CN109886285A - A kind of cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks - Google Patents
A kind of cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks Download PDFInfo
- Publication number
- CN109886285A CN109886285A CN201811629625.9A CN201811629625A CN109886285A CN 109886285 A CN109886285 A CN 109886285A CN 201811629625 A CN201811629625 A CN 201811629625A CN 109886285 A CN109886285 A CN 109886285A
- Authority
- CN
- China
- Prior art keywords
- cholelithiasis
- medical image
- image
- training
- convolutional neural
- 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
Landscapes
- Apparatus For Radiation Diagnosis (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of the cholelithiasis CT medical image method for quickly identifying based on YOLO convolutional neural networks, designed image processing, medical big data, deep learning field.Include: 1) to acquire medical image, constructs training set；2) image in training set is handled, training sample required for generating；3) the medical image recognition training based on deep learning is carried out using training sample, generates the quick identification model of trained medical image；4) new medical image, building verifying collection are acquired；5) image is concentrated to verify model using verifying.The invention avoids in current deep learning the problem of medical image data sets redundancy, and realize the quick identification of medical image, recognition speed is fast.
Description
Technical field
The present invention relates to a kind of cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks belongs to
In artificial intelligence field
Background technique
Cholelithiasis is a kind of common disease of digestive system, and disease is various, and pathogenic factors is intricate, has disease incidence
The features such as height, molten row's stone are difficult.In addition, the type and form of cholelithiasis are varied, the lesion form of part cholelithiasis
Be it is closely similar, this hinders the correct diagnosis and treatment of cholelithiasis significantly.In this case, the hepatology doctor of some youths needs
Prolonged study skillfully can grasp the technical ability for diagnosing cholelithiasis, this brings huge to the clinical diagnosis of hepatology doctor
Big challenge.And for the patient, cholelithiasis cannot cannot such as be controlled in time according to itself illness in online access data
It treats, there is carcinogenic danger.
It is createed safe and efficient auxiliary by artificial intelligence deep learning in combination with cholelithiasis big data and knowledge etc.
The diagnostic method of doctor is helped, and then provides more effective help for the correct diagnosis and treatment of cholelithiasis, this is a kind of based on lightweight
The cholelithiasis CT medical image method for quickly identifying of convolutional neural networks.
Cholelithiasis type and form are varied, so that needing very more cholelithiasis medical treatment during model training
Image data.Common practice is to have initially set up cholelithiasis CT medical image training set, in the method benefit using deep learning
A large amount of training set of images is trained with model, generates trained model, then constructs new image authentication collection to mould
Type is verified, and final recognition result is obtained.But the type of cholelithiasis and form are varied so that its clinical diagnosis
Process is extremely difficult, brings great challenge to the clinical diagnosis of cholelithiasis.When doctor is in the medical image of observation cholelithiasis
When, it needs to take a significant amount of time observation, check, determine the type, form and position of cholelithiasis.According to investigations, existing market
The deep learning model of the upper overwhelming majority can't complete above-mentioned task.
Summary of the invention
The purpose of the present invention is to solve difficulties present in above-mentioned disease in the liver and gallbladder field, provide a kind of based on lightweight
The cholelithiasis CT medical image method for quickly identifying of convolutional neural networks can help the doctor of cholelithiasis clinic to carry out correct
Diagnosis and treatment improve the accuracy rate of cholelithiasis diagnosis.
The present invention is to solve difficulty present in above-mentioned disease in the liver and gallbladder field, and used technical solution is: Yi Zhongji
In the cholelithiasis CT medical image method for quickly identifying of lightweight convolutional neural networks.
A kind of cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks, which is characterized in that
Include:
1) cholelithiasis CT medical image training set is constructed；
2) training set is handled, training sample required for generating；
3) the cholelithiasis CT medical image recognition training based on deep learning is carried out using training sample, generates cholelithiasis CT
The quick identification model of medical image；
4) it constructs new cholelithiasis CT medical image and verifies collection；
5) the trained quick identification model of cholelithiasis CT medical image is verified.
