CN1934589A - Systems and methods providing automated decision support for medical imaging - Google Patents
Systems and methods providing automated decision support for medical imaging Download PDFInfo
- Publication number
- CN1934589A CN1934589A CNA2005800093058A CN200580009305A CN1934589A CN 1934589 A CN1934589 A CN 1934589A CN A2005800093058 A CNA2005800093058 A CN A2005800093058A CN 200580009305 A CN200580009305 A CN 200580009305A CN 1934589 A CN1934589 A CN 1934589A
- Authority
- CN
- China
- Prior art keywords
- view
- quality
- characteristic
- image
- extracted
- 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
Images
Landscapes
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
Systems and methods are provided for processing a medical image to automatically identify the anatomy and view (or pose) from the medical image and automatically assess the diagnostic quality of the medical image. In one aspect a method for automated decision support for medical imaging includes obtaining image data, extracting feature data from the image data, and automatically performing anatomy identification, view identification and/or determining a diagnostic quality of the image data, using the extracted feature data.
Description
The cross reference of related application
The sequence number that the application requires to submit on March 23rd, 2004 is No.60/555,620 U.S. Provisional Application No., and this application is incorporated into this fully with for referencial use.
Technical field
The system and method that provides automatic decision to support for medical imaging is provided.More specifically, the present invention relates to be used to handle medical image automatically to discern anatomy and view (or attitude) from medical image and automatically to estimate the system and method for the quality of diagnosis of medical image.
Background technology
In the medical imaging field, various imaging patterns and system are used to generate the medical image of individual anatomical structure to be used for examination and assessment medical conditions.These imaging systems for example comprise CT (computer tomography) imaging, MRI (magnetic resonance imaging), NM (nuclear-magnetism) resonance image-forming, x-ray system, US (ultrasonic) system, PET (PET (positron emission tomography)) system or the like.For in these patterns each, the specific part of human body is aimed to be used for imaging, and this can carry out in various manners.Utilizing ultrasonicly, is target from the sound wave of transducer with the specific part (for example heart) of health.In MRI, gradient coil is used to the part of " selection " health, writes down nuclear resounce in this part of health.This part of the health that imaging pattern aimed at is usually corresponding to the interesting zone of seeking and visiting of physician.Each imaging pattern can provide the distinct advantages that is better than other imaging pattern, to be used for disease, medical conditions or the anatomical abnormalities of examination and some type of assessment, for example comprise carcinous Microcalcification or lump and various other damage or unusual of cardiomyopathy, polyp of colon, aneurysm, lung tubercle, heart or arterial tissue's calcification, breast tissue.
Typically, physician, clinician, radiologist etc. rebuild artificially check and assessment from the image data set of being gathered medical image (X-ray film, printed matter, photo or the like) is to recognize interested characteristic feature and detection, diagnosis or to discern potential medical conditions in other mode.For example, the CT view data of being gathered during CT examination can be used to produce one group of 2D medical image (X-ray film), can observe described 2D medical image with identification for example potential abnormal anatomical structures or damage.Yet, depending on technology and the knowledge of checking physician, clinician, radiologist etc., the manual evaluation of medical image may cause the mistaken diagnosis medical conditions because of simple personal error.In addition, when the medical image of being gathered has low quality of diagnosis, even there is the check person of high professional qualification may also be difficult to assess effectively such medical image and the potential medical conditions of identification.
Summary of the invention
Usually, one exemplary embodiment of the present invention comprises and is used to handle medical image automatically to discern anatomy and view (or pose) from medical image and automatically to estimate the system and method for the quality of diagnosis of medical image.For example, in an one exemplary embodiment, the method that is used for the automatic decision support of medical imaging comprises obtains view data, extract characteristic and utilize the characteristic that is extracted automatically to carry out anatomy identification, view identification and/or determine the quality of diagnosis of view data from described view data.
In another one exemplary embodiment of the present invention, utilize the correlation classifier of handling the characteristic that is extracted to carry out automatic anatomy identification, view identification and/or image quality evaluation.Can use machine learning method, realize sorter based on the method for model or machine learning with based on any combination of the method for model.
In another one exemplary embodiment of the present invention, can come case like the recognition category by the database that utilizes the characteristic inquiry known cases extracted and use the information relevant to carry out separately function, thereby carry out automatic anatomy identification, view is discerned and/or image quality evaluation with the similar case that is identified.The training data that is extracted from the database of known cases can be used to training classifier to be used to carry out such function.
In another one exemplary embodiment of the present invention, come template like the recognition category and use the information relevant to carry out separately function by the database that utilizes the template that the characteristic inquiry extracted derived from the information of known cases, thereby carry out automatic anatomy identification, view is discerned and/or image quality evaluation with the similar template that is identified.
In another one exemplary embodiment of the present invention, during image acquisition, carry out feature extraction, anatomy identification, view identification and image quality evaluation automatically in real time, wherein during image acquisition, the result of image quality evaluation is offered the user in real time.For imaging pattern (such as the ultrasonic imaging that is used for cardiac imaging (for example 2D ultrasonic cardiography)), the time of ultrasonic inspection doctor images acquired during stress stage is very limited.By the real-time quality assessment of image acquisition is provided, the ultrasonic inspection doctor can determine whether the image of being gathered has enough quality of diagnosis, considers the variation in image acquisition thus in case of necessity.
These and other exemplary embodiment of the present invention, feature and advantage will be described or will become apparent from the following detailed description of one exemplary embodiment, and described detailed description should be understood in conjunction with the accompanying drawings.
Description of drawings
Fig. 1 provides the block diagram of the system that automatic decision supports according to one exemplary embodiment of the present invention for medical imaging.
Fig. 2 is the process flow diagram of explanation according to the method for the automatic decision support that is used for medical imaging of one exemplary embodiment of the present invention.
Fig. 3 is the process flow diagram that the method for supporting according to the automatic decision that utilizes the database query method to carry out medical imaging of one exemplary embodiment of the present invention is described.
To be explanation carry out the process flow diagram of the method that the automatic decision of medical imaging supports according to the utilization of one exemplary embodiment of the present invention to Fig. 4 based on the method for template.
Fig. 5 is explanation according to the classify process flow diagram of the method that the automatic decision of carrying out medical imaging supports of the utilization of one exemplary embodiment of the present invention.
Embodiment
Fig. 1 shows and provides the high level block diagram of the system (100) that automatic decision supports according to one exemplary embodiment of the present invention for medical imaging.Usually, example system (100) comprises data processing module (101), the whole bag of tricks that this data processing module is carried out the medical image (10) (for example ultrasound image data, MRI data, nuclear medicine data or the like) that is used for analyzing one or more imaging patterns to be automatically extracting and to handle from medical image for information about, thereby provides (multiple) different decision support function for the evaluate medical image.In this one exemplary embodiment, data processing module (101) comprises automated characterization analysis module (102), anatomy identification module (103), view identification module (104) and image quality evaluation module (105).
Usually, characteristics analysis module (102) realizes such method, and described method is used for automatically extracting the features/parameters of one or more types and to be applicable to that the mode of being handled by decision support module (103,104 and/or 105) makes up the features/parameters that is extracted from the input medical image.System (100) can processes digital image data (10), and the form of described Digital Image Data is raw image data, 2D data reconstruction (for example axial slices) or 3D data reconstruction (volumetric image data or many planar reformat), 4D data reconstruction or other image model/form.Should be understood that (multiple) type of the automatic decision support method that will be supported according to imaging pattern and/or by CAD system (100) by the performed method of feature extraction module (102) and the anatomical structure of being considered and difference.
Anatomy identification module (102) is carried out and to be used to utilize the features/parameters that the extracted anatomical object of recognition image data centralization (chambers of the heart, kidney or the like) and identify the method for mark (a plurality of) image with suitable anatomy automatically.In another one exemplary embodiment, anatomy identification module (102) execution is used for (at each anatomy/view ID mark) definite anatomy/view of being discerned and is estimated by the degree of confidence of correct labeling or likelihood.The result who is used for the anatomy identification of medical image can or provide other application of automatic diagnosis, treatment plan etc. to use by other automated process (for example view identification and quality evaluating method).
View identification module (103) is carried out and is used to utilize the features/parameters that is extracted automatically to discern the method for the view of the image of being gathered.In other words, view identification module (104) is carried out and to be used for that pose is estimated and to comprise the method which view of anatomy comes the mark medical image about medical image.For example, for the cardiac ultrasonic imaging, U.S. ultrasonic cardiography association (ASE) suggestion uses the standard ultrasound view (apex of the heart two chamber figure (A2C), apical four-chamber figure (A4C), apex of the heart major axis figure (ALAX), parasternal major axis figure (PLAX), parasternal minor axis figure (PSAX)) of B pattern to obtain enough cardiac image datas.The ultrasonoscopy of heart can obtain from different angles, but need distinguish by the position of the heart of imaging (view) can discern important cardiac structure the efficient analysis of cardiac ultrasound images.According to one exemplary embodiment of the present invention, view identification module (103) is carried out a method of discerning unknown cardiac image that is used for as normal view.In addition, view identification module (103) can be carried out and be used for the method that (at each view mark) determines that the view discerned is estimated by the degree of confidence of correct labeling or likelihood.
Quality assessment module (105) is carried out the level of the quality of diagnosis be used for utilizing the features/parameters that is extracted to estimate the image data set of being gathered and is determined whether to occur in image acquisition process the method for error.In another one exemplary embodiment of the present invention, the result of anatomy and/or view identification can be used to quality assessment.In addition, can manner of execution, be used for during image acquisition providing real-time feedback, thereby consider the variation in the image acquisition about the quality of diagnosis of the image gathered.In addition, can manner of execution, be used to determine that quality measure in predetermined span is to provide the indication as the quality level of the image of being gathered based on certain specified criteria.
System (100) further comprises database (106), template database (107) and the categorizing system (108) of the medical image of previous diagnosis the/institute's mark, and it can be individually or is used to carry out their functions separately by in the different automatic decision support module (102-105) of data handling system (101) one or more in the mode of combination.For example, in an one exemplary embodiment, disparate modules (103), (104) and (105) are carried out the database method for inquiring and are searched for by the case of similar mark in database (106) to use the characteristic that is extracted.Database (106) can comprise the medical image that a plurality of being labeled/quilt is diagnosed that is used for various clinical fields, wherein utilizes the multi-dimensional indexing scheme that described medical image is indexed based on features relevant/parameter.Under these circumstances, concentrate the features/parameters that is extracted to compare according to the characteristic of the known cases some module or criterion and the database (106) from the view data of being considered, discern the quality of the image that is extracted to discern specific anatomy or view or help.
In another one exemplary embodiment, different module (103), (104) and (105) can be carried out based on the method for template to use the characteristic that extracted template like the search class in template database (107).Especially, can use the different template of information architecture that is obtained from case database (106).For example, can utilize statistical technique to handle, to derive the characteristic of the template representative on the relevant case collection at the characteristic on a plurality of known cases of given sign and view.In this case, from the view data considered concentrate the features/parameters that extracted can according to some module or criterion with at the characteristic of the template the database (107) relatively, to discern specific anatomy or view or to help to discern the quality of the image that is extracted.
In another one exemplary embodiment, sorting technique can be carried out in different module (103), (104) and (105), and the characteristic that described sorting technique is utilized sort module (108) to handle to be extracted is to classify to the image data set of being considered.In the one exemplary embodiment of Fig. 1, sort module (108) comprises that study engine (109) and knowledge base (110) are to realize principle (machine) learning classification system.Study engine (109) comprises and is used for utilizing database (106) training data of being learnt from the case of before being diagnosed/being labeled to train/set up the method for one or more sorters.Sorter is by being used to carry out its different decision support module (102-105) realization of function separately.
