CN110827275A - Liver nuclear magnetic artery phase image quality grading method based on raspberry group and deep learning - Google Patents
Liver nuclear magnetic artery phase image quality grading method based on raspberry group and deep learning Download PDFInfo
- Publication number
- CN110827275A CN110827275A CN201911157464.2A CN201911157464A CN110827275A CN 110827275 A CN110827275 A CN 110827275A CN 201911157464 A CN201911157464 A CN 201911157464A CN 110827275 A CN110827275 A CN 110827275A
- Authority
- CN
- China
- Prior art keywords
- nuclear magnetic
- liver nuclear
- layer
- liver
- artery phase
- 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.)
- Granted
Links
- 210000004185 liver Anatomy 0.000 title claims abstract description 93
- 210000001367 artery Anatomy 0.000 title claims abstract description 69
- 238000000034 method Methods 0.000 title claims abstract description 30
- 240000007651 Rubus glaucus Species 0.000 title claims abstract description 27
- 235000011034 Rubus glaucus Nutrition 0.000 title claims abstract description 27
- 235000009122 Rubus idaeus Nutrition 0.000 title claims abstract description 27
- 238000013135 deep learning Methods 0.000 title claims abstract description 22
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 22
- 238000003908 quality control method Methods 0.000 claims abstract description 22
- 238000012800 visualization Methods 0.000 claims abstract description 17
- 238000011176 pooling Methods 0.000 claims abstract description 15
- 238000012549 training Methods 0.000 claims abstract description 15
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 238000012216 screening Methods 0.000 claims abstract description 8
- 230000007704 transition Effects 0.000 claims description 15
- 230000004913 activation Effects 0.000 claims description 12
- 238000010586 diagram Methods 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000003745 diagnosis Methods 0.000 description 12
- 208000014018 liver neoplasm Diseases 0.000 description 9
- 201000007270 liver cancer Diseases 0.000 description 8
- 238000002595 magnetic resonance imaging Methods 0.000 description 8
- 239000002872 contrast media Substances 0.000 description 6
- 208000019423 liver disease Diseases 0.000 description 5
- PWWVAXIEGOYWEE-UHFFFAOYSA-N Isophenergan Chemical compound C1=CC=C2N(CC(C)N(C)C)C3=CC=CC=C3SC2=C1 PWWVAXIEGOYWEE-UHFFFAOYSA-N 0.000 description 4
- 238000003384 imaging method Methods 0.000 description 4
- 238000002347 injection Methods 0.000 description 4
- 239000007924 injection Substances 0.000 description 4
- 238000013136 deep learning model Methods 0.000 description 3
- 238000000586 desensitisation Methods 0.000 description 3
- 230000003902 lesion Effects 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 3
- 238000005481 NMR spectroscopy Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000003759 clinical diagnosis Methods 0.000 description 2
- 238000011010 flushing procedure Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 210000005229 liver cell Anatomy 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 206010006322 Breath holding Diseases 0.000 description 1
- 206010019695 Hepatic neoplasm Diseases 0.000 description 1
- 206010067125 Liver injury Diseases 0.000 description 1
- 239000002616 MRI contrast agent Substances 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 230000036770 blood supply Effects 0.000 description 1
- 201000011510 cancer Diseases 0.000 description 1
- 208000019425 cirrhosis of liver Diseases 0.000 description 1
- 238000002591 computed tomography Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000013399 early diagnosis Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 210000002767 hepatic artery Anatomy 0.000 description 1
- 231100000753 hepatic injury Toxicity 0.000 description 1
- 210000002989 hepatic vein Anatomy 0.000 description 1
- 210000003494 hepatocyte Anatomy 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000001208 nuclear magnetic resonance pulse sequence Methods 0.000 description 1
- 239000002504 physiological saline solution Substances 0.000 description 1
- 210000003240 portal vein Anatomy 0.000 description 1
- 229960003910 promethazine Drugs 0.000 description 1
- 238000011158 quantitative evaluation Methods 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 210000003462 vein Anatomy 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30056—Liver; Hepatic
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Epidemiology (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a liver nuclear magnetic artery phase image quality grading method based on raspberry group and deep learning, which comprises the following steps: acquiring a liver nuclear magnetic artery phase image, preprocessing the liver nuclear magnetic artery phase image to obtain a gray image, and obtaining a training sample data set with graded quality; classifying the training sample data set and constructing a convolutional neural network model; thirdly, performing feature visualization operation by using global average pooling operation to obtain a feature visualization heat map; screening a characteristic visualization heat map; inputting the deep abstract features of the feature map into a classifier for secondary training to obtain a Pumei liver nuclear magnetic quality control hierarchical model; and step six, inputting the liver nuclear magnetic artery phase images of the patient to be classified to obtain the grading prediction result of the Pumei liver nuclear magnetic artery phase images.
Description
Technical Field
The invention relates to the field of medical image processing and analysis, in particular to a liver nuclear magnetic artery image quality grading method based on raspberry group and deep learning.
Background
China is a country with high incidence of liver diseases. Among them, liver cancer for short, is a common malignant tumor in Chinese population. According to statistics, about 33 thousands of people die of liver cancer in China every year, and the number of the dead people accounts for nearly 50% of the dead people of the liver cancer all over the world. Realizes early screening, early diagnosis and early treatment on liver cancer patients, and can greatly prolong the life cycle of the liver cancer patients. Various imaging examinations including computer tomography and Magnetic Resonance Imaging (MRI) can monitor the tissue morphology and lesion characteristics of liver diseases noninvasively, and become one of the important means for screening liver cancer at early stage.
Promethazine is a novel MRI contrast agent. About 50% of the promethal contrast agent in blood can be absorbed by liver cells, so that MRI can accurately provide blood supply information of liver lesions and liver cell information in the diagnosis process. The American multicenter phase 3 clinical trial has proved in 2010 that the praise obviously can improve the qualitative diagnosis rate of liver lesions and has good safety. Therefore promet has been viewed by clinicians as a valuable hepatobiliary specific contrast agent. With the wide application of promet in clinical liver disease diagnosis, the defects of the promet are gradually shown. Clinicians find liver MRI images in the arterial phase often show artifacts. The visual appearance of the artifacts varies and causes are different. These artifacts can significantly improve the imaging quality of MRI, and in severe cases can lead to missed diagnosis, misdiagnosis, or even failure to diagnose.
The acquisition of high-quality medical images is the premise of accurate image diagnosis, and the quality control of the liver universal display MRI arterial-stage images can improve the MRI image quality and improve the clinical diagnosis accuracy. The western countries have already implemented the medical image quality control system, which is mainly completed by manual review by medical physicists and engineers in the radiology department. However, the clinical quality control work in China is late, and the related system is still imperfect. Medical physicists or engineers in the radiology department lack clinical diagnosis experience and knowledge and cannot accurately grasp the key points of image quality control.
In recent years, with the rapid development of artificial intelligence technology, a large number of automatic diagnosis or quantitative evaluation of liver diseases based on imaging have emerged, such as automatic segmentation of liver and liver tumor regions based on the full convolution neural network FCN, liver cirrhosis classification and liver cancer classification based on the convolution neural network CNN, liver cancer life prediction and liver injury classification based on imaging omics and machine learning models, and the like. These preliminary studies demonstrate that the approach of mining the digital features of medical images and training artificial intelligence models can be used for the aided analysis of liver diseases.
In summary, it is necessary to fully automatically complete the MRI quality control of the universal liver for diagnosis by using the deep learning model, optimize the existing workflow, improve the MRI quality of the universal liver for diagnosis in clinical practice, practically help clinicians to reduce the working pressure, improve the diagnosis efficiency, and benefit patients.
Disclosure of Invention
The invention provides a liver nuclear magnetic artery phase image quality grading method based on raspberry group and deep learning, which is characterized in that the quality grading method is used for accurately grading the quality of a liver nuclear magnetic artery phase image based on Pumei-showing liver nuclear magnetic artery phase image data marked by a professional radiologist by taking the Pumei-showing liver nuclear magnetic artery phase image data as a training sample and constructing a convolutional neural network model.
The invention also aims to introduce a characteristic dense connection strategy, increase the characteristic input of each layer, distinguish the nuclear magnetic artery stage artifact and improve the accuracy of quality control grading.
A liver nuclear magnetic artery phase image quality grading method based on raspberry group and deep learning comprises the following steps:
acquiring a liver nuclear magnetic artery phase image, preprocessing the liver nuclear magnetic artery phase image to obtain a gray image, and marking grading results of the liver nuclear magnetic artery phase image respectively to obtain a training sample data set with graded quality;
classifying the training sample data set, constructing a convolutional neural network model, and extracting hidden features in the gray level image to obtain a pre-trained convolutional neural network model;
thirdly, performing feature visualization on all convolutional layers in the pre-trained convolutional neural network model by using a gradient weighted class activation mapping method to obtain a feature visualization heat map;
screening the characteristic visualization heat map, selecting a characteristic layer with a highlight capture artifact area, and extracting depth abstract characteristics of the characteristic map;
inputting the deep abstract features of the feature map into a classifier for secondary training to obtain a trained Pumei liver nuclear magnetic quality control hierarchical model;
and step six, building a Pumei liver nuclear magnetic quality control hierarchical model in raspberry group equipment, and inputting the liver nuclear magnetic arterial phase images of the patient to be classified to obtain a hierarchical prediction result of the Pumei liver nuclear magnetic arterial phase images.
Preferably, the preprocessing process of the nuclear magnetic artery phase images of the liver in the first step includes:
firstly, signal normalization is carried out on the collected liver nuclear magnetic artery phase image, and the calculation formula is as follows:
wherein, IiIs the signal value of the ith liver nuclear magnetic artery phase image I'iIs a normalized liver nuclear magnetic artery image signal value,mean value, σ, of all acquired nuclear magnetic imagesISignal standard deviation representing the entire nuclear magnetic image;
and then, carrying out graying processing on the liver nuclear magnetic artery phase image and carrying out pixel point segmentation to obtain a grayscale image.
Preferably, the second step includes:
inputting the gray level image of the liver nuclear magnetic artery phase image into a convolutional neural network model as an input layer vector; the output layer of the network model is a Pumei liver nuclear magnetic artery phase image quality grading label;
the convolutional neural network model comprises a first dense connection module, a second dense connection module and a third dense connection module;
a first transition module is arranged between the first dense connection module and the second dense connection module;
a second transition module is arranged between the second dense connection module and the third dense connection module;
the dense connection modules respectively comprise 6 convolution layers with convolution kernels of 3 x 3, and can be used for carrying out feature extraction on the gray level image of the liver nuclear magnetic artery phase image;
the transition modules each include 1 transition convolutional layer with convolutional kernel size of 1 × 1 and 1 average pooling layer with kernel size of 2 × 2, and can compress and select convolutional layer features.
Preferably, the convolution layer is calculated by the following steps:
sliding a convolution kernel on a gray level image of an input layer, and performing convolution operation on a pixel point (i, j) on the gray level image to obtain an output characteristic diagram, wherein the convolution operation formula is as follows:
wherein x isl(i, j) is a feature of an arbitrary layer l, xl=Hl(x0,x1,…xi…,xl-1),xiBeing a feature of any preamble layer, Hl() For batch normalization operations, consisting of an activation function and a convolution operation of size 3 x 3, xl+1(i, j) is the output characteristic of any layer l, wl+1(i, j) is the weight parameter of layer l +1, blThe deviation value of the first layer;
the dimensions of the output feature map are:
wherein L isl+1Is the output feature size, L, of an arbitrary layer LlThe characteristic dimension of any layer l of any layer is defined, p is a corresponding filling parameter, f is a corresponding convolution kernel size, and s is a corresponding convolution step length.
Preferably, the convolution kernel size is 3 × 3.
Preferably, the operation formula of the pooling layer is as follows:
wherein x isl(i, j) is the characteristic of any layer l, d is the corresponding pooling kernel size, and r is the corresponding pooling step.
Preferably, the feature visualization heat map calculation formula is:
where H is the feature visualization heatmap, relu is the activation function,for the gradient weights of the nth signature to class c, is the mean, y, of n feature maps of dimensions i x jcThe nth feature map scores category c, for the weight of class c to the nth feature map,wherein c is 3.
Preferably, the classifier calculation formula in the step five is as follows:
wherein, ak(X) is the activation function operation at characteristic channel k and X ∈ Ω pixel location, Pk(X) output values at characteristic channel k and X ∈ Ω pixel positions.
Preferably, the prairie liver nuclear magnetic quality control hierarchical model is stored in a raspberry pi device.
The invention has the advantages of
The invention provides a supervised learning mode by utilizing a deep learning model, trains on a large-scale Pumei liver nuclear magnetic data set labeled by a professional radiologist, and integrates human knowledge and experience to realize feature screening, thereby effectively compressing features. The trained liver nuclear magnetic quality control grading model can distinguish nuclear magnetic arterial artifacts with higher specificity, and the accuracy of quality control grading is improved.
Drawings
Fig. 1 is a flowchart of a liver nuclear magnetic artery phase image quality grading method based on raspberry pi and deep learning according to the present invention.
Fig. 2 is a network structure diagram of the deep learning neural network model according to the present invention.
FIG. 3 is a schematic diagram of the present invention showing the classification of the nuclear magnetic artery phase image quality of liver in Pumei 1 minute sample.
FIG. 4 is a schematic diagram of 2-point sample for nuclear magnetic artery phase image quality grading of Pumei liver according to the present invention.
FIG. 5 is a schematic diagram of 3-point sample for nuclear magnetic artery phase image quality grading of Pumei liver according to the present invention.
FIG. 6 is a diagram of a raspberry pi device connection structure according to the present invention.
Detailed Description
The present invention is further described in detail below with reference to the attached drawings so that those skilled in the art can implement the invention by referring to the description text.
As shown in fig. 1, the liver nuclear magnetic artery phase image quality grading method based on raspberry pi and deep learning provided by the present invention includes:
step S110, collecting liver nuclear magnetic artery phase images, respectively marking grading results, preprocessing the liver nuclear magnetic artery phase images to obtain gray level images, and obtaining a training sample data set with graded quality;
the liver nuclear magnetic artery phase image acquisition method comprises the following steps:
patients were trained to hold breath prior to examination. And patients were asked to discontinue water in the morning of the examination day and to fasted for 4 hours prior to examination.
The detection equipment adopts a superconducting ultrahigh field nuclear magnetic scanner, the maximum gradient field intensity is 40mT/m on the X axis and the Y axis, 45mT/m on the Z axis, the gradient field switching rate is 200mT/m/m, and an 8-channel phased array body coil is adopted as a surface coil.
Before and after the contrast agent is injected, T1WI of a rapid three-dimensional spoiled gradient echo pulse sequence is respectively carried out, wherein the specific parameters are that TR/TE is 4.19/4.17ms, the flip angle is 9 degrees, the visual field is 300 multiplied by 400mm, the matrix is 168 multiplied by 320, and the recombination minimum layer thickness is 2.5-3.5 mm.
The Primei contrast agent is injected by an artificial vein high-pressure injection tube, the injection dosage is 0.1mL/kg, the flow rate is 2mL/s, 20mL of physiological saline is used for flushing immediately after the contrast agent is injected, the flushing flow rate is 2mL/s, the liver multi-stage dynamic scanning of the liver after injection is respectively carried out for 16-25 s in the hepatic artery stage, 46-55 s in the portal vein stage, 86-100 s in the hepatic vein stage, 150-180 s in the dynamic late stage and 20min after injection in the hepatocyte stage, and the single breath holding time is 16 +/-1 s.
The desensitization and preprocessing process of the image data comprises the following steps: the acquired liver nuclear magnetic artery phase image data enhanced by the common contrast agent is in a DICOM file, and desensitization operation is performed on text information in the DICOM header file. Personal information such as patient telephone numbers, addresses and the like and hospital information in the medical information are erased so as to guarantee privacy. The age and sex information of the patient is of reference significance for disease diagnosis, and therefore the information is reserved.
Performing signal normalization on the collected image data in all DICOM, wherein the calculation formula is as follows:
wherein, IiIs the signal value of the ith liver nuclear magnetic artery phase image I'iIs a normalized liver nuclear magnetic artery image signal value,mean value, σ, of all acquired nuclear magnetic imagesISignal standard deviation representing the entire nuclear magnetic image;
as shown in fig. 3 to 5, the image data labeling: at least three clinical radiologists were selected to form a panel of experts. The panel is required to cover different levels of experience, which roughly consists of: at least one attending physician with a work experience of 8 years or more; at least one hospitalization physician with a working experience of 5 years and more; at least one physician, with a working experience of 3 years and more. The classification method adopts 3 points recommended by the American radiology Association (ACR), as shown in FIG. 3, wherein 3 points represent images with good quality and suitable for diagnosis; 2 points represent images of general quality but still diagnosable; score 1 represents an image of poor image quality that cannot be used for diagnosis. And (4) carrying out back-to-back quality control grading on the data in the image by the expert panel member, if the opinions are diverged, calling the discussion to determine a grading result, and if the opinions cannot be unified after the discussion, taking the opinion of the doctor with the highest position as the standard.
Image cases subjected to data desensitization, preprocessing and data labeling are arranged in time sequence, the liver nuclear magnetic artery phase image is subjected to graying processing and pixel segmentation to obtain a grayscale image, and the resolution of the grayscale image is preferably 512 x 512.
As shown in fig. 2, in step S120, 20% of data closest to the acquisition time is selected as test set data, and the remaining 80% of data is selected as training set data, a convolutional neural network model is constructed, and hidden features in the gray-scale image are extracted to obtain a pre-trained convolutional neural network model;
the convolutional neural network model introduces a characteristic dense connection strategy on the basis of hierarchical connection of the traditional convolutional neural network, increases the characteristic input of each layer, and aims at the l-th layer, the characteristic x of the l-th layerlCan be expressed as:
xl=Hl(x0,x1,…xi…,xl-1)
wherein x isiFor the characteristics of any preamble layer, the deep learning network is constructed with 24 layers, wherein the deep learning network comprises 3 dense connection modules and 2 transition modules. Each densely connected module contains 6 convolutional layers with a convolution kernel of 3 x 3. Each dense connection module is connected with 1 transition module. The transition module contains 1 convolutional layer with a convolutional kernel size of 1 × 1 and 1 average pooling layer with a kernel size of 2 × 2.
Inputting the gray level image of the liver nuclear magnetic artery phase image into the convolutional neural network model as an input layer vector; the convolution spirit enables an output layer of the network model to be a Pumei liver nuclear magnetic artery phase image quality grading label;
the convolutional neural network model comprises a first dense connection module, a second dense connection module and a third dense connection module; a first transition module is arranged between the first dense connection module and the second dense connection module; a second transition module is arranged between the second dense connection module and the third dense connection module;
the dense connection modules respectively comprise 6 convolution layers with convolution kernels of 3 x 3, and can be used for carrying out feature extraction on the gray level image of the liver nuclear magnetic artery image;
the operation process of the convolutional layer is as follows:
sliding a convolution kernel on the gray level image of the input layer, and performing convolution operation on a pixel point (i, j) on the gray level image to obtain an output characteristic diagram, wherein the convolution operation formula is as follows:
wherein x isl(i, j) is a feature of an arbitrary layer l, xl=Hl(x0,x1,…xi…,xl-1),xiBeing a feature of any preamble layer, Hl() For batch normalization operations, consisting of an activation function and a convolution operation of size 3 x 3, xl+1(i, j) is the output characteristic of any layer l, wl+1(i, j) is the weight parameter of layer l +1, blThe deviation value of the first layer;
the dimensions of the output feature map are:
wherein L isl+1Is the output feature size, L, of an arbitrary layer LlThe characteristic dimension of any layer l of any layer is defined, p is a corresponding filling parameter, f is a corresponding convolution kernel size, and s is a corresponding convolution step length.
The transition modules each include 1 transition convolutional layer with convolutional kernel size of 1 × 1 and 1 average pooling layer with kernel size of 2 × 2, enabling the convolutional layer features to be compressed and selected.
S130, performing feature visualization on all convolutional layers in the pre-trained convolutional neural network model by utilizing global average pooling operation to obtain a feature visualization heat map;
the operation formula of the pooling layer is as follows:
wherein x isl(i, j) is the characteristic of any layer l, d is the corresponding pooling kernel size, and r is the corresponding pooling step.
Preferably, the feature visualization heat map calculation formula is:
where H is the feature visualization heatmap, relu () is the activation function,for the gradient weights of the nth signature to class c, is the mean, y, of n feature maps of dimensions i x jcThe nth feature map scores category c, for the weight of class c to the nth feature map,wherein c is 3.
Step S140, screening the characteristic visualization heat map, selecting a characteristic layer with a highlight capture artifact area, and extracting a depth abstract characteristic of the characteristic map;
s150, inputting the deep abstract features of the feature map into a classifier for secondary training to obtain a trained convolutional neural network model, namely a Pumei liver nuclear magnetic quality control hierarchical model;
the classifier calculation formula in the step five is as follows:
wherein, ak(X) is the activation function operation at characteristic channel k and X ∈ Ω pixel location, Pk(X) represents the output values at the characteristic channel k and X ∈ Ω pixel locations.
Step S160, inputting the liver nuclear magnetic artery phase images of the patient to be classified into a Pumei liver nuclear magnetic quality control hierarchical model to obtain a hierarchical prediction result of the Pumei liver nuclear magnetic artery phase images.
In another embodiment, a raspberry pi 4 is included, and the shutdown deep learning peme liver nuclear magnetic arterial phase image quality grading neural network model of the invention can be transplanted into a hardware device, namely a raspberry pi. And respectively connecting the raspberry pie with a computer module and a display of the nuclear magnetic resonance spectrometer. The method comprises the steps of directly inputting the pulegorian liver nuclear magnetic artery phase images collected and reconstructed by a nuclear magnetic resonance instrument computer module into raspberry group equipment, and then transmitting the quality grading results of the pulegorian liver nuclear magnetic artery phase images judged by a neural network model in the raspberry group to a display screen, so that direct quality control of a data generation end is realized, clinical workflow is optimized, and the working efficiency of doctors and technicians is improved.
