CN109636819A - A kind of Lung neoplasm growth rate prediction meanss based on deep learning - Google Patents
A kind of Lung neoplasm growth rate prediction meanss based on deep learning Download PDFInfo
- Publication number
- CN109636819A CN109636819A CN201811543927.4A CN201811543927A CN109636819A CN 109636819 A CN109636819 A CN 109636819A CN 201811543927 A CN201811543927 A CN 201811543927A CN 109636819 A CN109636819 A CN 109636819A
- Authority
- CN
- China
- Prior art keywords
- lung neoplasm
- lung
- neoplasm image
- historical data
- neural network
- 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
- 208000020816 lung neoplasm Diseases 0.000 title claims abstract description 130
- 238000013135 deep learning Methods 0.000 title claims abstract description 11
- 238000003062 neural network model Methods 0.000 claims abstract description 33
- 210000004072 lung Anatomy 0.000 claims description 30
- 238000012549 training Methods 0.000 claims description 15
- 238000012360 testing method Methods 0.000 claims description 11
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 230000002708 enhancing effect Effects 0.000 claims description 9
- 238000003708 edge detection Methods 0.000 claims description 7
- 238000000638 solvent extraction Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 6
- 238000005520 cutting process Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 230000006870 function Effects 0.000 description 11
- 238000004590 computer program Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 7
- 238000000034 method Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 3
- 230000007170 pathology Effects 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 206010058467 Lung neoplasm malignant Diseases 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000000151 deposition Methods 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
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 201000005202 lung cancer Diseases 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000505 pernicious effect Effects 0.000 description 1
- 230000002685 pulmonary effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000001225 therapeutic effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- 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
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- 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
- G06T2207/10081—Computed x-ray tomography [CT]
-
- 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
- G06T2207/30061—Lung
- G06T2207/30064—Lung nodule
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Geometry (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The Lung neoplasm growth rate prediction meanss based on deep learning that the invention discloses a kind of, comprising: computing unit;And storage unit, the instruction that storage can be executed by computing unit, instruction make device execute following operation when being executed by computing unit: obtaining the Lung neoplasm image historical data and current data of Different Individual;It is labelled by Lung neoplasm image current data to Lung neoplasm image historical data;Picture pretreatment is carried out to the Lung neoplasm image historical data of tape label;The initial neural network model being pre-created is trained according to pretreated Lung neoplasm image historical data and obtains final neural network model;And further Lung neoplasm image data is handled using final neural network model, to be predicted growth rate of the Lung neoplasm in further Lung neoplasm image data in subsequent preset time period and obtain prediction result.The device of the invention can further increase the antidiastole ability to Lung neoplasm image, make reference for doctor's analysing patient's condition.
Description
Technical field
The present invention relates to field of artificial intelligence, more specifically, particularly relating to a kind of Lung neoplasm based on deep learning
Growth rate prediction meanss.
Background technique
Lung neoplasm is primarily referred to as the circle or similar round tubercle shadow that single-shot or multiple diameter in pulmonary parenchyma are no more than 3cm, no
It can exclude the possibility of the early stage of lung cancer.Therefore, clinician needs accurately to describe Lung neoplasm, and especially those have pernicious possible
Tubercle.Clinician can formulate suitable treatment plan according to tubercle property, continue to observe or clarify a diagnosis and treat.It is multiple rows of
Spiral CT space with higher and density resolution are the prefered methods of current detection and diagnosis Lung neoplasm.However, in lung knot
It saves rate of rise and predicts field, there is no correlation predictive technology still, presently mainly by the professional ability of doctor and experience to every
The Lung neoplasm CT images of a patient are analyzed to prejudge the development trend in Lung neoplasm future.At this stage, to realize to lung knot
The Accurate Prediction of growth rate is saved, the professional ability and experience for relying solely on doctor are practically impossible to completing for task.Moreover,
With increasing for sufferer, CT images data are also more and more, this increases a large amount of work to doctor;In addition, due to patient's
The state of an illness tends to complicate, therefore CT picture hides pathology and also becomes increasingly complex, this causes to analyze and determine that difficulty increases.
The depth learning technology of artificial intelligence field is research hot topic instantly, has been successfully applied to computer view at present
The fields such as feel, speech recognition, natural language processing.However, pre- for solving Lung neoplasm growth rate using depth learning technology
The problem of survey, there are no any trials for this field.
In view of the above-mentioned defects in the prior art, this field urgently needs one kind that can divide in conjunction with current depth learning art
Analysis Lung neoplasm CT images and the scheme that Accurate Prediction is made to the growth rate in Lung neoplasm future.
Summary of the invention
In view of this, the purpose of the embodiment of the present invention is to propose that a kind of Lung neoplasm growth rate based on deep learning is pre-
Device is surveyed, professional ability and empirical analysis Lung neoplasm CT images heavy workload by doctor is able to solve, judges difficulty greatly simultaneously
And the problem of Accurate Prediction can not being made to the growth rate in Lung neoplasm future.
Based on above-mentioned purpose, the one side of the embodiment of the present invention provides a kind of Lung neoplasm growth speed based on deep learning
Rate prediction meanss, comprising:
Computing unit;With
Storage unit, the instruction that storage can be executed by computing unit, instruction execute device when being executed by computing unit with
Lower operation:
Obtain the Lung neoplasm image historical data and current data of Different Individual;
It is labelled by the Lung neoplasm image current data to Lung neoplasm image historical data;
Picture pretreatment is carried out to the Lung neoplasm image historical data of tape label;
The initial neural network model being pre-created is trained according to pretreated Lung neoplasm image historical data and is obtained
Final neural network model;And
Further Lung neoplasm image data is handled using final neural network model, to the further lung knot
Growth rate of the Lung neoplasm in subsequent preset time period in section image data is predicted and obtains prediction result.
In some embodiments, the Lung neoplasm image historical data of Different Individual and the instruction packet of current data are obtained
It includes:
Collect lung's CT images of Different Individual;
Lung neoplasm image is extracted from lung's CT images using edge detection partitioning algorithm;And
Lung neoplasm image is uniformly zoomed to setting pixel to obtain Lung neoplasm image historical data and current data.
In some embodiments, Lung neoplasm image is extracted from lung's CT images using edge detection partitioning algorithm
Instruction includes:
Lung's CT images are filtered to remove noise spot;
Enhancing processing in edge is carried out to filtered lung's CT images by gradient operator;
The marginal information of the enhanced lung's CT images in edge is obtained according to preset threshold;And
It is split according to target area of the marginal information to lung's CT images to extract Lung neoplasm image.
In some embodiments, it is labelled by the Lung neoplasm image current data to Lung neoplasm image historical data
Instruction include:
According to before half a year Lung neoplasm image data and the current Lung neoplasm image data half a year of obtaining over it is each individual
Lung neoplasm actual growth rate, and the Lung neoplasm image that Lung neoplasm actual growth rate is added to corresponding individual as label is gone through
In history data.
In some embodiments, picture pretreatment includes picture enhancing and/or picture normalized.
In some embodiments, picture enhancing includes denoising, overturning, distortion, and/or the cutting of picture.
In some embodiments, the initial mind being pre-created according to the training of pretreated Lung neoplasm image historical data
Include: through network model and the instruction that obtains final neural network model
The initial neural network model being pre-created by the training of a part of Lung neoplasm image historical data;
By the initial neural network model after the test training of another part Lung neoplasm image historical data to obtain test
As a result;And
Final neural network model is determined according to test result.
In some embodiments, a part of Lung neoplasm image historical data accounting is 90%, and another part lung knot
Saving image historical data accounting is 10%.
In some embodiments, label is the Lung neoplasm actual growth rate of half a year each individual in the past,
Final neural network model is configured to the growth rate to the Lung neoplasm of individual to be predicted within subsequent half a year
It is predicted and obtains prediction result.
In some embodiments, final neural network model connects entirely comprising 3 convolutional layers, 3 sample levels and 2
Connect layer.
The present invention has a kind of following advantageous effects: Lung neoplasm based on deep learning provided in an embodiment of the present invention
Growth rate prediction meanss analyze Lung neoplasm image and the life to Lung neoplasm in subsequent a period of time using depth learning technology
Long rate makes Accurate Prediction, can further increase the antidiastole ability to Lung neoplasm image, does for doctor's analysing patient's condition
It refers to out, and reduces the workload of doctor, reduce the difficulty that clinician analyzes and determines Lung neoplasm.Clinician can be with
In conjunction with the prediction result that prediction meanss of the invention export, suitable therapeutic scheme is timely and effectively formulated according to tubercle property.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
Other embodiments are obtained according to these attached drawings.
Fig. 1 is the calculating according to the Lung neoplasm growth rate prediction meanss based on deep learning of one embodiment of the invention
The schematic diagram of storage architecture;
Fig. 2 is that the device of Fig. 1 predicts the schematic flow chart of Lung neoplasm growth rate;
Fig. 3 is that the device of Fig. 1 extracts showing for Lung neoplasm image using edge detection partitioning algorithm from lung's CT images
Meaning property flow chart;And
Fig. 4 is the schematic diagram according to the convolutional neural networks model of one embodiment of the invention.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with specific embodiment, and reference
The embodiment of the present invention is further described in attached drawing.
Based on above-mentioned purpose, the invention proposes a kind of one of the Lung neoplasm growth rate prediction meanss based on deep learning
A embodiment.Shown in fig. 1 is the schematic diagram of the calculating storage architecture of the device.
As shown in fig. 1, should the calculating storage architecture of Lung neoplasm growth rate prediction meanss based on deep learning can be with
Including computing unit 101 and storage unit 102.Computing unit 101 can be made of CPU and 2 GPU, and GPU can be used
1080Ti accelerates trained or reasoning calculating process using its powerful computing capability.Storage unit 102 can be using common
Storage hard disk (for example, local hard drive).Storage unit 102 stores the instruction that can be executed by computing unit 101, which is being counted
Calculate the operating process for making the device execute prediction Lung neoplasm growth rate when unit 101 executes.
Fig. 2 shows be Fig. 1 device prediction Lung neoplasm growth rate schematic flow chart.
As shown in Figure 2, firstly, obtaining the Lung neoplasm image historical data and current data (S201) of different patients,
In, which can be present data, be also possible to the data put in those years after the historical data.At this
In step, which can execute following operation: collect lung's CT images of different patients, which can be each
Patient is in the lung's CT images put in those years;Lung knot is extracted from lung's CT images using edge detection partitioning algorithm
Save image;And Lung neoplasm image is uniformly zoomed to setting pixel (such as 64*64 pixel) to obtain Lung neoplasm image history
Data and current data, here, the purpose for uniformly zooming to setting pixel are to guarantee certain accuracy rate for the ease of processing
Memory space can be saved simultaneously.
Fig. 3 shows that the device of Fig. 1 extracts Lung neoplasm shadow using edge detection partitioning algorithm from lung's CT images
The schematic flow chart of picture.As shown in Figure 3, which is first filtered to remove noise spot lung's CT images;Then
Enhancing processing in edge is carried out to filtered lung's CT images by gradient operator;Next edge is obtained according to preset threshold to increase
The marginal information of lung's CT images after strong;Finally it is split according to target area of the marginal information to lung's CT images to mention
Take out Lung neoplasm image.
Fig. 2 is returned next to, which passes through the Lung neoplasm image after getting Lung neoplasm image historical data
Current data labels (S202) to Lung neoplasm image historical data.The process to label is mainly calculated over a certain section
The Lung neoplasm actual growth rate of each patient in time, and be added to accordingly using the Lung neoplasm actual growth rate as label
In the Lung neoplasm image historical data of patient.For example, it is assumed that Lung neoplasm image historical data is the Lung neoplasm image number before half a year
According in this step, which can execute following operation: according to the Lung neoplasm image data and current Lung neoplasm shadow before half a year
As the Lung neoplasm actual growth rate of the data half a year of obtaining over each patient is (that is, halve year by current Lung neoplasm image data
The Lung neoplasm actual growth rate of the preceding Lung neoplasm image data half a year of obtaining over each patient), and Lung neoplasm is practical raw
Long rate is added in the Lung neoplasm image historical data of corresponding patient as label.
Next, as shown in Figure 2, which carries out picture pretreatment to the Lung neoplasm image historical data of tape label
(S203).In this step, pretreatment mainly including pictures enhancings such as conventional picture denoising, overturning, distortion, cuttings and is returned
The processing such as one change, hdf5 format needed for pretreated data are finally stored as such as caffe.
Then, the initial neural network which is pre-created according to the training of pretreated Lung neoplasm image historical data
Model simultaneously obtains final neural network model (S204).In this step, which can execute following operation: will be a part of
Lung neoplasm image historical data is sent into initial neural network model, and according to a part of Lung neoplasm image historical data training
Initial neural network model;Another part Lung neoplasm image historical data is sent into the initial neural network model after training,
And according to the initial neural network model after the test training of another part Lung neoplasm image historical data to obtain test result;With
And final neural network model is determined according to test result.Specifically, training process is training data by neural network meter
It calculates, result is exported compared with label value, by error back propagation, iteration, until error convergence to lesser value.If test knot
Fruit does not meet forecasting accuracy requirement, then the network structure of the initial neural network model after training is optimized or adjusted
Ginseng obtains final neural network model until model reaches expected forecasting accuracy.
In a preferred embodiment, which is 90%, and another portion
Dividing Lung neoplasm image historical data accounting is 10%.That is, it is assumed that the Lung neoplasm image historical data of 100 patients is had collected,
Wherein the Lung neoplasm image historical data of 90 patients may be used as training data, in addition the Lung neoplasm image history of 10 patients
Data may be used as test data, and after training convergence, prediction error can receive.
Finally, as shown in Figure 2, which handles further Lung neoplasm image number using final neural network model
According to be predicted simultaneously growth rate of the Lung neoplasm in further Lung neoplasm image data in subsequent preset time period
Obtain prediction result (S205).For example, for predicting the growth rate of subsequent half a year, then the label half a year of should be over each disease
The Lung neoplasm actual growth rate of people.The device is using final neural network model to the Lung neoplasm of patient to be predicted subsequent
Growth rate in half a year is predicted and obtains prediction result.
It should be understood that the situation of change in doctor's Primary Reference Lung neoplasm half a year at present comes analysing patient's condition, therefore this
Invention is illustrated with the growth rate for predicting subsequent half a year.However, the length of predetermined period can not influence this here
The protection scope of invention.
Fig. 4 shows the schematic diagram of CNN (convolutional neural networks) network model.As shown in Figure 4, preferred real at one
It applies in example, final neural network model may include 3 convolutional layers, 3 sample levels and 2 full articulamentums, finally by
Euler loss is returned.Wherein, convolutional layer completes the convolution operation to picture by convolution kernel, and main function is the depth for extracting picture
Feature;Sample level completes sampling operation to picture, and main function is to carry out dimensionality reduction, reduces calculation amount.The present invention devises one
Relatively simple network, the simple network can also retain part gross feature information while extracting picture depth information, than
Relatively it is suitble to the scene, secondly calculating speed is fast, can handle a large amount of pictures simultaneously.
The Lung neoplasm growth rate prediction meanss combination current depth learning art of the embodiment of the present invention is analyzed through pathology
The CT sign of confirmation further increases the ability of the antidiastole to CT sign.CT images input need to be only analysed to when use
The device just can be carried out Lung neoplasm segmentation, then export growth rate prediction result.The device can assist doctor to carry out pathology
Analysis, also can provide reference for the other application of medical imaging intellectual analysis.Meanwhile the device can also add newly-increased sample
Strong training, the predictive ability of further lift scheme.
It should be understood that the present invention is not limited by the embodiment of attached drawing, also that is, those skilled in the art is at this
The embodiment of attached drawing can be made under the introduction of invention suitably modified.
It should be noted that those of ordinary skill in the art will appreciate that realizing the whole in above-described embodiment operation or portion
Split flow can instruct related hardware to complete by computer program, and it is computer-readable that the program can be stored in one
It takes in storage medium, the program is when being executed, it may include such as the process of the embodiment of aforesaid operations.Wherein, the storage is situated between
Matter can be magnetic disk, CD, read-only memory (Read-Only Memory, ROM) or random access memory (Random
Access Memory, RAM) etc..The computer program can achieve that corresponding aforementioned operation embodiment is identical or phase
Similar effect.
In addition, disclosed operation is also implemented as the computer program executed by CPU according to embodiments of the present invention, it should
Computer program may be stored in a computer readable storage medium.When the computer program is executed by CPU, the present invention is executed
The above-mentioned function of being limited in the disclosed operation of embodiment.
In addition, above-mentioned steps also can use controller and for storing so that controller realizes above-mentioned steps function
The computer readable storage medium of computer program is realized.
In addition, it should be appreciated that realizing computer readable storage medium used by operation of the invention (for example, depositing
Reservoir) it can be volatile memory or nonvolatile memory, or may include volatile memory and non-volatile deposit
Both reservoirs.As an example and not restrictive, nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory.Volatile memory
It may include random access memory (RAM), which can serve as external cache.As an example rather than limit
Property, RAM can be obtained in a variety of forms, such as synchronous random access memory (DRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double
Data rate SDRAM (DDR SDRAM), enhancing SDRAM (ESDRAM), synchronization link DRAM (SLDRAM) and direct Rambus
RAM(DRRAM).The storage equipment of disclosed aspect is intended to the memory of including but not limited to these and other suitable type.
Those skilled in the art will also understand is that, various illustrative logical blocks, mould in conjunction with described in disclosure herein
Block, circuit and algorithm steps may be implemented as the combination of electronic hardware, computer software or both.It is hard in order to clearly demonstrate
This interchangeability of part and software, with regard to various exemplary components, square, module, circuit and step function to its into
General description is gone.This function is implemented as software and is also implemented as hardware depending on concrete application and application
To the design constraint of whole system.Those skilled in the art can realize described in various ways for every kind of concrete application
Function, but this realization decision should not be interpreted as causing a departure from range disclosed by the embodiments of the present invention.
Various illustrative logical blocks, module and circuit, which can use, in conjunction with described in disclosure herein is designed to
The following component of function described here is executed to realize or execute: general processor, digital signal processor (DSP), dedicated collection
At circuit (ASIC), field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, divide
Any combination of vertical hardware component or these components.General processor can be microprocessor, but alternatively, processing
Device can be any conventional processors, controller, microcontroller or state machine.Processor also may be implemented as calculating equipment
Combination, for example, the combination of DSP and microprocessor, multi-microprocessor, one or more microprocessors combination DSP and/or any
Other this configurations.
It can be directly contained in hardware the step of algorithm in conjunction with described in disclosure herein, by the soft of processor execution
In part module or in combination of the two.Software module may reside within RAM memory, flash memory, ROM memory,
Eprom memory, eeprom memory, register, hard disk, removable disk, any other shape of CD-ROM or known in the art
In the storage medium of formula.Illustrative storage medium is coupled to processor, enables a processor to read from the storage medium
Win the confidence breath or to the storage medium be written information.In an alternative, the storage medium can be integral to the processor
Together.Pocessor and storage media may reside in ASIC.ASIC may reside in user terminal.In an alternative
In, it is resident in the user terminal that pocessor and storage media can be used as discrete assembly.
In one or more exemplary designs, the function can be real in hardware, software, firmware or any combination thereof
It is existing.If realized in software, can be stored in using the function as one or more instruction or code computer-readable
It is transmitted on medium or by computer-readable medium.Computer-readable medium includes computer storage media and communication media,
The communication media includes any medium for helping for computer program to be transmitted to another position from a position.Storage medium
It can be any usable medium that can be accessed by a general purpose or special purpose computer.As an example and not restrictive, the computer
Readable medium may include RAM, ROM, EEPROM, CD-ROM or other optical disc memory apparatus, disk storage equipment or other magnetic
Property storage equipment, or can be used for carry or storage form be instruct or data structure required program code and can
Any other medium accessed by general or specialized computer or general or specialized processor.In addition, any connection is ok
It is properly termed as computer-readable medium.For example, if using coaxial cable, optical fiber cable, twisted pair, digital subscriber line
(DSL) or such as wireless technology of infrared ray, radio and microwave to send software from website, server or other remote sources,
Then above-mentioned coaxial cable, optical fiber cable, twisted pair, DSL or such as wireless technology of infrared ray, radio and microwave are included in
The definition of medium.As used herein, disk and CD include compact disk (CD), laser disk, CD, digital versatile disc
(DVD), floppy disk, Blu-ray disc, wherein disk usually magnetically reproduce data, and CD using laser optics reproduce data.On
The combination for stating content should also be as being included in the range of computer-readable medium.
It is exemplary embodiment disclosed by the invention above, the disclosed sequence of the embodiments of the present invention is just to retouching
It states, does not represent the advantages or disadvantages of the embodiments.It should be noted that the discussion of any of the above embodiment is exemplary only, it is not intended that
Imply that range disclosed by the embodiments of the present invention (including claim) is limited to these examples, what is limited without departing substantially from claim
Under the premise of range, it may be many modifications and modify.According to the function of the claim of open embodiment described herein,
Step and/or movement are not required to the execution of any particular order.In addition, although element disclosed by the embodiments of the present invention can be with individual
Form description requires, but is unless explicitly limited odd number, it is understood that is multiple.
Claims (10)
1. a kind of Lung neoplasm growth rate prediction meanss based on deep learning characterized by comprising
Computing unit;With
Storage unit, stores the instruction that can be executed by the computing unit, and described instruction makes institute when being executed by the computing unit
It states device and executes following operation:
Obtain the Lung neoplasm image historical data and current data of Different Individual;
It is labelled by the Lung neoplasm image current data to the Lung neoplasm image historical data;
Picture pretreatment is carried out to the Lung neoplasm image historical data of tape label;
The initial neural network model being pre-created is trained according to the pretreated Lung neoplasm image historical data and is obtained
Final neural network model;And
Further Lung neoplasm image data is handled using the final neural network model, to the further lung knot
Growth rate of the Lung neoplasm in subsequent preset time period in section image data is predicted and obtains prediction result.
2. the apparatus according to claim 1, which is characterized in that obtain the Lung neoplasm image historical data of Different Individual and work as
The instruction of preceding data includes:
Collect lung's CT images of Different Individual;
Lung neoplasm image is extracted from lung's CT images using edge detection partitioning algorithm;And
The Lung neoplasm image is uniformly zoomed to setting pixel to obtain the Lung neoplasm image historical data and current data.
3. the apparatus of claim 2, which is characterized in that utilize edge detection partitioning algorithm from lung's CT images
In extract the instruction of Lung neoplasm image and include:
Lung's CT images are filtered to remove noise spot;
Enhancing processing in edge is carried out to filtered lung's CT images by gradient operator;
The marginal information of the enhanced lung's CT images in edge is obtained according to preset threshold;And
It is split according to target area of the marginal information to lung's CT images to extract the Lung neoplasm image.
4. the apparatus according to claim 1, which is characterized in that give the lung knot by the Lung neoplasm image current data
Saving the instruction that image historical data labels includes:
According to before the half a year Lung neoplasm image data and the current Lung neoplasm image data half a year of obtaining over per each and every one
The Lung neoplasm actual growth rate of body, and described in using the Lung neoplasm actual growth rate as label being added to corresponding individual
In Lung neoplasm image historical data.
5. the apparatus according to claim 1, which is characterized in that the picture pretreatment includes picture enhancing and/or picture
Normalized.
6. device according to claim 5, which is characterized in that the picture enhancing includes denoising, overturning, the torsion of picture
Bent, and/or cutting.
7. the apparatus according to claim 1, which is characterized in that according to the pretreated Lung neoplasm image historical data
It trains the initial neural network model being pre-created and the instruction for obtaining final neural network model includes:
Pass through the initial neural network model being pre-created described in the training of a part of Lung neoplasm image historical data;
By the initial neural network model after the test training of another part Lung neoplasm image historical data to obtain test
As a result;And
The final neural network model is determined according to the test result.
8. device according to claim 7, which is characterized in that a part of Lung neoplasm image historical data accounting is
90%, and another part Lung neoplasm image historical data accounting is 10%.
9. the apparatus according to claim 1, which is characterized in that the label is that the Lung neoplasm of half a year each individual in the past is real
Border growth rate,
The final neural network model is configured to the growth rate to the Lung neoplasm of individual to be predicted within subsequent half a year
It is predicted and obtains prediction result.
10. the apparatus according to claim 1, which is characterized in that the final neural network model includes 3 convolution
Layer, 3 sample levels and 2 full articulamentums.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811543927.4A CN109636819A (en) | 2018-12-17 | 2018-12-17 | A kind of Lung neoplasm growth rate prediction meanss based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811543927.4A CN109636819A (en) | 2018-12-17 | 2018-12-17 | A kind of Lung neoplasm growth rate prediction meanss based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109636819A true CN109636819A (en) | 2019-04-16 |
Family
ID=66074757
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811543927.4A Pending CN109636819A (en) | 2018-12-17 | 2018-12-17 | A kind of Lung neoplasm growth rate prediction meanss based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109636819A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110889836A (en) * | 2019-11-22 | 2020-03-17 | 中国人民解放军第四军医大学 | Image data analysis method and device, terminal equipment and storage medium |
CN111353973A (en) * | 2020-02-11 | 2020-06-30 | 中国医学科学院肿瘤医院 | Lymph node medical image-based identification method and lymph node medical image-based identification device |
CN111383767A (en) * | 2020-03-05 | 2020-07-07 | 上海市肺科医院(上海市职业病防治院) | System and method for evaluating progress of small nodules in lung based on deep learning |
CN113688975A (en) * | 2021-08-24 | 2021-11-23 | 北京市商汤科技开发有限公司 | Neural network training method and device, electronic equipment and storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150379709A1 (en) * | 2013-02-14 | 2015-12-31 | The Research Foundation For The State University Of New York | Method for adaptive computer-aided detection of pulmonary nodules in thoracic computed tomography images using hierarchical vector quantization and apparatus for same |
CN107665736A (en) * | 2017-09-30 | 2018-02-06 | 百度在线网络技术(北京)有限公司 | Method and apparatus for generating information |
CN108629803A (en) * | 2018-04-17 | 2018-10-09 | 杭州依图医疗技术有限公司 | A kind of determination method and device of tubercle doubling time |
CN108648192A (en) * | 2018-05-17 | 2018-10-12 | 杭州依图医疗技术有限公司 | A kind of method and device of detection tubercle |
CN108898588A (en) * | 2018-06-22 | 2018-11-27 | 中山仰视科技有限公司 | Therapeutic effect appraisal procedure based on time series, electronic equipment |
-
2018
- 2018-12-17 CN CN201811543927.4A patent/CN109636819A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150379709A1 (en) * | 2013-02-14 | 2015-12-31 | The Research Foundation For The State University Of New York | Method for adaptive computer-aided detection of pulmonary nodules in thoracic computed tomography images using hierarchical vector quantization and apparatus for same |
CN107665736A (en) * | 2017-09-30 | 2018-02-06 | 百度在线网络技术(北京)有限公司 | Method and apparatus for generating information |
CN108629803A (en) * | 2018-04-17 | 2018-10-09 | 杭州依图医疗技术有限公司 | A kind of determination method and device of tubercle doubling time |
CN108648192A (en) * | 2018-05-17 | 2018-10-12 | 杭州依图医疗技术有限公司 | A kind of method and device of detection tubercle |
CN108898588A (en) * | 2018-06-22 | 2018-11-27 | 中山仰视科技有限公司 | Therapeutic effect appraisal procedure based on time series, electronic equipment |
Non-Patent Citations (2)
Title |
---|
郭宝龙 等: "《数字图像处理***工程导论》", 31 July 2012 * |
闵秋莎 等: "《医学图像压缩算法与应用研究》", 31 August 2018 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110889836A (en) * | 2019-11-22 | 2020-03-17 | 中国人民解放军第四军医大学 | Image data analysis method and device, terminal equipment and storage medium |
CN111353973A (en) * | 2020-02-11 | 2020-06-30 | 中国医学科学院肿瘤医院 | Lymph node medical image-based identification method and lymph node medical image-based identification device |
CN111383767A (en) * | 2020-03-05 | 2020-07-07 | 上海市肺科医院(上海市职业病防治院) | System and method for evaluating progress of small nodules in lung based on deep learning |
CN113688975A (en) * | 2021-08-24 | 2021-11-23 | 北京市商汤科技开发有限公司 | Neural network training method and device, electronic equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109636819A (en) | A kind of Lung neoplasm growth rate prediction meanss based on deep learning | |
Liu et al. | A cascaded deep convolutional neural network for joint segmentation and genotype prediction of brainstem gliomas | |
CN110475505A (en) | Utilize the automatic segmentation of full convolutional network | |
You et al. | Bootstrapping semi-supervised medical image segmentation with anatomical-aware contrastive distillation | |
Liu et al. | Counting people by estimating people flows | |
EP3975117A1 (en) | Image segmentation method and apparatus, and training method and apparatus for image segmentation model | |
Zhao et al. | Versatile framework for medical image processing and analysis with application to automatic bone age assessment | |
Chen et al. | Combining edge guidance and feature pyramid for medical image segmentation | |
CN114782398A (en) | Training method and training system for learning network for medical image analysis | |
Li et al. | Hierarchical deep network with uncertainty-aware semi-supervised learning for vessel segmentation | |
KR20220129521A (en) | Training method, device, program for predictive model that can predict cancer prognosis by analyzing tumor image | |
Tummala et al. | Liver tumor segmentation from computed tomography images using multiscale residual dilated encoder‐decoder network | |
CN114612662A (en) | Polyp image segmentation method based on boundary guidance | |
CN115190999A (en) | Classifying data outside of a distribution using contrast loss | |
Pal et al. | A fully connected reproducible SE-UResNet for multiorgan chest radiographs segmentation | |
Droste et al. | Towards capturing sonographic experience: cognition-inspired ultrasound video saliency prediction | |
WO2024093099A1 (en) | Thyroid ultrasound image processing method and apparatus, medium and electronic device | |
Ragnarsdottir et al. | Interpretable prediction of pulmonary hypertension in newborns using echocardiograms | |
CN111445456A (en) | Classification model, network model training method and device, and identification method and device | |
CN110189309A (en) | A kind of image processing method and device, electronic equipment and storage medium | |
US20220319707A1 (en) | System, Method and Computer Readable Medium for Video-Based Facial Weakness Analysis for Detecting Neurological Deficits | |
CN112862785B (en) | CTA image data identification method, device and storage medium | |
CN115661193A (en) | Ultrasonic motion tracking method and system based on time context and information bottleneck | |
US20210192732A1 (en) | Deep learning for optical coherence tomography segmentation | |
CN114881941A (en) | Spine health change prediction method and system based on big data |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190416 |