Further, the image in cholelithiasis CT medical image training set is analyzed, include the following steps in appoint
Anticipate one or more:
1) the step of image in the training set being zoomed in or out；
2) spin step is carried out to the image in the training set；
3) translation step is carried out to the image in the training set；
4) the step of radiation transformation being carried out to the image in the training set；
5) the step of image degree of comparing in the training set being enhanced；
6) the step of focal area label being carried out to the image in the training set.
A kind of further cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks, volume
Accumulating neural network structure includes:
1) data preprocessing phase, including zooming in or out, rotate, translate and radiating to cholelithiasis CT medical image
Variation and etc., the pretreatment of complete paired data；
2) it in the local convolution stage, is operated including one-dimensional convolution sum pond etc., utilizes lightweight convolutional neural networks speed
Fast feature, the label of the focal area of contrast enhancement processing and cholelithiasis CT medical image to cholelithiasis CT medical image
Operation；
3) global convolution stage, multidimensional convolution and pondization including lightweight convolutional neural networks operate, to including to mark
The cholelithiasis CT medical image in region carries out feature extraction, then carries out last knowledge in the full articulamentum for being transferred to network last layer
Other places reason.
A kind of further cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks, it is special
Sign is to comprise the following modules:
1) input unit, for inputting cholelithiasis CT medical image data sets；
2) analytical unit generates multiple training samples for handling the cholelithiasis CT medical image；
3) training unit carries out the cholelithiasis CT medical image recognition training based on deep learning using training sample, raw
At the quick identification model of cholelithiasis CT medical image；
4) detection unit, using the cholelithiasis CT medical image verifying collection newly constructed, to trained cholelithiasis CT medical treatment
The quick identification model of image is verified.
Advantages and advantages of the invention are as follows:
The present invention constantly rises for China's cholelithiasis illness rate, professional medical resource provision is insufficient, traditional cholelithiasis doctor
The problems such as image analysis effect is bad is treated, it is fast to provide a kind of cholelithiasis CT medical image based on lightweight convolutional neural networks
Fast recognition methods is realized using technologies such as cholelithiasis CT medical image data and deep learnings to cholelithiasis CT medical treatment figure
The Automatic signature extraction and intelligent diagnostics of picture, it is entire to diagnose process quickly, efficiently, improve the accuracy of cholelithiasis diagnosis.Base
In the cholelithiasis CT medical image method for quickly identifying of deep learning, solve cholelithiasis type and form it is varied so that
Difficult problem in its clinical diagnostic process.Relative to traditional autonomous diagnostic method of doctor, the present invention can be extremely short
In time, the type, form and position of cholelithiasis are determined, and provide diagnostic result.
Detailed description of the invention
Fig. 1 is a kind of cholelithiasis CT medical image based on lightweight convolutional neural networks of embodiment according to the present invention
The flow diagram of method for quickly identifying
Specific embodiment
Further technical solution of the present invention is illustrated below in conjunction with specific embodiment.
This system subject schemes major embodiment intelligent diagnostics, quickly, the high basic thought of accuracy rate.As shown in Figure 1, one
Cholelithiasis CT medical image method for quickly identifying of the kind based on lightweight convolutional neural networks, comprises the following modules:
1) input unit, for inputting cholelithiasis CT medical image data sets；
2) analytical unit generates multiple training samples for handling the cholelithiasis CT medical image；
3) training unit carries out the cholelithiasis CT medical image recognition training based on deep learning using training sample, raw
At the quick identification model of cholelithiasis CT medical image；
4) detection unit, using the cholelithiasis CT medical image verifying collection newly constructed, to trained cholelithiasis CT medical treatment
The quick identification model of image is verified.
A kind of cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks, basic step is such as
Under；
1) cholelithiasis CT medical image training set is constructed；
2) training set is handled, training sample required for generating；
3) the cholelithiasis CT medical image recognition training based on deep learning is carried out using training sample, generates cholelithiasis CT
The quick identification model of medical image；
4) it constructs new cholelithiasis CT medical image and verifies collection；
5) the trained quick identification model of cholelithiasis CT medical image is verified.
A kind of diagnosis process of the cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks is
One is constantly trained the process of study, and the convolutional neural networks structure of cholelithiasis intelligent auxiliary diagnosis algorithm therein includes data
Pretreatment stage, local convolution stage and global convolution stage, the data preprocessing phase includes to cholelithiasis CT medical treatment figure
The data enhancement operations such as rotation, scaling and the translation of picture；The local convolution stage includes the operation such as one-dimensional convolution sum pond, right
The marking operation of the focal area of the contrast enhancement processing and cholelithiasis CT medical image of cholelithiasis CT medical image；It is global
The convolution stage includes that the convolution of multidimensional and pondization operate, and carries out feature to the cholelithiasis CT medical image comprising marked region and mentions
It takes, then carries out last identifying processing in the full articulamentum for being transferred to network last layer.
A kind of cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks, first has to cholelith
Image in sick CT medical image training set is analyzed, include the following steps in any one or it is multiple:
1) the step of image in the training set being zoomed in or out；
2) spin step is carried out to the image in the training set；
3) translation step is carried out to the image in the training set；
4) the step of radiation transformation being carried out to the image in the training set；
5) the step of image degree of comparing in the training set being enhanced；
6) the step of focal area label being carried out to the image in the training set.
Method of the invention can be introduced into major mainstream hospital, and the equipment for combining major mainstream hospital by we, into one
Step improves into a whole set of cholelithiasis CT medical image intelligent diagnostics process, and comprehensive various big hospital is worked throughout the year in the liver of a line
The medical knowledge of many years of gallbladder section domain expert constantly removes the cholelithiasis CT medical data collection for expanding us, mentions for the present invention
The data source that height is constantly updated, optimizes the iterative process of this model training and verifying, not to the parameter of deep learning network model
Disconnected tuning and update, to greatly improve the present invention to the rate of correct diagnosis of cholelithiasis CT medical image.By to magnanimity disease
The study of example simultaneously excavates important information therein and disease inherent law, it is established that a set of perfect cholelithiasis feature conclude and
Diagnosing model, the final artificial intelligence auxiliary diagnosis realized to cholelithiasis.
The above is only preferred embodiment of the invention, not does limitation in any form to the present invention.
It is any to be familiar with this profession although then the present invention is not intended to limit the invention with preferred embodiment is disclosed above
Technical staff, without departing from the scope of the present invention, when can use above-mentioned technology contents be modified or
The embodiment for being modified to equivalent variations, but but without departing from the technical solutions of the present invention, technology according to the present invention is real
Any simple modification, equivalent change and modification that the above implementation content of verifying is done, still fall within the model of technical solution of the present invention
In enclosing.
Claims (4)
1. a kind of cholelithiasis CT medical image method for quickly identifying based on YOLO convolutional neural networks characterized by comprising
1) cholelithiasis CT medical image training set is constructed；
2) training set is handled, training sample required for generating；
3) the cholelithiasis CT medical image recognition training based on deep learning is carried out using training sample, generates cholelithiasis CT medical treatment
The quick identification model of image；
4) it constructs new cholelithiasis CT medical image and verifies collection；
5) the trained quick identification model of cholelithiasis CT medical image is verified.
2. the cholelithiasis CT medical image method for quickly identifying according to claim 1 based on YOLO convolutional neural networks,
It is characterized in that, analyze the image in the cholelithiasis CT medical image training set, further comprise the steps
Any one or it is multiple:
1) the step of image in the training set being zoomed in or out；
2) spin step is carried out to the image in the training set；
3) translation step is carried out to the image in the training set；
4) the step of radiation transformation being carried out to the image in the training set；
5) the step of image degree of comparing in the training set being enhanced；
6) the step of focal area label being carried out to the image in the training set.
3. the cholelithiasis CT medical image method for quickly identifying according to claim 4 based on YOLO convolutional neural networks,
It is characterized in that, convolutional neural networks structure includes:
1) data preprocessing phase, including zooming in or out, rotating, translating and radiating variation to cholelithiasis CT medical image
And etc., the pretreatment of complete paired data；
2) it in the local convolution stage, is operated including one-dimensional convolution sum pond etc., utilizes the fireballing spy of YOLO convolutional neural networks
Point, the marking operation of the focal area of contrast enhancement processing and cholelithiasis CT medical image to cholelithiasis CT medical image；
3) the global convolution stage, convolution and pondization including YOLO convolutional neural networks operate, to the cholelith comprising marked region
Sick CT medical image carries out feature extraction, then carries out last identifying processing in the full articulamentum for being transferred to network last layer.
4. the cholelithiasis CT medical image method for quickly identifying according to claim 1 based on YOLO convolutional neural networks,
It is characterized by comprising with lower module:
1) input module, for inputting cholelithiasis CT medical image data sets；
2) analytic unit generates multiple training samples for handling the cholelithiasis CT medical image；
3) training assembly carries out the cholelithiasis CT medical image recognition training based on deep learning using training sample, generates gallbladder
The quick identification model of stone disease CT medical image；
4) detection components, using the cholelithiasis CT medical image verifying collection newly constructed, to trained cholelithiasis CT medical image
Quick identification model is verified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811629625.9A CN109886285A (en) | 2018-12-29 | 2018-12-29 | A kind of cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811629625.9A CN109886285A (en) | 2018-12-29 | 2018-12-29 | A kind of cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109886285A true CN109886285A (en) | 2019-06-14 |
Family
ID=66925355
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811629625.9A Pending CN109886285A (en) | 2018-12-29 | 2018-12-29 | A kind of cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109886285A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111091559A (en) * | 2019-12-17 | 2020-05-01 | 山东大学齐鲁医院 | Depth learning-based auxiliary diagnosis system for small intestine sub-scope lymphoma |
-
2018
- 2018-12-29 CN CN201811629625.9A patent/CN109886285A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111091559A (en) * | 2019-12-17 | 2020-05-01 | 山东大学齐鲁医院 | Depth learning-based auxiliary diagnosis system for small intestine sub-scope lymphoma |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ren et al. | Interleaved 3D‐CNN s for joint segmentation of small‐volume structures in head and neck CT images | |
| CN109215021A (en) | A kind of cholelithiasis CT medical image method for quickly identifying based on deep learning | |
| CN109712705A (en) | A kind of cholelithiasis intelligent diagnostics APP based on deep learning | |
| AU2017292642B2 (en) | System and method for automatic detection, localization, and semantic segmentation of anatomical objects | |
| CN109872814A (en) | A kind of cholelithiasis intelligent auxiliary diagnosis system based on deep learning | |
| Le et al. | Pancreatic cancer detection in whole slide images using noisy label annotations | |
| RU2017137802A (en) | Method and system for supporting the adoption of medical decisions using mathematical models of patient presentation | |
| WO2023040164A1 (en) | Method and apparatus for training pet/ct-based lung adenocarcinoma and squamous carcinoma diagnosis model | |
| CN110459300A (en) | A kind of lung cancer pathology type diagnostic method based on computer vision and CT images | |
| Huang et al. | Applying deep learning in recognizing the femoral nerve block region on ultrasound images | |
| CN112365980A (en) | Brain tumor multi-target point auxiliary diagnosis and prospective treatment evolution visualization method and system | |
| CN111079901A (en) | Acute stroke lesion segmentation method based on small sample learning | |
| Li et al. | Prototypical few-shot segmentation for cross-institution male pelvic structures with spatial registration | |
| CN109712706A (en) | A kind of observation method and device based on deep learning | |
| Ni et al. | Asymmetry disentanglement network for interpretable acute ischemic stroke infarct segmentation in non-contrast CT scans | |
| Yang et al. | Two-stream regression network for dental implant position prediction | |
| CN109886285A (en) | A kind of cholelithiasis CT medical image method for quickly identifying based on lightweight convolutional neural networks | |
| Tang et al. | Lesion segmentation and RECIST diameter prediction via click-driven attention and dual-path connection | |
| CN102663728B (en) | Dictionary learning-based medical image interactive joint segmentation | |
| Yang et al. | Quantification of muscle, bones, and fat on single slice thigh CT | |
| Ran et al. | Surgical instrument detection algorithm based on improved YOLOv7x | |
| Nugaliyadde et al. | RCNN for region of interest detection in whole slide images | |
| Hu et al. | A location constrained dual-branch network for reliable diagnosis of jaw tumors and cysts | |
| CN109817310A (en) | A kind of cholelithiasis CT medical image data Enhancement Method based on YOLO convolutional neural networks | |
| CN109003659A (en) | Stomach Helicobacter pylori infects pathological diagnosis and supports system and method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190614 |