The invention provides a supervised learning mode by utilizing a deep learning model, trains on a large-scale Pumei liver nuclear magnetic data set labeled by a professional radiologist, and integrates human knowledge and experience to realize feature screening, thereby effectively compressing features. The trained liver nuclear magnetic quality control grading model can distinguish nuclear magnetic arterial artifacts with higher specificity, and the accuracy of quality control grading is improved.
The invention provides a liver nuclear magnetic artery phase image quality grading method based on raspberry group and deep learning, which takes the image data of the pulegorian liver nuclear magnetic artery phase marked by professional radiologists as training samples to construct a convolutional neural network model, can accurately grade the quality of the pulegorian liver nuclear magnetic artery phase images, introduces a characteristic dense connection strategy, increases the characteristic input of each layer, can distinguish nuclear magnetic artery phase artifacts, and improves the accuracy of quality control grading.
While embodiments of the invention have been described above, it is not limited to the applications set forth in the description and the embodiments, which are fully applicable in various fields of endeavor to which the invention pertains, and further modifications may readily be made by those skilled in the art, it being understood that the invention is not limited to the details shown and described herein without departing from the general concept defined by the appended claims and their equivalents.
Claims (9)
1. A liver nuclear magnetic artery phase image quality grading method based on raspberry group and deep learning is characterized by comprising the following steps:
acquiring a liver nuclear magnetic artery phase image, preprocessing the liver nuclear magnetic artery phase image to obtain a gray image, and marking grading results of the liver nuclear magnetic artery phase image respectively to obtain a training sample data set with graded quality;
classifying the training sample data set, constructing a convolutional neural network model, and extracting hidden features in the gray level image to obtain a pre-trained convolutional neural network model;
thirdly, performing feature visualization on all convolutional layers in the pre-trained convolutional neural network model by using a gradient weighted class activation mapping method to obtain a feature visualization heat map;
screening the characteristic visualization heat map, selecting a characteristic layer with a highlight capture artifact area, and extracting depth abstract characteristics of the characteristic map;
inputting the deep abstract features of the feature map into a classifier for secondary training to obtain a trained Pumei liver nuclear magnetic quality control hierarchical model;
and step six, inputting the liver nuclear magnetic artery phase images of the patient to be classified into a Pumei liver nuclear magnetic quality control hierarchical model to obtain a hierarchical prediction result of the Pumei liver nuclear magnetic artery phase images.
2. The liver nuclear magnetic artery phase image quality grading method based on raspberry pi and deep learning of claim 1, wherein the liver nuclear magnetic artery phase image preprocessing process in the first step comprises:
firstly, signal normalization is carried out on the collected liver nuclear magnetic artery phase image, and the calculation formula is as follows:
wherein, IiIs the signal value of the ith liver nuclear magnetic artery phase image I'iIs a normalized liver nuclear magnetic artery image signal value,mean value, σ, of all acquired nuclear magnetic imagesISignal standard deviation representing the entire nuclear magnetic image;
and then, carrying out graying processing on the liver nuclear magnetic artery phase image and carrying out pixel point segmentation to obtain a grayscale image.
3. The liver nuclear magnetic artery phase image quality grading method based on raspberry pi and deep learning of claim 2, wherein the second step comprises:
inputting the gray level image of the liver nuclear magnetic artery phase image into a convolutional neural network model as an input layer vector; the output layer of the network model is a Pumei liver nuclear magnetic artery phase image quality grading label;
the convolutional neural network model comprises a first dense connection module, a second dense connection module and a third dense connection module;
a first transition module is arranged between the first dense connection module and the second dense connection module;
a second transition module is arranged between the second dense connection module and the third dense connection module;
the dense connection modules respectively comprise 6 convolution layers with convolution kernels of 3 x 3, and can be used for carrying out feature extraction on the gray level image of the liver nuclear magnetic artery phase image;
the transition modules each include 1 transition convolutional layer with convolutional kernel size of 1 × 1 and 1 average pooling layer with kernel size of 2 × 2, and can compress and select convolutional layer features.
4. The liver nuclear magnetic artery phase image quality grading method based on raspberry pi and deep learning of claim 3, wherein the convolution layer is calculated by the following steps:
sliding a convolution kernel on a gray level image of an input layer, and sequentially performing convolution operation on pixel points on the gray level image to obtain an output characteristic diagram, wherein the convolution operation formula is as follows:
wherein x isl(i, j) is a feature of an arbitrary layer l, xl=Hl(x0,x1,…xi…,xl-1),xiBeing a feature of any preamble layer, Hl() For batch normalization operations, consisting of an activation function and a convolution operation of size 3 x 3, xl+1(i, j) is the output characteristic of any layer l, wl+1(i, j) is the weight parameter of layer l +1, blThe deviation value of the first layer; the sizes of the output characteristic diagram are as follows:wherein L isl+1Is the output feature size, L, of an arbitrary layer LlThe characteristic dimension of any layer l of any layer is defined, p is a corresponding filling parameter, f is a corresponding convolution kernel size, and s is a corresponding convolution step length.
