An auxiliary examination tool for Parkinson's disease based on mitochondrial metabolic analysis
Through optical coherence tomography, the outer plexus layer of the retinal plexus was analyzed, the mitochondrial spatial cluster index was obtained, and the support vector machine model was established, which solved the complexity and non-invasive problems of Parkinson's disease diagnosis in the existing technology, and achieved rapid Accurate Parkinson's disease auxiliary test.
Patent Information
- Application Number
- CN202410488165.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-23
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-04-23
AI Technical Summary
The existing Parkinson's disease diagnosis methods have the disadvantages of complex operation, long-term operation, relying on doctor experience, and inability to achieve non-invasive metabolic analysis, and are particularly difficult to perform fluorescence imaging and cell metabolism analysis of the human brain.
Optical coherence tomography technology is used to image the outer plexus layer of the retinal plexus, through pixel cloning, Fourier transform, power spectral density calculation and linear fitting, mitochondrial spatial cluster index is obtained, and a support vector machine model is established to achieve auxiliary Parkinson's disease. examine.
It has achieved rapid, non-invasive and low-cost Parkinson's disease diagnosis, which can accurately analyze the mitochondrial metabolism status of the extraretinal plexus, and improves the accuracy and application potential of the diagnosis.
Smart Images

Figure CN118436308B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of biological tissue image analysis and disease detection, and in particular relates to a Parkinson's disease auxiliary examination tool based on mitochondrial metabolic analysis. Background Art
[0002] Parkinson's disease (PD) is a chronic neurodegenerative disease that affects the central nervous system, mainly the motor nervous system. Symptoms usually appear slowly over time. The most obvious early symptoms are tremors, limb stiffness, motor dysfunction and abnormal gait. There may also be cognitive and behavioral problems. The diagnosis of Parkinson's disease relies on medical history and neurological examinations. Currently, there is no test method that can effectively confirm Parkinson's disease. It can only be confirmed by excluding other diseases with similar symptoms through brain imaging, or by observing the improvement of motor symptoms after the patient takes L-dopa. Existing methods have obvious disadvantages. The operation is complicated, time-consuming, and highly dependent on the doctor's experience.
[0003] At present, it is known that Parkinson's disease can cause brain neuron death, accompanied by functional damage and reduced activity of neuronal mitochondria. In the field of cell metabolism, analyzing cell mitochondrial metabolism by spectral analysis of fluorescence images is a relatively mature method. Mitochondria can undergo continuous cycles of fusion and fission. In the two-photon fluorescence images of cells, metabolically damaged mitochondria will show more fissions and clusters, and fewer fusions and cross-links. This characteristic will lead to increased details in the fluorescence image. Based on this characteristic, the fluorescence image is analyzed in the frequency domain. The increase in high-frequency components in the power spectral density spectrum indicates an increase in the details of the fluorescence image, indicating that the mitochondria show more fissions and clusters, and the cell metabolism is hindered.
[0004] Optical Coherence Tomography (OCT) is a new and most promising tomographic imaging technology that has developed rapidly in recent years and has been widely used in clinical ophthalmology examinations. OCT technology can quickly obtain high-resolution images of the human retina. As an extension of the nervous system, the mitochondria in the outer plexiform layer of the retina will also suffer from metabolic damage after suffering from Parkinson's disease. Compared with fluorescence imaging technology, OCT images are easier to obtain, can perform non-invasive in vivo imaging, and are already very mature in clinical examinations.
[0005] Existing diagnostic methods for Parkinson's disease are still very scarce and have certain limitations. First, it takes too long to confirm Parkinson's disease by observing the improvement of the condition after taking medication, and it is impossible to make an immediate clinical diagnosis; second, when using the emerging positron emission tomography / X-ray computed tomography (PET / CT) technology to diagnose Parkinson's disease, radioactive tracers need to be injected, and label-free imaging diagnosis cannot be achieved. It is costly and has great damage to the human body; third, traditional two-photon fluorescence imaging is currently limited to in vivo imaging of mice, and cannot be used for fluorescence imaging and cell metabolism analysis of the human body, especially the brain, in clinical practice. Summary of the invention
[0006] In view of the shortcomings of the prior art, the present invention proposes an auxiliary examination tool for Parkinson's disease based on mitochondrial metabolic analysis. This method uses a clinical OCT system to image the retina, manually segment the outer plexiform layer, and obtain the mitochondrial spatial cluster index of the outer plexiform layer by performing pixel cloning, Fourier transform, power spectral density calculation and linear fitting on the mitochondria in the outer plexiform layer. A classification model is established based on this index to achieve auxiliary examination of Parkinson's disease. This method avoids the shortcomings of using traditional methods to diagnose Parkinson's disease, which are time-consuming and labor-intensive, and uses OCT images for metabolic analysis, breaking through the difficulty of fluorescence imaging in achieving in vivo metabolic analysis of humans.
[0007] The technical solution adopted by the present invention is:
[0008] The present invention discloses an auxiliary inspection tool for Parkinson's disease based on mitochondrial metabolic analysis, which is characterized in that the mitochondrial metabolism of the outer plexiform layer (OPL) of the retina is analyzed by optical coherence tomography (OCT) images to establish an auxiliary inspection model. The inspection tool comprises the following steps:
[0009] 1) Use optical coherence tomography to radially scan human eyes with known disease information to obtain retinal OCT images with micron-level resolution;
[0010] 2) Preprocess the retinal OCT images by interpolation and cropping, and manually extract the outer plexiform layer of the retina to remove the background and signals from other retinal layers to obtain the outer plexiform layer image;
[0011] 3) Divide the outer plexiform layer image into two layers, upper and lower, and perform frequency domain analysis on the upper and lower outer plexiform layer images respectively to obtain the power spectrum density spectra of the upper and lower outer plexiform layer;
[0012] 4) Linear fitting was performed on the power spectrum density line to obtain the spatial cluster indexes β1 and β2 of the upper and lower mitochondria of the outer plexiform layer, as well as the difference β3 between β2 and β1, which were used as multivariate characteristic parameters;
[0013] 5) constructing a support vector machine model based on the numerical results of the multivariate characteristic parameters obtained in step 4) and the known disease information in step 1);
[0014] 6) In the support vector machine model, each independent sample is vectorized according to its multivariate parameter values. The multivariate parameters include the spatial cluster indexes β1 and β2 of the upper and lower mitochondria of the outer plexiform layer, and the difference β3. The multivariate calculation results constitute the vector Among them, v i ,ω i , are the values of β1, β2 and β3 calculated in step 4) respectively;
[0015] The decision plane equation obtained according to multivariate parameters is:
[0016]
[0017] in is the normal vector of the decision plane, b is the intercept of the decision plane, and the judgment conditions for whether the human eye sample suffers from Parkinson's disease are as follows:
[0018]
[0019] y i represents the known disease state of Parkinson's disease, y i =1 means the image is normal to human eyes, y i =-1 means the image is a human eye image with Parkinson's disease;
[0020] 7) Through training with a large number of multivariate parameters and label information, the decision surface normal vector can be obtained And intercept b, the specific training is to find the conditional maximum problem:
[0021]
[0022] Where st means simultaneous satisfaction, and m is the total number of samples;
[0023] 8) After training, the decision surface normal vector is obtained and intercept b, thus obtaining an auxiliary examination tool for Parkinson's disease based on mitochondrial metabolic analysis.
[0024] As a further improvement, in step 3) of the present invention, the outer plexiform layer image is divided into upper and lower layers, specifically, the outer plexiform layer is divided into an upper layer and a lower layer with the low reflection band existing in the outer plexiform layer OCT image as the boundary. There are few mitochondria in this low reflection band. The upper layer of the outer plexiform layer above the low reflection band is populated with mitochondria of bipolar cells, while the lower layer of the outer plexiform layer below the low reflection band is populated with mitochondria of photoreceptor cells. The mitochondria of the two types of cells have different morphological distributions and metabolic conditions.
[0025] As a further improvement, in step 3) of the present invention, the frequency domain analysis of the upper and lower outer plexiform layer images is specifically performed by first cloning the mitochondrial pixels, filling the entire image with mitochondrial signals, and then performing Fourier transform on the filled image to obtain the image power spectrum. It should be noted that when performing pixel cloning, since a single image contains fewer outer plexiform layer mitochondrial signals, multiple OCT images of the same human eye sample are cloned together as signal sources for cloned pixels and used for subsequent analysis.
[0026] As a further improvement, in step 4) of the present invention, the power spectrum density curve is linearly fitted, and the slope value of the straight line is the spatial cluster index of mitochondria in the outer plexiform layer, which represents the spatial clustering of mitochondrial distribution and reflects the strength of cell metabolism. Specifically, in cell metabolism, mitochondria in cells with weak metabolism divide more and then cluster together, and the mitochondrial spatial cluster index is high, while mitochondria in cells with strong metabolism are more cross-linked and fused, and the mitochondrial spatial cluster index is low.
[0027] As a further improvement, in step 8) of the present invention, the decision surface normal vector is obtained through training Intercept b = 2.59044859.
[0028] The present invention uses optical coherence tomography technology, based on the mitochondrial metabolism of the outer plexiform layer of the retina, with high precision, rapid and non-invasive, and can be examined in vivo. By performing power spectral density analysis of the outer plexiform layer mitochondria in the frequency domain and calculating the spatial cluster index of the mitochondria, the metabolic analysis of the outer plexiform layer of the retina is achieved, and the established Parkinson's disease auxiliary examination tool is used to assist doctors in examinations, further enhancing the understanding of the pathology of human Parkinson's disease.
[0029] Due to the application of the above technical solution, the present invention has the following advantages compared with the existing technical solution:
[0030] The present invention discloses a Parkinson's disease auxiliary examination tool based on mitochondrial metabolic analysis, which is used to analyze the mitochondrial metabolic status of the outer plexiform layer of the retina of the human eye and diagnose metabolic abnormalities caused by Parkinson's disease. The present invention can perform Fourier analysis on mitochondrial images to obtain image power spectra, and then perform linear fitting on the power spectrum curve to obtain characteristic parameters representing the degree of mitochondrial clustering. According to this parameter, a Parkinson's disease auxiliary examination model is established, and the model can automatically perform Parkinson's disease examinations. The present invention can extract the distribution information of mitochondria in the outer plexiform layer of the retina. Aiming at the difference in the mitochondrial metabolic status of the outer plexiform layer of the retina between normal people and Parkinson's patients, the present invention proposes a new examination method. The method can process and analyze high-resolution images obtained by optical coherence tomography commonly used in clinical practice, and can more accurately analyze the mitochondrial metabolic status. Compared with traditional Parkinson's diagnosis methods, it has greater application potential.
[0031] 1. The present invention abandons the previous method of diagnosing Parkinson's disease by observing brain metabolism through metabolic imaging and structural imaging of the human brain, and instead performs imaging and metabolic analysis on the retina, an extension of the nervous system. Due to the fragility of the brain, imaging it not only requires sufficient penetration depth, but also requires the utmost effort to minimize damage. In comparison, imaging and analyzing the retina is obviously easier.
[0032] 2. The present invention extracts and analyzes the features of the OCT images of the outer plexiform layer of the retina to obtain a quantitative mitochondrial spatial cluster index, and analyzes the mitochondrial metabolism from the index. Using OCT images for metabolic analysis breaks the previous practice of using fluorescent imaging or nuclear medicine imaging with radioactive tracers for metabolic imaging analysis, and can achieve rapid, accurate, non-invasive, and low-cost diagnosis of Parkinson's disease.
[0033] 3. The present invention analyzes the mitochondrial image of the outer plexiform layer in the frequency domain. By calculating the power spectrum density and fitting the power spectrum density curve, the distribution of the high-frequency and low-frequency information of the image can be judged more globally. At the same time, according to the size of the mitochondria, the partial power spectrum curve corresponding to the mitochondrial information can be accurately found, and this part can be observed and analyzed emphatically. Compared with directly observing and quantifying the morphological distribution of mitochondria at the spatial domain level, this frequency domain analysis method is more comprehensive and global, and can more effectively quantify the spatial cluster index of mitochondrial information.
[0034] 4. The present invention performs frequency domain analysis on the mitochondrial signals of the outer plexiform layer. When performing pixel cloning and filling, since a single image contains fewer mitochondrial signals of the outer plexiform layer, multiple OCT images of the same human eye sample are cloned together as signal sources for cloned pixels and used for subsequent analysis. Compared with frequency domain analysis of a single image, this method improves the information utilization of the OCT image, and the calculated parameter values have higher accuracy and stability, effectively increasing the accuracy of the inspection tool.
[0035] 5. The present invention realizes the layered analysis of the OCT image of the outer plexiform layer. The basis for the layering of the outer plexiform layer is that there is a low-reflection band in the OCT image of the outer plexiform layer. There are few mitochondria in this low-reflection band. Mitochondria of bipolar cells are gathered above the low-reflection band, and mitochondria of photoreceptor cells are gathered below the low-reflection band. Therefore, the mitochondria of the upper and lower layers of the outer plexiform layer have different morphology and metabolism. The upper and lower layers of the outer plexiform layer are analyzed separately, so that the present invention can calculate the characteristic parameters of both the upper and lower layers of the outer plexiform layer, and can analyze the inter-layer difference of the characteristic parameters, which increases the dimension of the present invention to extract image features, which is of great significance for revealing more detailed metabolic information in the outer plexiform layer. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] Figure 1 A flowchart for constructing an auxiliary examination tool for Parkinson's disease based on mitochondrial metabolic analysis;
[0037] Figure 2 This is a comparison of the retinal reflection band under OCT and the retinal mitochondria under electron microscopy;
[0038] Figure 3 This is the frequency domain analysis result of the OCT image of the outer plexiform layer mitochondria;
[0039] Figure 4 Schematic diagram of classification accuracy based on mitochondrial spatial cluster indicators. DETAILED DESCRIPTION
[0040] In order to make the purpose, technical solution and advantages of the present invention more clear, the technical solution of the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention and are not used to limit the present invention.
[0041] The present invention is a Parkinson's disease auxiliary examination tool based on mitochondrial metabolic analysis, which specifically comprises the following steps: Figure 1 As shown:
[0042] 1) Retinal OCT imaging: Use clinical OCT instruments to image the retina of the human eye and display the signals of each layer of the retina. The clinical OCT system uses radial scanning to obtain retinal B-scan images of the fundus around each person, a total of 18 images, such as Figure 3 (a) shown.
[0043] 2) Preprocess the obtained high-resolution image. Perform the same preprocessing on the 18 images of each sample. Since the subsequent frequency domain analysis requires the image to have consistent horizontal and vertical resolution, the image is first nonlinearly interpolated to obtain a retinal image with consistent horizontal and vertical resolution. Then, take the left part of the fovea, crop the image, and obtain an image of the left part of the fovea. Finally, manually segment the image of the left part of the fovea, manually select the outer plexiform layer area of the image, and extract the upper OPL and lower OPL parts of each image.
[0044] 3) After preprocessing, the OPL upper layer and OPL lower layer of the image are processed separately, and the 18 OPL upper layer images are superimposed together to form mask1, as shown in Figure 3 (b) The same processing is performed on the image in the lower layer of OPL to obtain mask2.
[0045] 4) Clone the image pixels of mask1 using Figure 3 The pixels with OPL upper layer signals in (b) are used to fill the blank areas without signals to obtain the cloned image clone1, as shown in Figure 3 (c) As shown. Then perform Fourier transform on clone1 and obtain Figure 3 (d) The spectrum shown. Perform power spectrum density analysis on clone1, fit the inverse power law decay spectrum of the power spectrum density, and obtain the slope β1 of the fitting line, as shown in Figure 3 (e) is shown. Mask2 is also processed as above to obtain the slope of the fitted line β2. It should be noted that since we are analyzing mitochondrial metabolism, the signal area represented by mitochondria should be selected for analysis. It is known that the size of mitochondria is less than 10 microns, so the horizontal axis 10 is selected for analysis. -1 The spectrum line on the right is used for fitting, such as Figure 3 (e) as shown.
[0046] 5) Calculate the mitochondrial cluster indices β1 and β2 of the upper and lower layers of the OPL through step 4), and calculate the difference between β2 and β1 as the new index β3.
[0047] 6) Use β1, β2, and β3 to establish a single parameter classification model, and use the single classification model to classify the existing samples to obtain the single parameter classification accuracy as follows: Figure 4As shown in (a), (b), and (c). Since the classification accuracy that can be achieved by a single parameter is limited, a multi-parameter auxiliary examination model for Parkinson's disease is established by using a multi-parameter fusion analysis method based on the numerical results of the multi-parameter β1, β2, and β3 and the known disease information labels. This model is a support vector machine model, which has improved accuracy compared to single parameter analysis and can achieve high-accuracy auxiliary examination applications for Parkinson's disease.
[0048] 7) In the support vector machine model, each independent sample is vectorized according to its multivariate parameter values. The multivariate parameters include the spatial cluster indexes β1 and β2 of the upper and lower mitochondria of the outer plexiform layer, and the difference β3. The multivariate calculation results constitute the vector Among them, v i ,ω i , are the values of β1, β2 and β3 calculated in step 4) respectively;
[0049] The decision plane equation obtained according to multivariate parameters is:
[0050]
[0051] in is the normal vector of the decision plane, b is the intercept of the decision plane, and the judgment conditions for whether the human eye sample suffers from Parkinson's disease are as follows:
[0052]
[0053] y i represents the known disease state of Parkinson's disease, y i =1 means the image is normal to human eyes, y i =-1 means the image is a human eye image with Parkinson's disease;
[0054] 8) Through training with a large number of multivariate parameters and label information, the decision surface normal vector can be obtained And intercept b, the specific training is to find the conditional minimum problem:
[0055]
[0056] Where st means simultaneous satisfaction, and m is the total number of samples;
[0057] 9) After training, the decision surface normal vector is obtained and intercept b, thus obtaining an auxiliary examination tool for Parkinson's disease based on mitochondrial metabolic analysis.
[0058] In the process of training the multi-parameter Parkinson's disease auxiliary examination model, a large number of multivariate parameter calculation results and known disease information are used to construct a data set, and the data set is used to train the multi-parameter Parkinson's disease auxiliary examination model. The data set is divided into a training set and a test set according to the leave-one-out method. The training set is used to build and train the discriminant model, and the test set is used to test the model's discriminant ability for Parkinson's disease examination and optimize the detection model. The final multi-parameter Parkinson's disease auxiliary examination model has an OCA (original classification accuracy) of 83.3% and a CVCA (cross-validation classification accuracy) of 72.2%. The area under the ROC curve is as high as 0.901. The specific classification situation is as follows: Figure 4 As shown in (d) and (e).
[0059] It is worth mentioning that the use of clinical OCT images for mitochondrial analysis is a new approach to metabolic analysis. We compared the images of retinal mitochondria under electron microscopy with those under OCT to demonstrate the feasibility of this method. Figure 2 As shown in (a), the OPL layer signal is intercepted in the clinical OCT image and calculated as follows: Figure 2 (b) The longitudinal strength distribution curve. Figure 2 The 9 groups of sample images shown in (c) are all processed in this way and averaged to obtain the following Figure 2 (d) shows the average intensity curve of the OPL layer. When calculating the longitudinal distribution curve of the retinal mitochondrial signal under electron microscopy, the mitochondrial image obtained by the authors using electron microscopy and deep learning segmentation in the article (Volume electron microscopy reveals human retinal mitochondria that align with reflective bands in optical coherence tomography) was cited, as shown in the figure. Figure 2 As shown in (e), the longitudinal intensity distribution curve of the image is calculated and the following is obtained: Figure 2 (f) Mitochondrial distribution curve. Figure 2 As shown in (g), the longitudinal intensity distribution curve of OCT is compared with the intensity curve of mitochondria under electron microscopy. It is found that the trends of the two curves in the OPL layer are very consistent, proving that the OPL layer image under OCT also carries a large number of mitochondrial signals, which can be used to calculate mitochondrial cluster indicators and analyze mitochondrial metabolism.
[0060] Based on this idea, the present invention finally obtained a Parkinson's disease auxiliary examination tool based on mitochondrial metabolic analysis through the steps of optical imaging, image preprocessing, manual segmentation, frequency domain analysis, and construction of a multi-parameter fusion classification model. From the results, this method can process and analyze high-resolution images obtained by optical coherence tomography commonly used in clinical practice, can more accurately analyze the mitochondrial metabolic status, and can accurately classify normal and Parkinson's disease samples, which is of great significance in the auxiliary examination of the disease.
[0061] The above description of the embodiments is to facilitate the understanding and application of the present invention by those skilled in the art. It is obvious that those skilled in the art can easily make various modifications to the above embodiments and apply the general principles described herein to other embodiments without creative work. Therefore, the present invention is not limited to the above embodiments. Improvements and modifications made by those skilled in the art to the present invention based on the disclosure of the present invention should be within the protection scope of the present invention.
Claims
1. A Parkinson's disease auxiliary examination tool based on mitochondrial metabolic analysis, It is characterized in that The mitochondrial metabolism of the outer plexiform layer (OPL) of the retina is analyzed by optical coherence tomography (OCT) images, and an auxiliary inspection model is established. The inspection tool includes the following steps: 1) Use optical coherence tomography to radially scan human eyes with known disease information to obtain retinal OCT images with micron-level resolution; 2) Preprocess the retinal OCT images by interpolation and cropping, and manually extract the outer plexiform layer of the retina to remove the background and signals from other retinal layers to obtain the outer plexiform layer image; 3) Divide the outer plexiform layer image into two layers, upper and lower, and perform frequency domain analysis on the upper and lower outer plexiform layer images respectively to obtain the power spectrum density spectra of the upper and lower outer plexiform layer; 4) Linear fitting of the power spectral density spectrum was performed to obtain the spatial cluster index β of the upper and lower mitochondria in the outer plexiform layer 1 and β 2 , and β 2 and β 1 The difference β 3 , taking them as multivariate feature parameters; 5) constructing a support vector machine model based on the numerical results of the multivariate characteristic parameters obtained in step 4) and the known disease information in step 1); 6) In the support vector machine model, each independent sample is vectorized according to its multivariate parameter values. The multivariate parameters include the spatial clustering index β of the upper and lower mitochondria of the outer plexiform layer 1 and β 2 , and the difference β 3 , the multi-parameter calculation results form a vector Among them, v i ,ω i , are the β calculated in step 4) 1 and β 2 and β 3 The value of The decision plane equation obtained according to multivariate parameters is: in is the normal vector of the decision plane, b is the intercept of the decision plane, and the judgment conditions for whether the human eye sample suffers from Parkinson's disease are as follows: y i represents the known disease state of Parkinson's disease, y i =1 means the image is normal to human eyes, y i =-1 means the image is a human eye image with Parkinson's disease; 7) Through training with a large number of multivariate parameters and label information, the decision surface normal vector is obtained And intercept b, the specific training is to find the conditional maximum problem: Where st means simultaneous satisfaction, and m is the total number of samples; 8) After training, the decision surface normal vector is obtained and intercept b, thus obtaining an auxiliary examination tool for Parkinson's disease based on mitochondrial metabolic analysis.
2. The Parkinson's disease auxiliary examination tool based on mitochondrial metabolic analysis according to claim 1, Features: In the step 3), the outer plexiform layer image is divided into two layers, specifically, the outer plexiform layer is divided into an upper layer and a lower layer based on the low reflection band in the outer plexiform layer OCT image. There are few mitochondria in this low reflection band. The mitochondria of bipolar cells are concentrated in the upper layer of the outer plexiform layer above the low reflection band, while the mitochondria of photoreceptor cells are concentrated in the lower layer of the outer plexiform layer below the low reflection band. The mitochondria of the two types of cells have different morphological distributions and metabolic conditions.
3. The Parkinson's disease auxiliary examination tool based on mitochondrial metabolic analysis according to claim 1, Features: In the step 3), the frequency domain analysis is performed on the upper and lower outer plexiform layer images respectively, specifically, the mitochondrial pixels are first cloned, the entire image is filled with mitochondrial signals, and then the filled image is Fourier transformed to obtain the image power spectrum.
4. The Parkinson's disease auxiliary examination tool based on mitochondrial metabolic analysis according to claim 1, 2 or 3, Features: In the step 4), a linear fit is performed on the power spectrum density curve, and the slope value of the straight line is the spatial cluster index of mitochondria in the outer plexiform layer. The index represents the spatial clustering of mitochondrial distribution and reflects the strength of cellular metabolism.
5. The Parkinson's disease auxiliary examination tool based on mitochondrial metabolic analysis according to claim 1, Features: In step 8), the decision surface normal vector is obtained through training. Intercept b = 2.59044859.