The different result that module produced by data processing module (101) can for good and all be stored in the storage vault (112) that is associated with corresponding image data set.Result can comprise be used for mark overlapping, cut apart, metamessage that color or brightness change or the like, described metamessage can be reproduced as the covering to the associated picture data.
System (100) comprises also that in addition the reconstruction of image and visualization system (111) are with the Digital Image Data (10) of handling the image data set (or its part) gathered and produce and show 2D and/or 3D rendering on computer monitor.More specifically, imaging system (111) 3D/2D that can provide view data (10) reproduces and is visual and have the Any Application of carrying out on the universal or special computer workstation of monitor.In addition, imaging system (111) comprises for example GUI (graphic user interface), and described GUI allows the user by 3D rendering or a plurality of 2D navigate.
The data handling system (101) and the reconstruction of image and visualization system (111) may be implemented as the single application program of carrying out in computing system (for example workstation).Selectively, system (101) and (111) can be the autonomous devices that is distributed on the computer network, and wherein known communication protocol (for example DICOM, PACS or the like) is used to communicate by letter between system and pass through the Network Transmission view data.
Should be understood that the exemplary method that is used for identification of automatic anatomy and view and image quality evaluation is to provide effective ways auxiliary substantially and decision support in medical imaging collection and assessment.In fact, when gathering medical image, importantly use correct anatomy and view marking image correctly, make the physician can carry out correct diagnosis.At present, carry out mark by the technician of acquisition scans or by the physician artificially.Utilize described exemplary labeling method here, system automatically discerns by the anatomy of imaging and view, and this provides various advantages.For example, anatomy identification is manually carried out the workflow that mark has improved the physician by getting rid of automatically.In addition, anatomy identification has promoted other robot brain of automated quality control and assisted diagnosis, treatment plan or other application to use automatically.
In addition, can use (for example 2D ultrasonic cardiography and particularly stress-echo) for medical imaging according to automatic view recognition methods of the present invention provides significant workflow to strengthen.In stress-echo, the time of ultrasonic inspection doctor images acquired from four different views very limited (is about 90 seconds for exercise stress).In order to save time, the ultrasonic inspection doctor usually only carries out record at 90 seconds pith, sets about view is carried out mark then after being embodied as picture.This is a heavy process, can be improved by automatically discerning view.
The result of anatomy identification and/or view identification can one exemplary embodiment according to the present invention be used to carry out automated graphics quality assessment process.Quality check will be estimated the quality of the image of being gathered, and how also estimate whether have any error aspect the images acquired.For example, in the 2D ultrasonic cardiography, be difficult to correctly make apex of the heart imaging from apical window.Typically, image is dwindled by scenography at the apex of the heart under the situation of drafting to be gathered, and that is to say that transducer is changeed an angle, makes the apex of the heart look in fact thicker than it, and this gives impression of people is thicker at the cardiac muscle at apex of the heart place.Yet other clue in the image has hinted that the apex of the heart is dwindled drafting by scenography really.
For example, distance that can be by measuring on ultrasonoscopy from the bottom of left ventricle the apex of the heart and come to discern to dwindling by scenography to draw at the thickness of the cardiac muscle of the apex of the heart.This distance can with the distance of being gathered in formerly the inspection relatively, perhaps the representative value that is associated with the heart of patient's identical size is relatively estimated to determine to dwindle by scenography exist (if any) of drafting then.In addition, the thickness of the cardiac muscle at apex of the heart place be can be evaluated to discern potential another module of dwindling drafting by scenography.If the apical myocardium obviously remainder than heart is thick, then the conclusion that can draw is that view may dwindle drafting by scenography.Other method may be to have one group to dwindle the image of drafting and another group by scenography and do not exist by scenography and dwindle the image that the quilt of drafting is correctly gathered, and they are stored in the database.Using similar module, can search database be more to be similar to by scenography to dwindle the image sets of drafting or more be similar to the image sets of correctly being gathered with definite present image.
Another problem may be the identification of the correction of motion artefacts among the MRI.Because MRI is in " k space " with gathered in the real space, so the motion during image acquisition can cause peculiar result.By the quality of analysis image, can discern correction of motion artefacts.
Except the searching problem, can carry out the automated graphics quality assessment so that the general feedback to the quality of diagnosis of image to be provided.For example, utilize 2D ultrasonic cardiography, particularly stress-echo, the time of ultrasonic inspection doctor images acquired during stress stage is very limited.Acquisition of diagnostic quality image on a plurality of views as far as possible apace importantly for the ultrasonic inspection doctor.Because the time pressure of stress-echo does not repeatedly obtain diagnostic quality images, image is useless.By quality check is provided, the ultrasonic inspection doctor can guarantee the quality of diagnosis of the image gathered.The advantage of doing such quality check is the operator that can in real time feedback be provided back to imaging device, thus the variation in considering to gather.This is very important in medical situation.If feedback is provided apace, sent back home or can be had an opportunity until then to patient's imaging again by being sent the patient so.In addition, also may have other advantage.For example, in stress-echo, Another Application may be automatically to select the image of E.B.B. to check for the cardiologist once more.Usually in stress-echo, the ultrasonic inspection doctor gathers up to four (more sometimes) data rings for each view, and wherein each ring is represented the cardiac cycle, perhaps the systole part of cardiac cycle at least.Typically, ultrasonic inspection doctor or cardiologist select to provide and see from the position of diagnosis and to be the ring of best image and to use them.By quality check is provided, this can automatically carry out.
With reference now to Fig. 2,, the method that provides automatic decision to support for medical imaging according to one exemplary embodiment of the present invention is provided process flow diagram.Illustrative purposes for example, the exemplary method that is used for the automatic decision support will be described with reference to the example system of figure 1.At first, physician, clinician, radiologist or the like will obtain medical images data sets, and this medical images data sets comprises one or more medical images (step 200) of the area-of-interest of subject patient.Can utilize medical image system to obtain image data set, this medical image system is used for gathering in real time and handling raw image data (the original CT data (radon data) of for example being gathered, or the raw data of using other imaging pattern to gather) during CT scan.Selectively, can previous collection obtain image data set by visiting with image data set permanent storage.Digital Image Data (10) can comprise one or more 2D section or three-dimensional volumetric images, and it is rebuilt and for good and all be stored from raw image data.As mentioned above, exemplary CAD process can be supported one or more imaging patterns (for example MRI, PET or the like).View data can be 2D (for example X ray Mammography image), 3D (for example CT, MRI, PET), 4D (dynamically a plurality of views of 3D MRI, the beating heart gathered with the 3D ultrasonic probe) or the like.
Then, with the image data processing collection, to determine from image data set or to extract relevant characteristic (step 201) in other mode, described relevant characteristic is used to carry out one or more decision support functions, for example anatomy identification, view identification and/or image quality evaluation (step 202) automatically.As mentioned above, features relevant/the parameter that extracts/determine from image data set according to imaging pattern, the clinical field supported with for providing automatic decision to support that performed method is different, and those of ordinary skill in the art can easily predict various types of characteristics or parameter, and described characteristic or parameter can be extracted or determine to be used for automatic anatomy and view recognition methods and the image quality evaluating method according to one exemplary embodiment of the present invention from medical image.For example, can the extraction various parameters relevant with optical density and contrast.Feature extraction can carry out known cut apart and/or filter method to be used for utilizing known method that interested feature or anatomy are cut apart with reference to picture characteristics (for example variation of the variation of edge, discernible structure, border, color or brightness or transition, spectral information or transition or the like) known or expection.These features can comprise can be from the characteristic of any kind that image extracted, for example given shape or structure.In addition, can cross over the various types of characteristics of Image Acquisition, for example the motion of specified point or special characteristic are crossed over the variation of image.In other embodiments, characteristic can comprise along each axle (x, y, the combination of the gradient characteristic of z) being calculated from view data, pixel brightness contribution or other statistical nature or different characteristic.
Can utilize the method that be used for automatic anatomy identification, automatic view identification and image quality evaluation (step 202) of one or more technology execution according to one exemplary embodiment of the present invention, wherein said technology comprises database query method (for example Fig. 3), template facture (for example Fig. 4) and/or classification (for example Fig. 5), utilizes the feature that is extracted that the automatic decision support function is provided.
To carry out mark to image data set or with other mode classify (step 203) based on the result that is obtained.For example, for anatomy and view identification, will come the mark medical image with suitable anatomy and view identification.In addition, for each anatomy/view ID mark, definite anatomy/view of being discerned is estimated by the degree of confidence of correct labeling or likelihood.And for image quality evaluation, medical image can comprise (in preset range) quality score, and this quality score provides the indication of the quality of diagnosis level of medical image.
Fig. 3 is the process flow diagram that the method for supporting according to the automatic decision that is used to utilize the database query method to carry out medical imaging of one exemplary embodiment of the present invention is shown.The method of Fig. 3 can and be performed in the step 202 of Fig. 2 by module (103), (104) and/or (105) of Fig. 1.In an one exemplary embodiment, can use the characteristic that extracts from image data set to formulate inquiry, and can visit the database (step 300) of known cases and utilize inquiry to search for.The characteristic that is extracted that comprises inquiry can compare with the feature of known cases with case (step 301) like the recognition category.The content of the case that is identified can be used for object images then and determine most probable anatomy or view, perhaps the quality (step 302) of definite image of being gathered.
For example consider the problem of the apical four-chamber figure in the identification ultrasonic cardiography.One group of typical apical four-chamber figure will represent a plurality of features, for example the overall shape of the existence in four chambeies and heart.Also can be described by the disappearance (for example main artery efferent tract (it may be applicable to the so-called apex of the heart five chamber figure) does not exist) of further feature.These features can be extracted from test pattern, and compare with a stack features from known view.
Identical design can be used in the anatomy identification.For example consider the ultrasonoscopy of kidney.Feature can be extracted, and comprises that with expression the database of case of all types of anatomy of liver, gall-bladder, kidney or the like compares.A people even can in database, have right kidney and left kidney.Based on the comparison of these known cases, can report most probable anatomy.
Can utilize in the common U.S. Patent application of transferring the possession of disclosed technology to carry out to be used for to the database of image and index and use the method for low-level features search database, the sequence number of described U.S. Patent application is No.10/703,204, submitting day to is on November 6th, 2003, denomination of invention is " System and Method for Performing ProbabilisticClassification and Decision Support Using Multidimensional MedicalImage Databases (being used to utilize the multidimensional medical image databases to carry out the system and method for probabilistic classification and decision support) ", and described application is incorporated into this with for referencial use.In one embodiment, database can be with image or is only made up with the character representation of image.System can recognition category like image, determine anatomy, view and/or quality based on the content of similar image then.
Fig. 4 illustrates to be used to utilize the process flow diagram of carrying out the method that the automatic decision of medical imaging supports based on the method for template according to one exemplary embodiment of the present invention.In an one exemplary embodiment, can utilize the characteristic that extracts from image data set to formulate inquiry, and database (step 400) that can access templates and utilizing is inquired and is searched for.The characteristic that is extracted that comprises inquiry can compare with the feature of template with template (step 401) like the recognition category.The content of the template that is identified can be used for object images then and determine most probable anatomy or view, perhaps the quality (step 402) of definite image of being gathered.As mentioned above, the database of known cases can be used to make up template.For example can be different cardiac views: apical four-chamber, the apex of the heart two chambeies or the like make up template.System can estimate each the similarity with these modules then, and this provides the more shirtsleeve operation method of search database then.
Fig. 5 is the process flow diagram that the method for supporting according to the automatic decision that is used to utilize classification to carry out medical imaging of one exemplary embodiment of the present invention is shown.In this one exemplary embodiment, will be imported into (step 500) the sorter from the characteristic that image data set extracted, described sorter is trained or designs the processing feature data with view data is classified (step 501).Classification results will be used to determine most probable anatomy or view, or assess image quality (step 502).
For example, a classifiers can be fabricated, based on the feature that is extracted image is classified.That is to say, can be based on database " study " classifiers of case.These sorters will use this stack features as input, and classify the image as anatomy, view or the quality level that belongs to specific.In this one exemplary embodiment of Fig. 1, categorizing system (108) comprises knowledge base (110), and this knowledge base is used to handle the features/parameters that extracted and image is classified.Knowledge base (110) is preserved one or more disaggregated models of being trained, parameter and/or other data structure of the knowledge learnt or the like.
Should be understood that the term of Shi Yonging " sorter " is commonly referred to as various types of sorter frameworks here, for example classify device, integrated classifier or the like.In addition, classifier design can comprise the multiple classifition device, and described multiple classifition device is attempted data are divided into two groups and be organized in the mode of hierarchically organized or parallel running, is combined then to find optimal classification.In addition, sorter can comprise integrated classifier, wherein all a large amount of sorters (being called as " sorter group ") of attempting to carry out the same category task are learnt, but train with different pieces of information/variables/parameters, are combined then to produce final key words sorting.Performed sorting technique can be " black box ", and it can not be to their prediction of user interpretation (being exactly this situation if utilize neural network to set up sorter for example).Sorting technique can be " a white case ", and it adopts the readable form of people (being exactly this situation if utilize decision tree to set up sorter for example).In other embodiments, disaggregated model can be " a grey case ", and it can partly explain it is (for example, the combination of " white case " and " black box " type sorter) of how to derive solution.
Should be understood that described system and method can be realized with the multi-form of hardware, software, firmware, application specific processor or its combination according to the present invention here.For example, described here system and method can be used as the application program that comprises programmed instruction and realizes with the form of software, described programmed instruction visibly is embodied in (for example hard disk, flexible plastic disc, RAM, CD Rom, DVD, ROM and flash memory) on one or more procedure stores devices, and can be carried out by any equipment that comprises suitable framework or machine.In addition, because described composition system module and method step can be realized with the form of software in the accompanying drawings, so the mode that the actual connection between the system unit (or flow process of process steps) can be programmed according to application programs and difference.Be given in the instruction here, those of ordinary skill in the related art can predict of the present invention these and implement or configuration with similar.
Should be understood that further system and a method according to the invention can be used as conventional CAD method or other expansion that is used for the automatic diagnosis method of image data processing is implemented.In addition, should be understood that described example system and method can be utilized 3D medical imaging and CAD system or be suitable for being used on a large scale diagnosing and the application program of the imaging pattern (CT, MRI or the like) assessed and easily being implemented here.In this, although one exemplary embodiment can be here is described with reference to specific imaging pattern or specific anatomical features, nothing should be interpreted into is restriction to scope of the present invention.
Although exemplary embodiment of the present invention here has been described with reference to the drawings, but should be understood that, the present invention is not limited to those clear and definite embodiment, and do not depart from the scope of the present invention or the situation of spirit under, those skilled in the art can realize various other variation and modifications here.All such changes and modifications mean and are included in the scope of the present invention that is limited by claims.
Claims (30)
1. the method that is used to medical imaging to provide automatic decision to support comprises:
Obtain view data;
Extract characteristic from described view data; With
Utilize the characteristic that is extracted automatically to determine the quality of diagnosis of described view data.
2. according to the process of claim 1 wherein that described view data comprises the cardiac ultrasound images data.
3. according to the method for claim 1, further comprise the module of the level of automatically determining the indication quality of diagnosis.
4. automatically determine quality of diagnosis according to the process of claim 1 wherein to determine quality of diagnosis by the characteristic using sorter to handle to be extracted.
5. according to the method for claim 4, wherein use machine learning method, realize sorter based on the method for model or machine learning with based on any combination of the method for model.
6. come like the recognition category case and use the information relevant to determine quality of diagnosis by the database that utilizes the characteristic inquiry known cases extracted, automatically definite picture quality according to the process of claim 1 wherein with the similar case of being discerned.
7. according to the method for claim 1, wherein come like the recognition category template and use the information relevant to determine quality of diagnosis, automatically definite picture quality with the similar template of being discerned by the database that utilizes the template that the characteristic inquiry extracted derived from the information of known cases.
8. according to the method for claim 1, comprise that further characteristic that utilization is extracted automatically discerns the anatomical object with marked image data.
9. according to the method for claim 7, determine automatically that wherein the quality of diagnosis of view data further comprises the result who uses automatic anatomy identification.
10. according to the method for claim 1, comprise that further characteristic that utilization is extracted automatically discerns the view with marked image data.
11., determine automatically that wherein the quality of diagnosis of view data further comprises the result who uses automatic view identification according to the method for claim 10.
12. according to the method for claim 10, wherein said view is the normal view of interested clinical field.
13. during image acquisition, utilize the characteristic that is extracted automatically to determine the quality of diagnosis of view data in real time according to the process of claim 1 wherein.
14., quality of diagnosis is determined that the result offers the user in real time during further being included in image acquisition according to the method for claim 13.
15. the method that is used to medical imaging to provide automatic decision to support comprises:
Obtain view data;
Extract characteristic from described view data; With
Utilize the characteristic that is extracted to carry out automatic anatomy identification process, so that the anatomical object of being discerned in view data is carried out mark.
16., comprise that further the characteristic that utilization is extracted and the result of anatomy identification carry out the automatic view identifying, carry out mark with the view to view data according to the method for claim 15.
17. according to the method for claim 16, comprise that further the result of characteristic that utilization is extracted and anatomy and view identification carries out automated graphics quality assessment process, with the level of the quality of diagnosis of determining view data.
18., wherein during image acquisition, carry out automatic anatomy identification, view identification and image quality evaluation in real time according to the method for claim 17.
19. according to the method for claim 18, the result who further is included in during the image acquisition automatic anatomy identification, view identification and image quality evaluation process offers the user in real time.
20., wherein utilize the correlation classifier of handling the characteristic that is extracted to carry out automatic anatomy identification, view identification and image quality evaluation according to the method for claim 17.
21., wherein use machine learning method, realize sorter based on the method for model or machine learning with based on any combination of the method for model according to the method for claim 20.
22. method according to claim 17, wherein come case like the recognition category and use the information relevant to carry out separately function, carry out automatic anatomy identification, view is discerned and image quality evaluation with the similar case of being discerned by the database that utilizes the characteristic inquiry known cases extracted.
23. method according to claim 17, wherein come template like the recognition category and use the information relevant to carry out separately function, carry out automatic anatomy identification, view is discerned and image quality evaluation with the similar template of being discerned by the database that utilizes the template that the characteristic inquiry extracted derived from the information of known cases.
24. the method that is used to medical imaging to provide automatic decision to support comprises:
Obtain view data;
Extract characteristic from described view data; With
Utilize the characteristic that is extracted to carry out the automatic view identifying, with the view of recognition image data.
25. according to the method for claim 24, wherein said view data comprises the view data of utilizing the heart that ultrasonic imaging gathers.
26. according to the method for claim 25, wherein view is identified as the apex of the heart two chamber figure (A2C), apical four-chamber figure (A4C), apex of the heart major axis figure (ALAX), parasternal major axis figure (PLAX) or parasternal minor axis figure (PSAX).
27. according to the method for claim 25, comprise that further the result of characteristic that utilization is extracted and view identification carries out automated graphics quality assessment process, with the level of the quality of diagnosis of determining view data.
28. method according to claim 27, wherein by determining and the assessment module is carried out automated graphics quality assessment process, described module provides to comprise wherein about the cardiac image of being gathered and has the indication of view of dwindling the apex of the heart of drafting by scenography.
29., wherein during image acquisition, utilize the characteristic that is extracted automatically to determine the quality of diagnosis of view data in real time according to the method for claim 28.
30., quality of diagnosis is determined that the result offers the user in real time during further being included in image acquisition according to the method for claim 29.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US55562004P | 2004-03-23 | 2004-03-23 | |
| US60/555,620 | 2004-03-23 | ||
| US11/083,161 | 2005-03-17 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1934589A true CN1934589A (en) | 2007-03-21 |
Family
ID=37879390
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2005800093058A Pending CN1934589A (en) | 2004-03-23 | 2005-03-23 | Systems and methods providing automated decision support for medical imaging |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1934589A (en) |
Cited By (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102460444A (en) * | 2009-06-09 | 2012-05-16 | 皇家飞利浦电子股份有限公司 | Apparatus and method for ordering stored images |
| CN102612695A (en) * | 2009-10-15 | 2012-07-25 | 埃斯奥特欧洲有限公司 | Apparatus and method for performing diagnostic imaging examinations with tutorial means for the user, both in the preparatory step and in the operative step |
| CN103034779A (en) * | 2011-09-30 | 2013-04-10 | 通用电气公司 | Systems and methods for providing automated imaging feedback |
| CN103377369A (en) * | 2012-04-18 | 2013-10-30 | 国际商业机器公司 | Method and system for generating cardiovascular measurement for individual identification |
| CN102056547B (en) * | 2008-06-03 | 2014-05-14 | 株式会社日立医疗器械 | Medical image processing device and method for processing medical image |
| CN103946885A (en) * | 2011-11-17 | 2014-07-23 | 拜耳医疗保健公司 | Methods and techniques for collecting, reporting, and managing information about medical diagnostic procedures |
| CN104424385A (en) * | 2013-08-22 | 2015-03-18 | 上海联影医疗科技有限公司 | Method and device for evaluating medical images |
| CN104757993A (en) * | 2014-01-07 | 2015-07-08 | 三星麦迪森株式会社 | Method and medical imaging apparatus for displaying medical images |
| CN105118002A (en) * | 2015-07-30 | 2015-12-02 | 芜湖卫健康物联网医疗科技有限公司 | Five-step-method grading diagnosis system and method |
| CN106408610A (en) * | 2015-04-16 | 2017-02-15 | 西门子公司 | Method and system for machine learning based assessment of fractional flow reserve |
| CN106462662A (en) * | 2014-05-12 | 2017-02-22 | 皇家飞利浦有限公司 | Method and system for computer-aided patient stratification based on case difficulty |
| CN106650292A (en) * | 2017-01-04 | 2017-05-10 | 梁月强 | Personal health record system with process decision support function |
| CN107209809A (en) * | 2015-02-05 | 2017-09-26 | 皇家飞利浦有限公司 | Background for the report content of radiological report is created |
| WO2017193251A1 (en) * | 2016-05-09 | 2017-11-16 | 深圳迈瑞生物医疗电子股份有限公司 | Method and system for recognizing region of interest profile in ultrasound image |
| CN108056789A (en) * | 2017-12-19 | 2018-05-22 | 飞依诺科技(苏州)有限公司 | A kind of method and apparatus for the configuration parameter value for generating ultrasound scanning device |
| CN108171272A (en) * | 2018-01-12 | 2018-06-15 | 上海东软医疗科技有限公司 | A kind of evaluation method and device of Medical Imaging Technology |
| CN108701493A (en) * | 2016-02-29 | 2018-10-23 | 皇家飞利浦有限公司 | Equipment, system and method for the photographed image-related information for verifying medical image |
| CN108882917A (en) * | 2016-05-30 | 2018-11-23 | 深圳迈瑞生物医疗电子股份有限公司 | A kind of heart volume discriminance analysis system and method |
| CN109166108A (en) * | 2018-08-14 | 2019-01-08 | 上海融达信息科技有限公司 | A kind of automatic identifying method of CT images pulmonary abnormalities tissue |
| CN109190679A (en) * | 2018-08-08 | 2019-01-11 | 北京大学深圳医院 | Parathyroid gland learns and intelligent identifying system automatically in thyroid operation |
| CN109199438A (en) * | 2017-06-30 | 2019-01-15 | 通用电气公司 | For automatically determining the method and system of the anatomic measurement of ultrasound image |
| CN109362221A (en) * | 2016-03-09 | 2019-02-19 | 安科诺思公司 | Utilize the ultrasound image identifying system and method for smart network |
| CN109478325A (en) * | 2016-07-19 | 2019-03-15 | 唯盼健康科技有限公司 | System and device for clinical decision optimization |
| CN109758178A (en) * | 2017-11-10 | 2019-05-17 | 美国西门子医疗解决公司 | Machine-assisted workflow in ultrasound imaging |
| CN110692107A (en) * | 2017-05-31 | 2020-01-14 | 皇家飞利浦有限公司 | Machine learning of raw medical imaging data for clinical decision support |
| CN110717912A (en) * | 2019-10-18 | 2020-01-21 | 上海联影智能医疗科技有限公司 | Medical image quality control system, method, computer device and readable storage medium |
| CN110910342A (en) * | 2018-09-12 | 2020-03-24 | 西门子医疗有限公司 | Analyzing bone trauma by using deep learning |
| WO2021023051A1 (en) * | 2019-08-02 | 2021-02-11 | 无锡海斯凯尔医学技术有限公司 | Tissue evaluation method, apparatus and device, and computer-readable storage medium |
| CN113129343A (en) * | 2020-01-14 | 2021-07-16 | 通用电气精准医疗有限责任公司 | Method and system for anatomical structure/view classification in X-ray imaging |
| CN113143300A (en) * | 2020-01-22 | 2021-07-23 | 上海西门子医疗器械有限公司 | X-ray medical imaging system and method and computer storage medium |
-
2005
- 2005-03-23 CN CNA2005800093058A patent/CN1934589A/en active Pending
Cited By (41)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102056547B (en) * | 2008-06-03 | 2014-05-14 | 株式会社日立医疗器械 | Medical image processing device and method for processing medical image |
| CN102460444A (en) * | 2009-06-09 | 2012-05-16 | 皇家飞利浦电子股份有限公司 | Apparatus and method for ordering stored images |
| CN102460444B (en) * | 2009-06-09 | 2016-01-27 | 皇家飞利浦电子股份有限公司 | For the apparatus and method sorted to the image stored |
| CN102612695A (en) * | 2009-10-15 | 2012-07-25 | 埃斯奥特欧洲有限公司 | Apparatus and method for performing diagnostic imaging examinations with tutorial means for the user, both in the preparatory step and in the operative step |
| CN103034779A (en) * | 2011-09-30 | 2013-04-10 | 通用电气公司 | Systems and methods for providing automated imaging feedback |
| CN103946885A (en) * | 2011-11-17 | 2014-07-23 | 拜耳医疗保健公司 | Methods and techniques for collecting, reporting, and managing information about medical diagnostic procedures |
| CN103377369A (en) * | 2012-04-18 | 2013-10-30 | 国际商业机器公司 | Method and system for generating cardiovascular measurement for individual identification |
| CN103377369B (en) * | 2012-04-18 | 2016-12-28 | 国际商业机器公司 | Generate the method and system of the cardiovascular measurement identified for individuality |
| CN104424385B (en) * | 2013-08-22 | 2018-08-31 | 武汉联影医疗科技有限公司 | A kind of evaluation method and device of medical image |
| CN104424385A (en) * | 2013-08-22 | 2015-03-18 | 上海联影医疗科技有限公司 | Method and device for evaluating medical images |
| CN104757993A (en) * | 2014-01-07 | 2015-07-08 | 三星麦迪森株式会社 | Method and medical imaging apparatus for displaying medical images |
| US10585940B2 (en) | 2014-05-12 | 2020-03-10 | Koninklijke Philips N.V. | Method and system for computer-aided patient stratification based on case difficulty |
| CN106462662B (en) * | 2014-05-12 | 2020-10-02 | 皇家飞利浦有限公司 | Method and system for computer-assisted patient stratification based on case difficulty |
| CN106462662A (en) * | 2014-05-12 | 2017-02-22 | 皇家飞利浦有限公司 | Method and system for computer-aided patient stratification based on case difficulty |
| CN107209809A (en) * | 2015-02-05 | 2017-09-26 | 皇家飞利浦有限公司 | Background for the report content of radiological report is created |
| CN106408610B (en) * | 2015-04-16 | 2020-05-19 | 西门子公司 | Method and system for anatomical object detection using a marginal space deep neural network |
| CN106408610A (en) * | 2015-04-16 | 2017-02-15 | 西门子公司 | Method and system for machine learning based assessment of fractional flow reserve |
| CN105118002A (en) * | 2015-07-30 | 2015-12-02 | 芜湖卫健康物联网医疗科技有限公司 | Five-step-method grading diagnosis system and method |
| CN108701493A (en) * | 2016-02-29 | 2018-10-23 | 皇家飞利浦有限公司 | Equipment, system and method for the photographed image-related information for verifying medical image |
| CN109362221A (en) * | 2016-03-09 | 2019-02-19 | 安科诺思公司 | Utilize the ultrasound image identifying system and method for smart network |
| CN108701354A (en) * | 2016-05-09 | 2018-10-23 | 深圳迈瑞生物医疗电子股份有限公司 | Identify the method and system of area-of-interest profile in ultrasonoscopy |
| WO2017193251A1 (en) * | 2016-05-09 | 2017-11-16 | 深圳迈瑞生物医疗电子股份有限公司 | Method and system for recognizing region of interest profile in ultrasound image |
| CN108701354B (en) * | 2016-05-09 | 2022-05-06 | 深圳迈瑞生物医疗电子股份有限公司 | Method and system for identifying contour of interest region in ultrasonic image |
| CN108882917A (en) * | 2016-05-30 | 2018-11-23 | 深圳迈瑞生物医疗电子股份有限公司 | A kind of heart volume discriminance analysis system and method |
| CN109478325A (en) * | 2016-07-19 | 2019-03-15 | 唯盼健康科技有限公司 | System and device for clinical decision optimization |
| CN106650292A (en) * | 2017-01-04 | 2017-05-10 | 梁月强 | Personal health record system with process decision support function |
| CN110692107A (en) * | 2017-05-31 | 2020-01-14 | 皇家飞利浦有限公司 | Machine learning of raw medical imaging data for clinical decision support |
| CN110692107B (en) * | 2017-05-31 | 2023-11-14 | 皇家飞利浦有限公司 | Machine learning of raw medical imaging data for clinical decision support |
| CN109199438B (en) * | 2017-06-30 | 2022-12-27 | 通用电气公司 | Method and system for automatically determining anatomical measurements of an ultrasound image |
| CN109199438A (en) * | 2017-06-30 | 2019-01-15 | 通用电气公司 | For automatically determining the method and system of the anatomic measurement of ultrasound image |
| CN109758178A (en) * | 2017-11-10 | 2019-05-17 | 美国西门子医疗解决公司 | Machine-assisted workflow in ultrasound imaging |
| CN108056789A (en) * | 2017-12-19 | 2018-05-22 | 飞依诺科技(苏州)有限公司 | A kind of method and apparatus for the configuration parameter value for generating ultrasound scanning device |
| CN108171272A (en) * | 2018-01-12 | 2018-06-15 | 上海东软医疗科技有限公司 | A kind of evaluation method and device of Medical Imaging Technology |
| CN109190679A (en) * | 2018-08-08 | 2019-01-11 | 北京大学深圳医院 | Parathyroid gland learns and intelligent identifying system automatically in thyroid operation |
| CN109166108A (en) * | 2018-08-14 | 2019-01-08 | 上海融达信息科技有限公司 | A kind of automatic identifying method of CT images pulmonary abnormalities tissue |
| CN110910342A (en) * | 2018-09-12 | 2020-03-24 | 西门子医疗有限公司 | Analyzing bone trauma by using deep learning |
| CN110910342B (en) * | 2018-09-12 | 2023-11-14 | 西门子医疗有限公司 | Analysis of skeletal trauma by using deep learning |
| WO2021023051A1 (en) * | 2019-08-02 | 2021-02-11 | 无锡海斯凯尔医学技术有限公司 | Tissue evaluation method, apparatus and device, and computer-readable storage medium |
| CN110717912A (en) * | 2019-10-18 | 2020-01-21 | 上海联影智能医疗科技有限公司 | Medical image quality control system, method, computer device and readable storage medium |
| CN113129343A (en) * | 2020-01-14 | 2021-07-16 | 通用电气精准医疗有限责任公司 | Method and system for anatomical structure/view classification in X-ray imaging |
| CN113143300A (en) * | 2020-01-22 | 2021-07-23 | 上海西门子医疗器械有限公司 | X-ray medical imaging system and method and computer storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1934589A (en) | Systems and methods providing automated decision support for medical imaging | |
| US7672491B2 (en) | Systems and methods providing automated decision support and medical imaging | |
| EP3806744B1 (en) | Immediate workup | |
| JP6807820B2 (en) | Image search device, method and program | |
| US7529394B2 (en) | CAD (computer-aided decision) support for medical imaging using machine learning to adapt CAD process with knowledge collected during routine use of CAD system | |
| CN105760874B (en) | CT image processing system and its CT image processing method towards pneumoconiosis | |
| US20180061049A1 (en) | Systems and methods for analyzing in vivo tissue volumes using medical imaging data | |
| Hennessey et al. | Artificial intelligence in veterinary diagnostic imaging: A literature review | |
| CN1839392A (en) | Automated regional myocardial assessment for cardiac imaging | |
| CN114782307A (en) | Enhanced CT image colorectal cancer staging auxiliary diagnosis system based on deep learning | |
| US11854190B2 (en) | Similarity determination apparatus, similarity determination method, and similarity determination program | |
| WO2020110519A1 (en) | Similarity determination device, method, and program | |
| CN110838114A (en) | Pulmonary nodule detection method, device and computer storage medium | |
| Lin et al. | Using deep learning in ultrasound imaging of bicipital peritendinous effusion to grade inflammation severity | |
| Peña-Solórzano et al. | Findings from machine learning in clinical medical imaging applications–Lessons for translation to the forensic setting | |
| Wang et al. | Deep learning for automatic identification of nodule morphology features and prediction of lung cancer | |
| Dhalia Sweetlin et al. | Patient-Specific Model Based Segmentation of Lung Computed Tomographic Images. | |
| CN112508943A (en) | Breast tumor identification method based on ultrasonic image | |
| CN1820274A (en) | Cad (computer-aided decision) support for medical imaging using machine learning to adapt cad process with knowledge collected during routine use of cad system | |
| CN111265234A (en) | Method and system for judging properties of lung mediastinal lymph nodes | |
| Gheysari et al. | Lung ct image segmentation using reinforcement learning | |
| US20230096522A1 (en) | Method and system for annotation of medical images | |
| Ramos | Analysis of medical images to support decision-making in the musculoskeletal field | |
| Sudha et al. | A deep learning approach to identify and classify lung tumor from CT–Scans | |
| Moh'd Rasoul et al. | Pneumonia identification using organizing map algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20070321 |