5. The method for grading liver nuclear magnetic artery phase images based on raspberry pi and deep learning of claim 4, wherein the convolution kernel size is 3 x 3.
6. The liver nuclear magnetic artery phase image quality grading method based on raspberry pi and deep learning of claim 3, wherein the operation formula of the pooling layer is as follows:
wherein x isl(i, j) is the characteristic of any layer l, d is the corresponding pooling kernel size, and r is the corresponding pooling step.
7. The liver nuclear magnetic artery phase image quality grading method based on raspberry pi and deep learning according to claim 1 or 6, wherein the gradient weighted class activation mapping method in the third step comprises:
8. The liver nuclear magnetic artery phase image quality grading method based on raspberry pi and deep learning of claim 7, wherein the classifier calculation formula in the fifth step is:
wherein, ak(X) is the activation function operation at characteristic channel k and X ∈ Ω pixel location, Pk(X) represents the output values at the characteristic channel k and X ∈ Ω pixel locations.
9. The liver nuclear magnetic arterial phase image quality grading method based on raspberry pi and deep learning of claim 1, wherein the Pomey liver nuclear magnetic quality control grading model is stored in raspberry pi device.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911157464.2A CN110827275B (en) | 2019-11-22 | 2019-11-22 | Liver nuclear magnetic artery image quality grading method based on raspberry pie and deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911157464.2A CN110827275B (en) | 2019-11-22 | 2019-11-22 | Liver nuclear magnetic artery image quality grading method based on raspberry pie and deep learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110827275A true CN110827275A (en) | 2020-02-21 |
CN110827275B CN110827275B (en) | 2023-12-22 |
Family
ID=69558656
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911157464.2A Active CN110827275B (en) | 2019-11-22 | 2019-11-22 | Liver nuclear magnetic artery image quality grading method based on raspberry pie and deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110827275B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111554402A (en) * | 2020-04-24 | 2020-08-18 | 山东省立医院 | Machine learning-based method and system for predicting postoperative recurrence risk of primary liver cancer |
CN112257769A (en) * | 2020-10-19 | 2021-01-22 | 中国科学院自动化研究所 | Multilayer nuclear magnetic image classification method and system based on reinforcement learning type brain reading |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106651830A (en) * | 2016-09-28 | 2017-05-10 | 华南理工大学 | Image quality test method based on parallel convolutional neural network |
CN107506797A (en) * | 2017-08-25 | 2017-12-22 | 电子科技大学 | One kind is based on deep neural network and multi-modal image alzheimer disease sorting technique |
CN109472774A (en) * | 2018-10-11 | 2019-03-15 | 上海中医药大学 | A kind of tongue picture picture quality detection method based on deep learning |
CN109523514A (en) * | 2018-10-18 | 2019-03-26 | 西安电子科技大学 | To the batch imaging quality assessment method of Inverse Synthetic Aperture Radar ISAR |
CN109767429A (en) * | 2018-12-28 | 2019-05-17 | 上海联影智能医疗科技有限公司 | A kind of image screening method and device |
US10304193B1 (en) * | 2018-08-17 | 2019-05-28 | 12 Sigma Technologies | Image segmentation and object detection using fully convolutional neural network |
WO2019133538A2 (en) * | 2017-12-29 | 2019-07-04 | Leica Biosystems Imaging, Inc. | Processing of histology images with a convolutional neural network to identify tumors |
CN110246126A (en) * | 2019-06-14 | 2019-09-17 | 吉林大学第一医院 | A method of extracting terminal bronchi tree from lung CT image |
CN110335259A (en) * | 2019-06-25 | 2019-10-15 | 腾讯科技(深圳)有限公司 | A kind of medical image recognition methods, device and storage medium |
CN110414494A (en) * | 2019-01-25 | 2019-11-05 | 西安电子科技大学 | SAR image classification method with ASPP deconvolution network |
-
2019
- 2019-11-22 CN CN201911157464.2A patent/CN110827275B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106651830A (en) * | 2016-09-28 | 2017-05-10 | 华南理工大学 | Image quality test method based on parallel convolutional neural network |
CN107506797A (en) * | 2017-08-25 | 2017-12-22 | 电子科技大学 | One kind is based on deep neural network and multi-modal image alzheimer disease sorting technique |
