2
- 238000002795 fluorescence method Methods 0.000 description 2
- 239000007850 fluorescent dye Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 208000020084 Bone disease Diseases 0.000 description 1
- 239000003298 DNA probe Substances 0.000 description 1
- 108020004518 RNA Probes Proteins 0.000 description 1
- 239000003391 RNA probe Substances 0.000 description 1
- FOIXSVOLVBLSDH-UHFFFAOYSA-N Silver ion Chemical compound [Ag+] FOIXSVOLVBLSDH-UHFFFAOYSA-N 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 239000000090 biomarker Substances 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 1
- 229910052737 gold Inorganic materials 0.000 description 1
- 239000010931 gold Substances 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000003018 immunoassay Methods 0.000 description 1
- 208000019423 liver disease Diseases 0.000 description 1
- 238000002796 luminescence method Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 210000002966 serum Anatomy 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
- 231100000331 toxic Toxicity 0.000 description 1
- 230000002588 toxic effect Effects 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Abstract
The invention provides an alkaline phosphatase concentration detection method based on deep learning, which is used for processing data of pictures based on the deep learning; establishing a neural network model in deep learning and training; and inputting the alkaline phosphatase concentration detection image into the trained neural network model in deep learning to obtain the concentration value of the alkaline phosphatase. An application software is designed based on the alkaline phosphatase concentration detection method of deep learning, and a neural network model in deep learning is deployed at a mobile phone end by adopting a Pythrch deep learning framework; and performing functional design. The method realizes the concentration detection of the alkaline phosphatase on the electronic product, solves the inconvenience of the concentration detection of the alkaline phosphatase at present, can obtain the concentration value of the alkaline phosphatase only by taking a picture and uploading the picture by a mobile phone or other electronic equipment, greatly saves manpower and material resources compared with the prior colorimetric method and the like, and relieves the problem of medical resource shortage in some regions.
Description
Technical Field
The invention relates to the field of detection of alkaline phosphatase, and particularly provides a deep learning-based alkaline phosphatase concentration detection method and application software.
Background
Alkaline phosphatase (ALP) is a very important biomarker, high-level alkaline phosphatase expression in human serum is usually related to certain bone diseases and liver diseases, the current commonly used method is colorimetry, the ALP participates in a reaction system to cause the color absorbance of a solution to change, and the ALP concentration is obtained by analyzing absorbance, but the colorimetry has low sensitivity and poor anti-interference performance, and the colorimetry needs complex instruments and professional operators, and is particularly not convenient for certain areas with poor medical resources.
There are other ALP detection methods including a fluorescence method, an electrochemiluminescence method, a Surface Enhanced Raman Scattering (SERS) method, etc., in which:
fluorescence method: the presence of ALP causes a change (enhancement or reduction) in the fluorescence signal by synthesizing a fluorescent probe specifically responsive to ALP, and the ALP concentration is analyzed by the fluorescence signal, but the design of the fluorescent probe is complicated, the synthesis process is time-consuming and may be toxic.
Electrochemical luminescence method: the specific chemiluminescence reaction initiated by electrochemistry on the surface of the electrode comprises two processes of electrochemistry and chemiluminescence, the ALP detection is realized by a luminous signal, and the method not only can be applied to all immunoassays, but also can be used for detecting DNA or RNA probes.
Surface Enhanced Raman Scattering (SERS) method: in the presence of ALP, raman signals change, and are amplified through the surface Raman enhancement effect of the gold and silver nanoparticles, so that the detection effect is realized.
However, the above methods have the disadvantages of high cost, high requirements for detection process and operation conditions, and the like. Therefore, an ALP real-time detection technique with low cost, simplicity, rapidness, high sensitivity is urgently needed.
Disclosure of Invention
The invention aims to solve the problems, provides an alkaline phosphatase concentration detection method based on deep learning, designs a mobile phone application software based on the method, and aims to solve the inconvenience of detecting the alkaline phosphatase at present.
The invention provides an alkaline phosphatase concentration detection method based on deep learning, which comprises the following steps:
s1, carrying out data processing on the picture based on deep learning, wherein the specific process is as follows:
s11, scaling the size of the picture by adopting a bilinear interpolation method;
s12, performing center rotation on the picture;
s13, adjusting the brightness of the picture;
s14, smoothing the picture by adopting Gaussian blur;
s2, establishing a neural network model in deep learning and training, wherein the specific process is as follows:
s21, establishing ResNet-34, and adding a sigmoid activation function behind a full connection layer;
s22, training a neural network model in deep learning is completed by adopting a mean square loss function to minimize the square sum mean of the difference between the target value and the predicted value;
and S3, inputting the alkaline phosphatase concentration detection image into the trained neural network model in deep learning to obtain the concentration value of the alkaline phosphatase.
Preferably, in S11, the bilinear interpolation method specifically includes the following steps:
for the pixel of any unknown function f at any point P (x, y), the coordinates and pixels of four points surrounding P, namely Q, are determined 11 (x 1 ，y 1 )、Q 12 (x 1 ，y 2 )、Q 21 (x 2 ，y 1 )、Q 22 (x 2 ，y 2 )；
Interpolating in the X direction to obtain f (R) 1 ) And f (R) 2 ) The following were used:
R 1 ＝(x，y 1 )
R 2 ＝(x,y 2 )
interpolation is carried out in the Y direction, and the calculation formula of the pixel of the arbitrary unknown function f at the arbitrary point P (x, Y) is obtained as follows:
preferably, in S13, the pixels of the three RGB channels of the picture are multiplied by the adjustment factor percent, and if the adjustment factor percent >1, the brightness of the picture is increased; if the adjustment factor Percentage is less than or equal to 1, the brightness of the picture is reduced.
Preferably, in S14, the specific process of performing the smoothing process on the picture by using the gaussian blur is as follows:
adopting a Gaussian normal distribution density function to carry out Gaussian blur on the picture to obtain a normal distribution curve;
the normal distribution density function of gaussian is as follows:
where μ represents the mean of the normal distribution and σ represents the variance of the normal distribution.
Setting the midpoint of the normal distribution curve as an origin to obtain a one-dimensional Gaussian function as follows:
the two-dimensional gaussian function is obtained from the one-dimensional gaussian function as follows:
and distributing weights to other points except the origin according to a two-dimensional Gaussian function, and calculating to obtain a weighted average value of all points on the normal distribution curve.
Preferably, in S21, the sigmoid activation function is as follows:
the output value after adding the sigmoid activation function is a value between 0 and 1.
Preferably, the mean square loss function is as follows:
wherein N represents the total number of training samples, i is a label variable, i belongs to [1,N ], and theta is a model parameter.
Application software of an alkaline phosphatase concentration detection method based on deep learning is characterized in that a neural network model in deep learning is deployed at a mobile phone end by adopting a Pytrch deep learning framework; and (3) performing functional design to enable the alkaline phosphatase concentration detection image to be input by selecting or photographing in an album, cutting to obtain a region to be detected in the input process, and obtaining the alkaline phosphatase concentration value of the region to be detected by using a neural network model in deep learning.
Compared with the prior art, the invention can obtain the following beneficial effects:
the invention realizes the concentration detection of the alkaline phosphatase on the electronic product, solves the inconvenience of the concentration detection of the alkaline phosphatase at present, obtains the concentration value of the alkaline phosphatase only by taking a picture and uploading the picture by a mobile phone or other electronic equipment and calculating the picture by using a deep learning algorithm, greatly saves manpower and material resources compared with the existing colorimetric method and the like, and relieves the problem of lack of medical resources in some areas.
Drawings
FIG. 1 is a flow chart of a method for detecting alkaline phosphatase concentration based on deep learning according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of bilinear interpolation provided in accordance with an embodiment of the present invention;
FIG. 3 is a flowchart of the software application of the method for detecting alkaline phosphatase concentration based on deep learning according to the embodiment of the present invention;
fig. 4 is a schematic starting diagram of a mobile phone APP provided according to an embodiment of the present invention;
fig. 5 is a schematic main page of a mobile phone APP provided according to an embodiment of the present invention;
FIG. 6 is a schematic cut-out diagram of an alkaline phosphatase concentration detection image of a mobile phone APP provided according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of a detection page of an alkaline phosphatase concentration detection image of a mobile phone APP provided in an embodiment of the present invention;
fig. 8 is a schematic diagram of an alkaline phosphatase concentration detection image of a mobile phone APP provided according to an embodiment of the present invention;
fig. 9 is a schematic page of a detection result of an alkaline phosphatase concentration detection image of a mobile phone APP according to an embodiment of the present invention.
Detailed Description
Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following description, like modules are denoted by like reference numerals. In the case of the same reference numerals, their names and functions are also the same. Therefore, detailed description thereof will not be repeated.
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention.
Fig. 1 shows a flowchart of an alkaline phosphatase concentration detection method based on deep learning according to an embodiment of the present invention.
As shown in fig. 1, an alkaline phosphatase concentration detection method based on deep learning and application software provided by an embodiment of the present invention relate to deep learning and Android Studio-related development technologies, and data processed by deep learning is input into a deep learning neural network model for training, so that a deep learning neural network model with a relatively small error is obtained after training; in the development of an android program, a trained neural network model in deep learning is migrated to an android mobile phone APP by using an official deployment method, so that a picture is selected to be uploaded and a concentration value of alkaline phosphatase is detected through the mobile phone APP, wherein the detection method specifically comprises the following steps:
s1, carrying out data processing on the picture based on deep learning, wherein the data processing is used for expanding a data set on one hand and adding noise to training data on the other hand, and avoids overfitting during training as much as possible, and the specific process is as follows:
s11, in the prior art, the nearest linear interpolation method has the problem that the processed and obtained picture is not smooth enough, so the method adopts a bilinear interpolation method to scale the size of the picture, mainly carries out linear interpolation once in two directions of X, Y respectively, obtains the pixel of any unknown function f at any point P (x, y) during scaling, and specifically describes as follows:
fig. 2 illustrates a bilinear interpolation method provided according to an embodiment of the present invention.
For the pixel of any unknown function f at any point P (x, y), the coordinates and pixels of four points surrounding P, namely Q, are determined 11 (x 1 ，y 1 )、Q 12 (x 1 ，y 2 )、Q 21 (x 2 ，y 1 )、Q 22 (x 2 ，y 2 ) Usually, the four points are four point vertices of the picture, and when there is a special requirement, the points surrounding P may also adopt a larger number of points;
firstly, linear interpolation is carried out in the X direction to obtain f (R) 1 ) And f (R) 2 )、R 1 And R 2 The following were used:
R 1 ＝(x，y 1 )
R 2 ＝(x,y 2 )
interpolation is carried out in the Y direction, and the calculation formula of the pixel of the arbitrary unknown function f at the arbitrary point P (x, Y) is obtained as follows:
s12, in the rectangular plane coordinate system where the picture is located, the picture is subjected to center rotation, and a rotation matrix around the rotation center can be expressed as:
where center.x and center.y are coordinates of the rotation center, α = cos θ, β = sin θ, and θ represents the rotation angle.
S13, the picture has three RGB channels, the pixel values of the three channels are multiplied by an adjusting factor Percentage, and the adjusting direction of the brightness of the picture is judged according to the adjusting factor:
if the adjustment factor Percentage is larger than 1, increasing the brightness of the picture;
if the adjustment factor Percentage is less than or equal to 1, the brightness of the picture is reduced.
S14, smoothing the picture by adopting Gaussian blur, wherein the specific process is as follows:
in this embodiment, a gaussian normal distribution density function is used to perform gaussian blurring on the picture to obtain a normal distribution curve;
the Gaussian normal distribution density function is as follows:
where μ represents the mean of a normal distribution and σ represents the variance of the normal distribution.
The normal distribution curve is a bell-shaped curve, and is characterized in that the closer to the central point, the larger the value is, the farther away from the central point, the smaller the value is, the midpoint of the normal distribution curve is set as the origin, and the obtained one-dimensional Gaussian function is as follows:
the two-dimensional gaussian function is obtained from the one-dimensional gaussian function as follows:
and distributing weights to other points except the original point according to a two-dimensional Gaussian function, and calculating to obtain a weighted average value of all points on the normal distribution curve according to the distributed weights of all points and point coordinate data.
S2, establishing a neural network model in deep learning and training, wherein the neural network model in deep learning in the embodiment specifically adjusts a residual error network (ResNet-34) on the basis of a ResNet model which is popular in current picture processing, obtains required model parameters according to processed data training, and determines the final neural network model in deep learning, and the specific network and hierarchy of ResNet-34 are divided into the prior art, which is not described herein again, and the following only describes the innovative adjustment content:
s21, establishing ResNet-34 as a backbone network of a neural network model in deep learning, wherein an original task of ResNet-34 is picture classification, in the embodiment of the invention, a required target result is a fitting curve, each input value correspondingly outputs an output value of 0-20, the input value is processing data of an alkaline phosphatase concentration detection image, and the output value is a concentration value of alkaline phosphatase, so that a sigmoid activation function is added behind a full connection layer of ResNet-34, the original output result of ResNet-34 is activated to a value of 0-1, and the value of 0-1 is multiplied by a corresponding multiplying factor to enable a final output result to meet the concentration display requirement.
The sigmoid activation function adopted by the embodiment of the invention is as follows:
s22, because the original target of ResNet-34 is to classify the image, training is carried out through the loss function on intersection in the training stage, the maximum probability of a certain image class is finally output, the task of image classification is completed, and the loss function on intersection for classifying the task is as follows:
the invention aims to fit a concentration curve, so that a mean square Loss function (MSE-Loss) is adopted to minimize the mean of the sum of squares of the difference between a target value and a predicted value during training, and network training is completed.
The mean square loss function is as follows:
wherein N represents the total number of training samples, i is a label variable, i belongs to [1,N ], and theta is a model parameter.
And S3, inputting the alkaline phosphatase concentration detection image into the trained neural network model in deep learning to obtain the concentration value of the alkaline phosphatase.
The method is based on deep learning, trains an obtained neural network model in the deep learning and provides technical support for the concentration detection of the alkaline phosphatase on an electronic product, so that the method provides a framework of application software based on the method, and particularly deploys an alkaline phosphatase concentration detection APP at an android mobile phone terminal, thereby realizing the technical result conversion, wherein the specific framework is as follows:
fig. 3 shows a flow chart of application software of the alkaline phosphatase concentration detection method based on deep learning according to the embodiment of the present invention.
As shown in fig. 3, the trained model is subjected to parameter file conversion, and finally a parameter file with a suffix of ". Ptl" is obtained, and the neural network model in deep learning is deployed at a mobile phone end by using a Pytorch _ android library and a Pytorch _ andoird _ torchvision library provided by a Pytorch official, so that a functional design is required for a user to use conveniently, and an alkaline phosphatase concentration detection image can be selected from an album or input by photographing; in addition, in the input process, the region to be detected can be obtained by cutting, the function is realized in an Android Studio, a mode of obtaining a photo is set in a mode of creating Intent, in order to select the concentration value of the region to be detected which is expected to be obtained, a user can cut the picture after selecting one picture, a transmission mode of data of the specified region in an APP page is obtained, whether the picture needs to be cut or not is obtained, the cut picture is sent to a neural network model in deep learning, the concentration value of alkaline phosphatase of the region to be detected can be obtained, the interface design and the function design of software are the prior art, and no repeated description is given here.
To verify the effectiveness and convenience of the present invention, the following is described in conjunction with APP:
selecting a case picture, wherein the concentration value of alkaline phosphatase measured by the traditional detection method is 10.0mU/ml, starting an alkaline phosphatase concentration detection APP, and referring to a starting interface of the alkaline phosphatase concentration detection APP in FIG. 4, a main page of the alkaline phosphatase concentration detection APP is shown in FIG. 5, selecting the picture in an album as shown in FIG. 6, manually cutting the picture according to requirements, wherein the cut interface is shown in FIG. 6, entering a detection page picture 7 after determining a region to be detected, and clicking detection by a user, namely starting detection as shown in FIG. 8; fig. 9 shows the concentration values of alkaline phosphatase in case pictures. The concentration value is 9.77mU/ml, and the detection result of the alkaline phosphatase concentration detection APP is quite close to the existing complex detection result, so that the simple requirement is completely met, and the APP can be used as a medical reference.
Data of the APP for detecting the concentration of the alkaline phosphatase are normal within 10.00 +/-1.00, and when the quality of an obtained picture is influenced by factors such as illumination, environment and the like, individual experiments can be beyond a normal range, and the APP can be detected by photographing again. The detection of the concentration of the alkaline phosphatase APP does not need expensive cost, complex operation process and operation conditions, and compared with colorimetry, fluorescence, electrochemiluminescence and the like, the method realizes low-cost, simple, convenient, rapid and sensitive instant detection targets, can realize instant quantitative analysis through a mobile phone terminal, and has greater advantages compared with traditional reservation detection, and the applicability is wider.
In addition, the APP is not limited to the android mobile phone terminal, and can be deployed on the apple or other electronic equipment with the photographing or storage function.
While embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and should not be taken as limiting the invention. Variations, modifications, substitutions and alterations of the above-described embodiments may be made by those of ordinary skill in the art without departing from the scope of the present invention.
The above embodiments of the present invention should not be construed as limiting the scope of the present invention. Any other corresponding changes and modifications made according to the technical idea of the present invention should be included in the protection scope of the claims of the present invention.
Claims (7)
1. A method for detecting the concentration of alkaline phosphatase based on deep learning is characterized by comprising the following steps:
s1, carrying out data processing on the picture based on deep learning, wherein the specific process is as follows:
s11, scaling the size of the picture by adopting a bilinear interpolation method;
s12, performing center rotation on the picture;
s13, adjusting the brightness of the picture;
s14, smoothing the picture by adopting Gaussian blur;
s2, establishing a neural network model in deep learning and training, wherein the specific process is as follows:
s21, establishing ResNet-34, and adding a sigmoid activation function behind a full connection layer;
s22, training the neural network model in the deep learning is completed by adopting a mean square loss function to minimize the square sum mean of the difference between the target value and the predicted value;
and S3, inputting the alkaline phosphatase concentration detection image into the trained neural network model in the deep learning to obtain the concentration value of the alkaline phosphatase.
2. The method for detecting alkaline phosphatase concentration based on deep learning according to claim 1, wherein in S11, the bilinear interpolation method comprises the following specific steps:
for the pixel of any unknown function f at any point P (x, y), the coordinates and pixels of four points surrounding P, i.e. Q, are determined 11 (x 1 ，y 1 )、Q 12 (x 1 ，y 2 )、Q 21 (x 2 ，y 1 )、Q 22 (x 2 ，y 2 )；
Interpolating in the X direction to obtain f (R) 1 ) And f (R) 2 ) The following were used:
R 1 ＝(x，y 1 )
R 2 ＝(x,y 2 )
interpolation is carried out in the Y direction, and the calculation formula of the pixel of the arbitrary unknown function f at the arbitrary point P (x, Y) is obtained as follows:
3. the method for detecting the concentration of alkaline phosphatase based on deep learning according to claim 1, wherein in S13, the pixels of the three RGB channels of the picture are multiplied by an adjustment factor percent, and if the adjustment factor percent >1, the brightness of the picture is increased; and if the adjustment factor Percentage is less than or equal to 1, reducing the brightness of the picture.
4. The method for detecting the concentration of alkaline phosphatase based on deep learning according to claim 1, wherein in S14, the smoothing process of the picture by using gaussian blur is as follows:
adopting a Gaussian normal distribution density function to carry out Gaussian blur on the picture to obtain a normal distribution curve;
the normal distribution density function of gaussian is as follows:
wherein μ represents a mean of a normal distribution, and σ represents a variance of the normal distribution;
setting the midpoint of the normal distribution curve as an origin, and obtaining a one-dimensional Gaussian function as follows:
obtaining a two-dimensional Gaussian function according to the one-dimensional Gaussian function as follows:
and distributing weights to other points except the origin according to the two-dimensional Gaussian function, and calculating to obtain a weighted average value of all points on the normal distribution curve.
5. The deep learning-based alkaline phosphatase concentration detection method according to claim 1, wherein in S21, the sigmoid activation function is as follows:
the output value after the sigmoid activation function is added is a value between 0 and 1.
6. The method for detecting alkaline phosphatase concentration based on deep learning according to claim 1, wherein the mean square loss function is as follows:
wherein N represents the total number of training samples, i is a label variable, i belongs to [1,N ], and theta is a model parameter.
7. Application software of the alkaline phosphatase concentration detection method based on deep learning according to any one of claims 1 to 6, wherein a Pythrch deep learning framework is adopted to deploy the neural network model in deep learning at a mobile phone end; and performing functional design to enable the alkaline phosphatase concentration detection image to be input by selecting or photographing in an album, and cutting to obtain a region to be detected in the input process, wherein the alkaline phosphatase concentration value of the region to be detected can be obtained by the neural network model in deep learning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211450123.6A CN115937103A (en) | 2022-11-19 | 2022-11-19 | Alkaline phosphatase concentration detection method based on deep learning and application software |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211450123.6A CN115937103A (en) | 2022-11-19 | 2022-11-19 | Alkaline phosphatase concentration detection method based on deep learning and application software |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115937103A true CN115937103A (en) | 2023-04-07 |
Family
ID=86699981
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211450123.6A Pending CN115937103A (en) | 2022-11-19 | 2022-11-19 | Alkaline phosphatase concentration detection method based on deep learning and application software |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115937103A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110738641A (en) * | 2019-10-07 | 2020-01-31 | 福州大学 | Image processing and KELM based qualitative detection method for concentration of medical reagent |
| KR102193108B1 (en) * | 2019-10-10 | 2020-12-18 | 서울대학교산학협력단 | Observation method for two-dimensional river mixing using RGB image acquired by the unmanned aerial vehicle |
| CN112330682A (en) * | 2020-11-09 | 2021-02-05 | 重庆邮电大学 | Industrial CT image segmentation method based on deep convolutional neural network |
| CN114049370A (en) * | 2021-11-04 | 2022-02-15 | 马晓晨 | Intelligent mobile terminal colored solution concentration detection method based on deep learning method |
| CN114998202A (en) * | 2022-04-26 | 2022-09-02 | 杭州电子科技大学 | Semi-supervised deep learning defect detection method |
-
2022
- 2022-11-19 CN CN202211450123.6A patent/CN115937103A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110738641A (en) * | 2019-10-07 | 2020-01-31 | 福州大学 | Image processing and KELM based qualitative detection method for concentration of medical reagent |
| KR102193108B1 (en) * | 2019-10-10 | 2020-12-18 | 서울대학교산학협력단 | Observation method for two-dimensional river mixing using RGB image acquired by the unmanned aerial vehicle |
| CN112330682A (en) * | 2020-11-09 | 2021-02-05 | 重庆邮电大学 | Industrial CT image segmentation method based on deep convolutional neural network |
| CN114049370A (en) * | 2021-11-04 | 2022-02-15 | 马晓晨 | Intelligent mobile terminal colored solution concentration detection method based on deep learning method |
| CN114998202A (en) * | 2022-04-26 | 2022-09-02 | 杭州电子科技大学 | Semi-supervised deep learning defect detection method |
Non-Patent Citations (11)
| Title |
|---|
| *MAJ...: "均方误差损失函数（MSE，mean squared error)", 《HTTPS://BLOG.CSDN.NET/QQ_41375318/ARTICLE/DETAILS/106047208》, pages 1 - 6 * |
| 4869CONAN: "数据增强处理(图像平移, 缩放, 翻转, 旋转, 加噪, 亮度调节)", 《HTTPS://BLOG.CSDN.NET/QQ_42758024/ARTICLE/DETAILS/102822329》, pages 1 - 6 * |
| 4869CONAN: "数据增强处理(图像平移，缩放，翻转，旋转，加噪，亮度调节)", 《HTTPS://BLOG.CSDN.NET/QQ_42758024/ARTICLE/DETAILS/102822329》, pages 1 - 6 * |
| HARDMAN: "图像处理之高斯模糊", < HTTPS://WWW.JIANSHU.COM/P/F27BB99368C9>, pages 1 - 8 * |
| PAUL-LANGJUN: "激活函数与损失函数的配对问题", 《HTTPS://BLOG.CSDN.NET/GAOXUEYI551/ARTICLE/DETAILS/104462041》, pages 1 - 4 * |
| TYHJ_SF: "常用激活函数（激励函数）理解与总结", 《HTTPS://BLOG.CSDN.NET/TYHJ_SF/ARTICLE/DETAILS/79932893》, pages 1 - 9 * |
| 刘志浩等: "仿生视觉色度识别的浓度快速测定", 《应用化学》, pages 196 * |
| 成功上岸U: "《数字图像处理》实验之对图像进行双线性 （bilinear）插值缩放", 《HTTPS://BLOG.CSDN.NET/QQ_56982298/ARTICLE/DETAILS/127687122》, pages 1 - 9 * |
| 爽朗的SUNMENG: "图像亮度调整", 《HTTPS://WWW.CNBLOGS.COM/BIGDREAM6/P/8378343.HTML》, pages 1 - 3 * |
| 程序猿浩波: "ResNet——CNN经典网络模型详解(pytorch实现)", 《HTTPS://BAIJIAHAO.BAIDU.COM/S?ID=1674926316828109679&WFR=SPIDER&FOR=PC》, pages 1 - 11 * |
| 谜之_摄影爱好者: "神经网络中常用的损失函数", 《HTTPS://BLOG.CSDN.NET/BAIDU_31982893/ARTICLE/DETAILS/124433844》, pages 1 - 3 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8406554B1 (en) | Image binarization based on grey membership parameters of pixels | |
| Komyshev et al. | Evaluation of the SeedCounter, a mobile application for grain phenotyping | |
| CN105283902A (en) | Image processing device, image processing method, and image processing program | |
| KR20120023591A (en) | Image processing apparatus, method and program | |
| CN110222694B (en) | Image processing method, image processing device, electronic equipment and computer readable medium | |
| JP2007507803A (en) | Method and apparatus for determining parameters for adjusting an image | |
| EP1745438A1 (en) | Method for determining image quality | |
| JP2004310475A (en) | Image processor, cellular phone for performing image processing, and image processing program | |
| CN113781510A (en) | Edge detection method and device and electronic equipment | |
| AU2021273324A1 (en) | Assay reading method | |
| Wharton et al. | Logarithmic edge detection with applications | |
| Sivakumar et al. | An automated lateral flow assay identification framework: Exploring the challenges of a wearable lateral flow assay in mobile application | |
| CN115937103A (en) | Alkaline phosphatase concentration detection method based on deep learning and application software | |
| CN116157867A (en) | Neural network analysis of LFA test strips | |
| Nuutinen et al. | A framework for measuring sharpness in natural images captured by digital cameras based on reference image and local areas | |
| CN111967332B (en) | Visibility information generation method and device for automatic driving | |
| Hipp et al. | Computer-aided laser dissection: A microdissection workflow leveraging image analysis tools | |
| CN116735463A (en) | Directed target detection-based diatom size automatic measurement method | |
| JP2006031584A (en) | Area division of image | |
| CN110046607A (en) | A kind of unmanned aerial vehicle remote sensing image board house or building materials test method based on deep learning | |
| CN115546554A (en) | Sensitive image identification method, device, equipment and computer readable storage medium | |
| Lian et al. | Film and television animation sensing and visual image by computer digital image technology | |
| CN114943976A (en) | Model generation method and device, electronic equipment and storage medium | |
| CN115330610A (en) | Image processing method, image processing apparatus, electronic device, and storage medium | |
| JPWO2005075961A1 (en) | Glitter feeling evaluation method and glitter feeling evaluation apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |