CN116645565A - Classifier construction method based on Raman spectrum and support vector machine - Google Patents
Classifier construction method based on Raman spectrum and support vector machine Download PDFInfo
- Publication number
- CN116645565A CN116645565A CN202310725163.5A CN202310725163A CN116645565A CN 116645565 A CN116645565 A CN 116645565A CN 202310725163 A CN202310725163 A CN 202310725163A CN 116645565 A CN116645565 A CN 116645565A
- Authority
- CN
- China
- Prior art keywords
- spectrum
- support vector
- vector machine
- image data
- self
- 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
- 238000012706 support-vector machine Methods 0.000 title claims abstract description 46
- 238000001237 Raman spectrum Methods 0.000 title claims abstract description 32
- 238000010276 construction Methods 0.000 title claims abstract description 14
- 238000001228 spectrum Methods 0.000 claims abstract description 61
- 238000002189 fluorescence spectrum Methods 0.000 claims abstract description 33
- 238000012549 training Methods 0.000 claims abstract description 25
- 206010008342 Cervix carcinoma Diseases 0.000 claims abstract description 12
- 208000006105 Uterine Cervical Neoplasms Diseases 0.000 claims abstract description 12
- 201000010881 cervical cancer Diseases 0.000 claims abstract description 12
- 238000007781 pre-processing Methods 0.000 claims abstract description 9
- 238000012216 screening Methods 0.000 claims abstract description 7
- 238000012935 Averaging Methods 0.000 claims abstract description 3
- 238000000034 method Methods 0.000 claims description 18
- 230000003595 spectral effect Effects 0.000 claims description 15
- 239000013598 vector Substances 0.000 claims description 10
- 238000012937 correction Methods 0.000 claims description 9
- 238000001069 Raman spectroscopy Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 238000011282 treatment Methods 0.000 claims description 4
- 230000009977 dual effect Effects 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 2
- 238000000926 separation method Methods 0.000 claims 1
- 238000012545 processing Methods 0.000 abstract description 3
- 239000000523 sample Substances 0.000 description 17
- 206010028980 Neoplasm Diseases 0.000 description 11
- 230000003902 lesion Effects 0.000 description 5
- 238000000799 fluorescence microscopy Methods 0.000 description 3
- 230000004927 fusion Effects 0.000 description 3
- 230000037311 normal skin Effects 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000005286 illumination Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 208000015181 infectious disease Diseases 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 241000701806 Human papillomavirus Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000001574 biopsy Methods 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 201000011510 cancer Diseases 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000002458 infectious effect Effects 0.000 description 1
- 238000003780 insertion Methods 0.000 description 1
- 230000037431 insertion Effects 0.000 description 1
- 230000031700 light absorption Effects 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 230000036210 malignancy Effects 0.000 description 1
- 238000011580 nude mouse model Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 238000011321 prophylaxis Methods 0.000 description 1
- 102000004169 proteins and genes Human genes 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 238000004451 qualitative analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/03—Recognition of patterns in medical or anatomical images
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)
Abstract
The invention belongs to the technical field of medical image processing, and discloses a classifier construction method based on Raman spectrum and a support vector machine, which is applied to cervical cancer screening and comprises the following steps: averaging the spectrum values corresponding to the same pixel in the Raman spectrum images acquired from different positions to obtain self-fluorescence spectrum image data; preprocessing the self-fluorescence spectrum image data; extracting features of the self-fluorescence spectrum image data and taking the self-fluorescence spectrum image data as a training sample; and taking the minimum interval distance between the training sample and the classification hyperplane as a constraint condition of the support vector machine, and acquiring an optimal classification decision function of the support vector machine, thereby obtaining the support vector machine classifier. The Raman spectrum information of the cervical part of the subject is acquired, class labels of pixels to be classified are classified according to the Raman spectrum characteristics of different tissues, and the acquired Raman spectrum images are classified through a support vector machine, so that the full utilization of the Raman spectrum information is ensured, and a stable classification result is acquired rapidly.
Description
Technical Field
The invention relates to the technical field of medical image processing, in particular to a classifier construction method based on Raman spectrum and a support vector machine.
Background
Cervical cancer is the second most common cancer in women, and cervical cancer occurs in association with infection by human papillomaviruses. Some high-risk infectious patients are easy to generate precancerous lesions and cervical cancer when continuously infected, and seriously endanger life health. Cervical cancer is one of the few malignancies that can reduce morbidity and mortality by prophylaxis. The current standard flow for diagnosing cervical cancer is to find suspicious lesions through a colposcope, take biopsies of the suspicious lesions to make pathological sections and then diagnose the suspicious lesions.
Currently, classification methods based on spectrum information exist, but due to the lack of utilization of spatial context information, classification results of the methods generally have a large number of noise spots, and it is difficult to meet the application requirements of hyperspectral images. When dealing with ultra-complex surfaces, particularly when the pixels to be classified are in heterogeneous areas, the distinguishing performance of the current method based on spatial spectrum information fusion is reduced due to interference of heterogeneous pixels, and in addition, such methods generally require long operation time due to the involvement of spatial spectrum information fusion.
Therefore, how to provide a classifier applied to cervical cancer screening, and to improve the operation efficiency of the algorithm while ensuring the classification accuracy of the classification method based on spectral information applied to cervical cancer screening is a current urgent problem to be solved.
Disclosure of Invention
In view of the above, the embodiment of the invention provides a classifier construction method based on Raman spectrum and a support vector machine, which aims to solve the problem that the classification method based on spectrum information adopted for cervical cancer screening in the prior art has insufficient operation efficiency under the condition of spatial spectrum information fusion.
The embodiment of the invention provides a classifier construction method based on Raman spectrum and a support vector machine, which is applied to cervical cancer screening and comprises the following steps:
averaging the spectrum values corresponding to the same pixel in the Raman spectrum images acquired from different positions to obtain self-fluorescence spectrum image data;
preprocessing the self-fluorescence spectrum image data;
extracting features of the self-fluorescence spectrum image data and taking the self-fluorescence spectrum image data as a training sample;
and taking the minimum interval distance between the training sample and the classification hyperplane as a constraint condition of the support vector machine, and acquiring an optimal classification decision function of the support vector machine, thereby obtaining the support vector machine classifier.
Optionally, preprocessing the auto fluorescence spectrum image data includes:
normalizing the self-fluorescence spectrum image data;
carrying out noise reduction treatment on the self-fluorescence spectrum image data through the sliding average value;
baseline correction was performed on the auto-fluorescence spectrum image data.
Optionally, baseline correction of the auto-fluorescence spectral image data includes:
approximating the original spectrum by constructing polynomial fitting to obtain a fitting spectrum;
eliminating peaks in the fitted spectrum;
if the fitting residual meets the preset condition, outputting a fitting value as a base line;
subtracting the baseline to obtain a fitting spectrum after baseline correction;
wherein, the preset condition is set as follows: the fit residual was less than 0.05.
Optionally, removing peaks in the fitted spectrum, comprising:
if the actual value of the pixel in the original spectrum is larger than the fitting value in the corresponding fitting spectrum, eliminating the actual value;
if the actual value of the pixel in the original spectrum is smaller than the fitting value in the corresponding fitting spectrum, the actual value is reserved.
Optionally, the method further comprises:
if the fitting residual error does not meet the preset condition, performing spectrum reconstruction;
wherein the spectral reconstruction comprises:
the data of which the actual value of the pixel is larger than the corresponding fitting value in the original spectrum is replaced by the corresponding fitting value; data in the original spectrum with actual values of pixels smaller than corresponding fitting values retain the original data.
Optionally, feature extraction is performed on the auto fluorescence spectrum image data, including:
extracting a peak value in the fitting spectrum as a characteristic quantity; or, the area difference between the spectral line of the fitting spectrum and the coordinate axis is taken as the characteristic quantity.
Optionally, before preprocessing the auto fluorescence spectrum image data, the method further comprises:
and dividing the pixels to be classified into different class labels according to the Raman spectrum characteristics of different tissues.
Optionally, taking the minimum interval distance between the training sample and the classification hyperplane as a constraint condition of the support vector machine, including:
geometric spacing of training sample sets to classification hyperplanesA representation; converting the search for a classification hyperplane into the following constraint optimization problem:
wherein ω is a normal vector of the classification hyperplane; b is the intercept; x is x i Is the ith feature vector; n is the total number of training samples; y is i Is x i Category labels of (c); t is a training set;
performing dual optimization by using a Lagrangian multiplier method to obtain optimal solutions omega and b;
for the nonlinear classification problem, after nonlinear transformation is performed through a kernel function, a classification decision function is constructed to determine the attribute of each test sample.
Optionally, the method further comprises:
selecting a linear kernel function, an RBF kernel function, a polynomial kernel function and a Sigmoid kernel function to respectively construct four classification decision functions;
training the support vector machine model by adopting four classification decision functions respectively;
and forming a support vector machine classifier by adopting a classification decision function with highest classification accuracy.
The embodiment of the invention has the beneficial effects that:
the embodiment of the invention provides a classifier construction method based on Raman spectrum and a support vector machine, which is characterized in that Raman spectrum information of a cervical part of a detected person is obtained, raman spectrums are collected from different spatial positions, each spectrum is reduced to a corresponding pixel with only one value, and each Raman spectrum is expressed as one value of a corresponding pixel point. After imaging, classifying class labels of pixels to be classified according to Raman spectrum characteristics of different tissues, classifying the obtained Raman spectrum images through a support vector machine, and ensuring full utilization of Raman spectrum information and rapidly obtaining stable classification results.
Drawings
The features and advantages of the present invention will be more clearly understood by reference to the accompanying drawings, which are illustrative and should not be construed as limiting the invention in any way, in which:
FIG. 1 shows a flowchart of a classifier construction method based on Raman spectrum and support vector machine in an embodiment of the invention;
fig. 2 shows a block diagram of an endoscope system for simultaneous fluorescence imaging and spectrum acquisition in an embodiment of the present invention.
FIG. 3 shows spectral information of a normal skin in an embodiment of the invention;
FIG. 4 shows spectral information of a tumor region in an embodiment of the invention;
fig. 5 shows a test result of a classifier construction method based on raman spectrum and support vector machine in an embodiment of the invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to fall within the scope of the invention.
The embodiment of the invention provides a classifier construction method based on Raman spectrum and a support vector machine, which is applied to cervical cancer screening, as shown in figure 1, and comprises the following steps:
and step S10, the spectrum value corresponding to the same pixel in the Raman spectrum images acquired from different positions is averaged to obtain the self-fluorescence spectrum image data.
The image acquisition is performed by an endoscope system which synchronously realizes fluorescence imaging and complete spectrum acquisition as shown in fig. 2, wherein the fluorescence imaging is used for quickly browsing the region to be detected, and the spectrum is used for performing qualitative analysis when suspicious targets are found. In this system, an image-transmitting optical fiber bundle and a probe are used as an endoscope main body to transmit illumination light and signal light, and a notch filter and a beam splitter are used to ensure that only the signal light enters a CCD and a spectrometer for imaging and spectrum acquisition. The whole system adopts a rapid disassembly and assembly design, and can complete the replacement of the laser and the optical element group within 30 seconds, thereby realizing rapid matching of different applications related to wavelength on a hardware level. The endoscope probe is used as a unique insertion part through a series of optical fiber couplings, and is simultaneously used as a unique channel for laser illumination and fluorescence signal collection, and is respectively led into a camera and a spectrometer.
Step S11, different class labels are divided for the pixels to be classified according to the Raman spectrum characteristics of different tissues.
In this embodiment, different pixels, such as protein, fat, blood, etc., are distinguished by raman spectral features, so that the pixels in the non-lesion area can be rapidly screened out, and the subsequent calculation amount is reduced.
Step S20, preprocessing the auto fluorescence spectrum image data.
In this embodiment, the preprocessing includes: and carrying out normalization processing on the self-fluorescence spectrum image data. And carrying out noise reduction treatment on the self-fluorescence spectrum image data through the sliding average value. Baseline correction was performed on the auto-fluorescence spectrum image data.
In a specific embodiment, the spectral data is converted to a fraction between [0,1] by linear function normalization (Min-Max Normalization), the conversion function being as follows:
where x is the original data, min is the minimum value of all sample data, and max is the maximum value of all sample data.
The sliding average filtering includes: and translating a window along the spectrum vector, and obtaining the average value of elements in the window one by one to replace the element in the central position of the window in the original spectrum, so that the noise-reduced spectrum can be obtained. The formula is as follows:
wherein p is t Representing the filtered value at the center of the window, N is the total number of pixels in the window, g t Representing the spectral value of the t-th pixel within the window.
Wherein the baseline correction comprises: and approximating the original spectrum by constructing polynomial fitting to obtain a fitting spectrum. The peaks in the fitted spectrum are eliminated. And if the fitting residual meets the preset condition, outputting a fitting value as a base line. Subtracting the baseline to obtain a fitting spectrum after baseline correction. The preset conditions are set as follows: the fit residual was less than 0.05.
The elimination of peaks in the fitted spectrum includes: if the actual value of the pixel in the original spectrum is larger than the fitting value in the corresponding fitting spectrum, eliminating the actual value; if the actual value of the pixel in the original spectrum is smaller than the fitting value in the corresponding fitting spectrum, the actual value is reserved. If the fitting residual error does not meet the preset condition, performing spectrum reconstruction: the data of which the actual value of the pixel is larger than the corresponding fitting value in the original spectrum is replaced by the corresponding fitting value; data in the original spectrum with actual values of pixels smaller than corresponding fitting values retain the original data.
And step S30, extracting features of the self-fluorescence spectrum image data and taking the extracted features as a training sample.
In the present embodiment, a peak value in the fitted spectrum is extracted as a feature quantity; or, the area difference between the spectral line of the fitting spectrum and the coordinate axis is taken as the characteristic quantity.
And S40, taking the minimum interval distance between the training sample and the classification hyperplane as a constraint condition of the support vector machine, and acquiring an optimal classification decision function of the support vector machine, thereby obtaining the support vector machine classifier.
In this embodiment, taking the minimum interval distance between the training sample and the classification hyperplane as the constraint condition of the support vector machine includes:
geometric spacing of training sample sets to classification hyperplanesA representation; converting the search for a classification hyperplane into the following constraint optimization problem:
wherein ω is a normal vector of the classification hyperplane; b is the intercept; x is x i Is the ith feature vector; n is the total number of training samples; y is i Is x i Category labels of (c); t is a training set; at a given training data set t= { (x) 1 ，y 1 )，...，(x n ，y n )}，y i ∈{-1，+1}。
And performing dual optimization by using a Lagrangian multiplier method to obtain optimal solutions omega and b.
For the nonlinear classification problem, after nonlinear transformation is performed through a kernel function, a classification decision function is constructed to determine the attribute of each test sample.
The Support Vector Machine (SVM) has unique advantages in solving the problem of linearity unavailability due to high feature dimensions in small sample, hyperspectral medical images. The basic principle is to solve a classification hyperplane which can correctly divide the training data set into two classes and has the largest geometric interval.
Let ω be T x+b=0 is the classification hyperplane, determined by its normal vector ω and intercept b, and the support vector is the nearest point to this hyperplane.
After nonlinear transformation is carried out on the support vector machine through a kernel function, a classification decision function f (x) is constructed, and the attribute of each sample is determined according to the size of the classification decision function value:
where SVs is a set of support vectors representing samples corresponding to non-zero Lagrangian multipliers, K (·) is a kernel function.
As an alternative embodiment, further comprising:
selecting a linear kernel function, an RBF kernel function, a polynomial kernel function and a Sigmoid kernel function to respectively construct four classification decision functions;
training the support vector machine model by adopting four classification decision functions respectively;
and forming a support vector machine classifier by adopting a classification decision function with highest classification accuracy.
In this embodiment, the linear kernel function:
RBF kernel function:
polynomial kernel function:
sigmoid kernel function:
wherein σ and c are constants.
In a specific embodiment, a nude mouse with a tumor is placed in a shading container to simulate the working environment of a cervical mirror. The tumor part is found out quickly by using the image, and the spectrum information of the tumor area and the normal skin is acquired respectively. Fig. 3 and 4 are raw spectral data, and it can be seen that the spectral intensity of the tumor region is generally weaker by nearly an order of magnitude than that of normal tissue, presumably because of the light absorption of a large amount of tissue fluid in the tumor region and the flatter spectral trend.
The self-fluorescence spectra of the skin and the tumor at different positions are acquired for the same sample, and the classification treatment is carried out by the constructed classification method based on the Raman spectrum and the support vector machine. As can be seen from fig. 5, the spectrum of the tumor area is flatter in the red and green bands, the spectrum of normal skin has higher consistency, and the relative intensities of the tumors are very non-uniform in the red band.
Although embodiments of the present invention have been described in connection with the accompanying drawings, various modifications and variations may be made by those skilled in the art without departing from the spirit and scope of the invention, and such modifications and variations are within the scope of the invention as defined by the appended claims.
Claims (9)
1. A classifier construction method based on Raman spectrum and support vector machine is applied to cervical cancer screening, and is characterized by comprising the following steps:
averaging the spectrum values corresponding to the same pixel in the Raman spectrum images acquired from different positions to obtain self-fluorescence spectrum image data;
preprocessing the self-fluorescence spectrum image data;
extracting features of the self-fluorescence spectrum image data and taking the self-fluorescence spectrum image data as a training sample;
and taking the minimum interval distance between the training sample and the classification hyperplane as a constraint condition of the support vector machine, and obtaining an optimal classification decision function of the support vector machine, thereby obtaining the support vector machine classifier.
2. The method for constructing a classifier based on raman spectroscopy and support vector machine according to claim 1, wherein preprocessing the self-fluorescence spectrum image data comprises:
normalizing the self-fluorescence spectrum image data;
carrying out noise reduction treatment on the self-fluorescence spectrum image data through a sliding average value;
baseline correction is performed on the self-fluorescence spectrum image data.
3. The method of constructing a classifier based on raman spectroscopy and support vector machine according to claim 2, wherein baseline correction is performed on the self-fluorescence spectrum image data, comprising:
approximating the original spectrum by constructing polynomial fitting to obtain a fitting spectrum;
eliminating peaks in the fitted spectrum;
if the fitting residual meets the preset condition, outputting a fitting value as a base line;
subtracting the baseline to obtain the fitting spectrum after baseline correction;
wherein the preset conditions are set as follows: the fit residual was less than 0.05.
4. A raman spectrum and support vector machine based classifier construction method according to claim 3, wherein eliminating peaks in the fitted spectrum comprises:
if the actual value of the pixel in the original spectrum is larger than the fitting value in the corresponding fitting spectrum, eliminating the actual value;
and if the actual value of the pixel in the original spectrum is smaller than the fitting value in the corresponding fitting spectrum, reserving the actual value.
5. A raman spectrum and support vector machine based classifier construction method according to claim 3, further comprising:
if the fitting residual error does not meet the preset condition, performing spectrum reconstruction;
wherein the spectral reconstruction comprises:
the data of which the actual value of the pixel is larger than the corresponding fitting value in the original spectrum is replaced by the corresponding fitting value; and the data of which the actual pixel value is smaller than the corresponding fitting value in the original spectrum retain the original data.
6. The method for constructing a classifier based on raman spectroscopy and support vector machine according to claim 5, wherein the feature extraction of the self-fluorescence spectrum image data comprises:
extracting a peak value in the fitted spectrum as a characteristic quantity; or taking the area difference between the spectral line of the fitting spectrum and the coordinate axis as the characteristic quantity.
7. The method of constructing a raman spectrum and support vector machine based classifier according to claim 1, further comprising, prior to preprocessing the auto-fluorescence spectrum image data:
and dividing the pixels to be classified into different class labels according to the Raman spectrum characteristics of different tissues.
8. The classifier construction method based on raman spectroscopy and support vector machine according to claim 1, wherein taking the minimum separation distance between the training sample and the classification hyperplane as a constraint condition of support vector machine comprises:
geometric spacing of training sample sets to the classification hyperplaneA representation; converting the search for a classification hyperplane into the following constraint optimization problem:
wherein ω is a normal vector of the classification hyperplane; b is the intercept; x is x i Is the ith feature vector; n is the total number of training samples; y is i Is x i Category labels of (c); t is a training set;
performing dual optimization by using a Lagrangian multiplier method to obtain optimal solutions omega and b;
for the nonlinear classification problem, after nonlinear transformation is performed through a kernel function, a classification decision function is constructed to determine the attribute of each test sample.
9. The method for constructing a classifier based on raman spectroscopy and support vector machine according to claim 8, further comprising:
selecting a linear kernel function, an RBF kernel function, a polynomial kernel function and a Sigmoid kernel function to respectively construct four classification decision functions;
training the support vector machine model by adopting four classification decision functions respectively;
and forming the support vector machine classifier by adopting a classification decision function with highest classification accuracy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310725163.5A CN116645565A (en) | 2023-06-19 | 2023-06-19 | Classifier construction method based on Raman spectrum and support vector machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310725163.5A CN116645565A (en) | 2023-06-19 | 2023-06-19 | Classifier construction method based on Raman spectrum and support vector machine |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116645565A true CN116645565A (en) | 2023-08-25 |
Family
ID=87615421
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310725163.5A Pending CN116645565A (en) | 2023-06-19 | 2023-06-19 | Classifier construction method based on Raman spectrum and support vector machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116645565A (en) |
-
2023
- 2023-06-19 CN CN202310725163.5A patent/CN116645565A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8483454B2 (en) | Methods for tissue classification in cervical imagery | |
| US6902935B2 (en) | Methods of monitoring effects of chemical agents on a sample | |
| US8401258B2 (en) | Method to provide automated quality feedback to imaging devices to achieve standardized imaging data | |
| US9436992B2 (en) | Method of reconstituting cellular spectra useful for detecting cellular disorders | |
| JP2015500722A (en) | Method and apparatus for detecting and quantifying skin symptoms in a skin zone | |
| Zheludev et al. | Delineation of malignant skin tumors by hyperspectral imaging using diffusion maps dimensionality reduction | |
| Vocaturo et al. | Features for melanoma lesions characterization in computer vision systems | |
| US20210374953A1 (en) | Methods for automated detection of cervical pre-cancers with a low-cost, point-of-care, pocket colposcope | |
| Obukhova et al. | Automated image analysis in multispectral system for cervical cancer diagnostic | |
| Srinivasan et al. | A probabilistic approach to segmentation and classification of neoplasia in uterine cervix images using color and geometric features | |
| CN116645565A (en) | Classifier construction method based on Raman spectrum and support vector machine | |
| Chand et al. | Identifying oral cancer using multispectral snapshot camera | |
| Pallavi et al. | Automated analysis of cervix images to grade the severity of cancer | |
| Yang et al. | Computer-aided cervical cancer screening method based on multi-spectral narrow-band imaging | |
| de Moura et al. | Skin lesions classification using multichannel dermoscopic Images | |
| Andrade et al. | Automatic Segmentation of the Cervical Region in Colposcopic Images. | |
| US20240065540A1 (en) | Apparatus and method for detecting cervical cancer | |
| JP7449004B2 (en) | Hyperspectral object image detection method using frequency bands | |
| Li et al. | Terraced Compression Method with Automated Threshold Selection for Multidimensional Image Clustering of Heterogeneous Bodies | |
| Gu et al. | Automatic image quality assessment for uterine cervical imagery | |
| Vega et al. | Feasibility Study of Hyperspectral Colposcopy as a Novel Tool for Detecting Precancerous Cervical Lesions | |
| Kumar et al. | Pioneering Skin Health: Harnessing Thermal Images for Cutting-Edge Computer-Aided Diagnosis | |
| Van Raad et al. | Emerging Technologies, Signal Processing and Statistical Methods for Screening of Cervical Cancer In Vivo: Are They Good Candidates for Cervical Screening? | |
| Natalia et al. | Image processing in clinical decision support system | |
| CN112116027A (en) | Skin cancer classification method based on optical intensity and gradient of OCT imaging image |
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 |