WO2019133538A2 (en) * | 2017-12-29 | 2019-07-04 | Leica Biosystems Imaging, Inc. | Processing of histology images with a convolutional neural network to identify tumors |
US10304193B1 (en) * | 2018-08-17 | 2019-05-28 | 12 Sigma Technologies | Image segmentation and object detection using fully convolutional neural network |
CN109472774A (en) * | 2018-10-11 | 2019-03-15 | 上海中医药大学 | A kind of tongue picture picture quality detection method based on deep learning |
CN109523514A (en) * | 2018-10-18 | 2019-03-26 | 西安电子科技大学 | To the batch imaging quality assessment method of Inverse Synthetic Aperture Radar ISAR |
CN109767429A (en) * | 2018-12-28 | 2019-05-17 | 上海联影智能医疗科技有限公司 | A kind of image screening method and device |
CN110414494A (en) * | 2019-01-25 | 2019-11-05 | 西安电子科技大学 | SAR image classification method with ASPP deconvolution network |
CN110246126A (en) * | 2019-06-14 | 2019-09-17 | 吉林大学第一医院 | A method of extracting terminal bronchi tree from lung CT image |
CN110335259A (en) * | 2019-06-25 | 2019-10-15 | 腾讯科技(深圳)有限公司 | A kind of medical image recognition methods, device and storage medium |
Non-Patent Citations (1)
Title |
---|
陈寿宏;柳馨雨;马峻;康怀强;: "深度卷积神经网络胸片肺结节分类识别研究" * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111554402A (en) * | 2020-04-24 | 2020-08-18 | 山东省立医院 | Machine learning-based method and system for predicting postoperative recurrence risk of primary liver cancer |
CN112257769A (en) * | 2020-10-19 | 2021-01-22 | 中国科学院自动化研究所 | Multilayer nuclear magnetic image classification method and system based on reinforcement learning type brain reading |
CN112257769B (en) * | 2020-10-19 | 2022-10-21 | 中国科学院自动化研究所 | Multilayer nuclear magnetic image classification method and system based on reinforcement learning type brain reading |
Also Published As
Publication number | Publication date |
---|---|
CN110827275B (en) | 2023-12-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11101033B2 (en) | Medical image aided diagnosis method and system combining image recognition and report editing | |
CN108573490B (en) | Intelligent film reading system for tumor image data | |
CN106372390B (en) | A kind of self-service healthy cloud service system of prevention lung cancer based on depth convolutional neural networks | |
CN110060774B (en) | Thyroid nodule identification method based on generative confrontation network | |
CN112086197B (en) | Breast nodule detection method and system based on ultrasonic medicine | |
CN112529894B (en) | Thyroid nodule diagnosis method based on deep learning network | |
CN108257135A (en) | The assistant diagnosis system of medical image features is understood based on deep learning method | |
CN109615633A (en) | Crohn disease assistant diagnosis system and method under a kind of colonoscopy based on deep learning | |
CN113781440B (en) | Ultrasonic video focus detection method and device | |
CN114782307A (en) | Enhanced CT image colorectal cancer staging auxiliary diagnosis system based on deep learning | |
CN109411084A (en) | A kind of intestinal tuberculosis assistant diagnosis system and method based on deep learning | |
CN111214255A (en) | Medical ultrasonic image computer-aided diagnosis method | |
CN110827275B (en) | Liver nuclear magnetic artery image quality grading method based on raspberry pie and deep learning | |
CN112508884A (en) | Comprehensive detection device and method for cancerous region | |
CN114398979A (en) | Ultrasonic image thyroid nodule classification method based on feature decoupling | |
CN115019049A (en) | Bone imaging bone lesion segmentation method, system and equipment based on deep neural network | |
CN114241187A (en) | Muscle disease diagnosis system, device and medium based on ultrasonic bimodal images | |
CN109493340A (en) | Esophagus fundus ventricularis varication assistant diagnosis system and method under a kind of gastroscope | |
CN209770401U (en) | Medical image rapid analysis processing system | |
CN112132790B (en) | DAC-GAN model construction method and application thereof in mammary gland MR image | |
CN113796850A (en) | Parathyroid MIBI image analysis system, computer device, and storage medium | |
Jha et al. | Interpretability of Self-Supervised Learning for Breast Cancer Image Analysis | |
CN115760824B (en) | Skin disease auxiliary diagnosis system and method, electronic equipment and storage medium | |
CN112489788B (en) | Multi-mode image analysis method and system for cancer diagnosis | |
CN115762722B (en) | Image review system based on artificial intelligence |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |