WO2011146994A2

WO2011146994A2 - Procedure, platform and system for the analysis of medical images

Info

Publication number: WO2011146994A2
Application number: PCT/AU2011/000636
Authority: WO
Inventors: Paul Clark
Original assignee: Magnepath Pty Ltd.
Priority date: 2010-05-28
Filing date: 2011-05-26
Publication date: 2011-12-01
Also published as: WO2011146994A3

Abstract

A procedure to perform an image analysis task for the analysis of at least one region of interest (ROI(s)) on one or more medical images (image(s)) by two or more image analysis tools without the duplication of common image analysis activities between tools, where a function list of each tool is compared with a function list of each other tool to determine if there is at least one common image analysis function between tools that operates on common input data, such that the common image analysis function need then be only executed once during the image analysis task by a single instantiation of that function that acts on the common input data. A workflow starts with image acquisition (1), where the resulting digital image or images (2) may be assessed with other clinical information (3) at the stage of image interpretation (5). The interpeted images and other interpreted data may then be sent for storage and distribution (8), and culminate with their use in treatment (9). The procedure, platform, and system can be readily integrated into the diagnostic workflow as an adjunct processing step (6) at the stage of image interpretation (5). The digital results (7) from the processing step (6) can be used to aid the interpretation (5). One or more of (pre-processing 210), core-processing (220) and/or post-processing (230) can be carried out on the image data. A tele-health system, preferably with tele-analysis, is also disclosed.

Description

PROCEDURE, PLATFORM AND SYSTEM

FOR THE ANALYSIS OF MEDICAL IMAGES

TECHNICAL FIELD

The present invention relates to analysis of medical images by multiple image analysis tools.

BACKGROUND

Advances in medical imaging technologies, coupled with the increasing complexity of the imaging procedures, have resulted in a growing number of images per patient examination. To improve radiology workflow and reduce risk of human error in the interpretation of the data sets, computer-aided analysis is increasingly adopted in imaging centres.

Computer-aided analysis also provides for the advanced visualisation of organs and tissues in three dimensions, plus the ability to extract quantitative information on the change in medical image scanner signal behaviour with the onset of pathology. The potential of computer-aided analysis to assist in the diagnosis and screening of disease, and the increasing demand for non-invasive diagnostic techniques, has thus lead to a rapid growth in the number of analytical and clinical imaging applications available.

Medical imaging applications are typically offered as stand-alone or add-on products to diagnostic imaging equipment, and are typically run on dedicated hardware and software platforms. There are thus significant costs associated with the purchase and maintenance of imaging applications, plus the hardware on which they operate. Many non-invasive diagnostic solutions are therefore outside the budgets of smaller healthcare providers, and may remain so unless pay-peruse access to these solutions can be made available through web-based application service providers.

The advantages of the application service provider (ASP) model for access to software on a pay-per-use or subscription basis are well known. The ASP model removes the capital and set up costs of ownership of software and hardware by providing remote processing on the dedicated systems of the service provider, which typically offer higher levels of performance than that which can be achieved in-house, in addition to guaranteed availability. Access to software through an ASP also removes the need to maintain and upgrade specially purchased software.

The potential of the ASP model for the provision of diagnostic imaging services has become a reality with the availability of high-bandwidth and low-cost internet access, making practical the transmission of medical image data sets over acceptable wait times. Picture Archiving and Communication Systems (PACS) are progressively transitioning from dedicated servers accessed through specially configured workstations, to web-based systems accessible by varied computing platforms over the Internet. PACS are now thus also being offered on an ASP basis (such as GE Medical Systems, Centricity® RIS/PACS, http://www.gehealthcare.com/euen/iis/products/radiology/centricity_ris_pacs/centr icity_ris_pacs.html, viewed 19 Apr. 2010 and Philips Medical Systems, iSite PACS,

http://www.healthcare.philips.com/main/products/healthcare_informatics/products/ enterprise_imaging_informatics/isite_pacs/index.wpd viewed 19 Apr. 2010).

Through ASP web-PACS, the diagnostic imaging service provider no longer has to maintain their own image servers and archives, but can store their images remotely through the hardware and software assets provided by the ASP.

The ASP model to the storage and retrieval of medical image data can also be applied to analysis of the image data through the provision of high performance computing services. This opportunity has been recognised by PACS manufacturers who offer an increasing number of medical image analysis applications through their systems. In fact, ASP access to 3D volume rendering of organs and tissues is now a reality. However, there are a number of deficiencies to the access of diagnostic image analysis applications through PACS. Not all applications are available through PACS, and not all PACS offer access to the same applications. Also, the applications themselves can be user intensive, requiring specific training and/or an understanding of the image analysis process. Very few applications offer fully automated image analysis - an ability that would lend them to use by a broader range and larger number of users. The potential for the provision of diagnostic image analysis services through an ASP model has been recognised (Cai W., Feng D., Fulton R., "Web- based digital medical images", IEEE Computer Graphics and Applications (Jan./Feb. 2001 ) 44-47; Zeng H., Fei D.Y., Fu C.T., Kraft, K.A., "Internet (WWW) based system of ultrasonic image processing tools for remote image analysis", Computer Methods and Programs in Biomedicine (2003) 71 :235-241 ; Liua B.J., Caob F., Zhoub M.Z., Mogelb G., Documet L, "Trends in PACS image storage and archive", Computerized Medical imaging and Graphics (2003) 207:165-174; Neri E., Bargellini I., Rieger M., Giachetti A., Vignali C, Tuveri M., Jaschke W., Bartolozzi C, "Abdominal aortic aneurysms: virtual imaging and analysis through a remote web server", European Journal of Radiology (2005) 15:348-352; Wei J.C., Daniel M.S., Valentino J., Bell D.S, Baker R.S., "A Web-based Telemedicine System for Diabetic Retinopathy Screening Using Digital Fundus Photography", Telemedicine and e-Health (2006) 12(1 ):50-57.). However, patents for such systems have only either been described in very general terms without any reference to how the image analysis is to be performed (DiFilippo F.P., Sivashankaran A., Behm S.M., Cottrill B.A., "Remote medical image analysis", US 6,829,378 B2), or only describe how the analysis is to be performed in a specific case, such as for vascular plaque detection (Huizenga J.T., Anderson R.W., Brotherton, T.W., "Automated methods and systems for vascular plaque detection and analysis", WO 2005/020790 A2).

However, no procedure, platform or system provides an ASP model for access to multiple image analysis tools in such a way that facilitates the optimum deployment and timely delivery of these tools in an automated manner, or that facilitates the inclusion of additional tools with minimal integration effort without duplicating the coding or execution of common activities between tools.

With the aforementioned in mind, it is desirable of the present invention to provide an improved procedure, platform and/or system for the analysis of medical images by multiple image analysis tools. SUMMARY OF THE INVENTION

One aspect of the present invention provides a procedure to perform an image analysis task for the analysis of one or more regions of interest (ROI(s)) on one or more medical images (image(s)) by two or more image analysis tools, without the duplication of common image analysis activities between tools, where a function list of each tool is compared with a function list of each other tool to determine if there is at least one common image analysis function between tools that operates on common input data, such that the common image analysis function need then be only executed once during the image analysis task by a single instantiation of that function that acts on the common input data.

Image analysis activities may include one or more of a pre-processing function, post-processing function or pre-processing and post processing functions.

A further aspect of the present invention provides a procedure to perform an image analysis task for the analysis of one or more regions of interest (ROI(s)) on one or more medical images (image(s)) by two or more image analysis tools, the procedure including:

for each tool, the image analysis activities segregated and sequenced into one or more pre-processing function(s), optional core-processing function(s), and post-processing function(s), whereby the image analysis activities of each tool are referenced by an ordered function list or such that the respective tool is instantiated by reference to an ordered list of its component functions; and, for the two or more tools, a function list of each said tool is compared to determine if there is any common image analysis function between tools that operates on common input data, such that the function need then be only executed once during the image analysis task by a single instantiation of that function that acts on the common input data.

Thus, the procedure may advantageously perform an image analysis task for the analysis of one or more regions of interest (ROI(s)) on one or more medical images (image(s)) by two or more image analysis tools without the duplication of common image analysis activities between tools, thereby saving on computing time and resources.

Preferably, the pre-processing functions involve the removal of medical image scanner dependencies and artefacts from the ROI(s) or image(s) to provide either or both:

improved image quality within the ROI(s) or image(s),

improved reliability and consistency of the results from core-processing activities, and

the post-processing functions provide for the enhanced visualisation or interpretation of the results from pre-processing activities or optional core- processing activities with either or both

multi-dimensional representation of the results,

quantitative information on the results.

Optionally, core-processing functions provide for transformation of the pre- processed ROI(s) or image(s) to give either or both:

the delineation of structure within the pre-processed ROI(s) or image(s), quantitative information on the medical image scanner signal behaviour captured within the pre-processed ROI(s) or image(s).

Segregation of a tool's image analysis activities into at least one or more of pre-processing functions, one or more core-processing functions, and/or one or more post-processing functions, may enable a variety of tools to be deployed that achieve their differences in operation primarily through differences in their core- processing functions, whilst potentially sharing any number of pre- and postprocessing functions.

Furthermore, segregation of an image analysis tool's activities into a sequence of at least one pre-processing, optional core-processing, and optional post-processing function(s), or combination thereof, may further facilitate automation of the image analysis tool's activities.

Comparison of function lists for multiple tools may also enable different functions acting on common input data to be identified and then executed in parallel.

Output of a pre-processing function may be input into another preprocessing function, a core-processing function, or post-processing function. Also, the output of a core-processing function may be input into another core- processing function or a post-processing function. Further, the output of a postprocessing function may be input into another post-processing function or as the output result for an image analysis task.

One or more core-processing activities may be combined for the quantification of specific parameters characterising the medical image scanner signal behaviour captured within the pre-processed ROI(s) or image(s).

Also, for an image analysis task employing multiple tools, any said preprocessing functions may be executed during a defined pre-processing stage of the analysis, any said core-processing functions may be executed during a defined core-processing stage of the analysis, and any said post-processing functions may be executed during a defined post-processing stage of the analysis.

For any said analysis stage, the first function to be executed for each tool in the stage may be executed during a defined first processing step for the stage, and the second possible function to be executed for each tool in the stage may be executed during a defined second processing step for the stage, and so on.

For each analysis stage, an input data list may be created for each processing step. This may include a number of input data items to be processed by one or more of the functions to be executed in the analysis step, where an input data item is selected from:

for the first processing step:

of the pre-processing stage, at least any one of:

^■ a medical image,

^■ a ROI defining a part of a medical image,

^■ and possibly any number of the image's acquisition settings, of the core-processing stage, at least any one of:

^■ a medical image that has been pre-processed,

^■ a ROI defining a part of a medical image that has been pre- processed,

^■ a series of medical images that have been pre-processed,

^■ a series of ROIs defining parts of medical images that have been pre-processed, ^■ and preferably any number of values derived from pre-processing; of the post-processing stage, at least any one of:

^■ a medical image that has been pre-processed or core-processed,

^■ a ROI defining a part of a medical image that has been pre- processed or core-processed,

^■ a series of medical images that have been pre-processed or core- processed,

^■ a series of ROIs defining parts of medical images that have been pre-processed or core-processed,

^■ and preferably any number of values derived from pre-processing or core-processing; and preferably further;

for any subsequent processing steps, an input generated from the output data of a previous processing step.

For any said analysis stage, the parallel execution of functions for any said processing step, without duplicating function execution for common functions between tools acting on a common input data item, may be achieved by a sub- procedure applied to the stage that for each input data item on the input data list: identifies the functions to be executed on the input data item from each tool's function list;

adds only one reference for each different function identified to a function list for the input data item; and

executes the functions on the input data item's function list in parallel, thereby ensuring that no function is executed more than once on the same input data item.

For each analysis step in any given analysis stage, the sub-procedure may execute the functions from the function list(s) of different input data items in parallel.

The pre-processing activities may include one or more of the following activities, such that core-processing activities to determine quantitative information on the medical image scanner signal behaviour can provide consistent results across images taken from any make and model of scanner for the same imaging modality: Modelling the signal behaviour of the background noise in the image(s), Subtraction of any background signal level offset from the image(s), Reduction of any scanner-induced artefacts within the image(s),

Reduction of any subject-induced artefacts within the image(s),

Modelling any non-uniform signal penetration throughout the image(s) owing to the scanner and/or subject, and

Rescaling the signal intensities within the image(s) relative to a reference signal within the image(s).

One or more forms of the present invention may provide or include a core- processing activity for the delineation of structure within the ROI(s) or image(s) that identifies an organ or tissue.

One or more forms of the present invention may also provide or include a core-processing activity for the quantification of medical image scanner signal behaviour within the ROI(s) or images(s) that is directly or indirectly used to quantify pathological or physiological change.

A further form of the present invention provides the procedure to perform an image analysis task including, for each analysis stage, an input data list created for each processing step that is comprised of a number of input data items to be processed by one or more of the functions to be executed in the analysis step, where an input data item is one or more of:

for the first processing step;

of the pre-processing stage, at least any one of:

• a medical image,

• a ROI defining a part of a medical image,

• and possibly any number of the image's acquisition settings, of the core-processing stage, at least any one of:

• a medical image that has been pre-processed or not pre-processed,

• a ROI defining a part of a medical image that has been pre- processed or not pre-processed,

• a series of medical images that have been pre-processed or not pre-processed,

• a series of ROIs defining parts of medical images that have been pre-processed or not pre-processed, • and possibly any number of values derived from pre-processing if pre-processing has been performed;

of the post-processing stage, at least any one of:

• a medical image that has been pre-processed or core-processed or not pre-processed or core processed,

• a ROI defining a part of a medical image that has been pre- processed or core-processed or not pre-processed or core processed,

• a series of medical images that have been pre-processed or core- processed or not pre-processed or core processed,

• a series of ROIs defining parts of medical images that have been pre-processed or core-processed or not pre-processed or core processed,

• and possibly any number of values derived from pre-processing or core-processing if pre-processing or core processing has been performed;

Where the image analysis task is specified at a separate location from where the analysis by one or more image analysis tools is performed, thus requiring the electronic communication of the image analysis task over a telecommunications network, the communication may include:

an analysis request, consisting of one or more regions of interest on one or more medical images and possibly any number of the images' acquisition settings, for analysis by one or more specified image analysis tools, and

either or both of:

i) an automated analysis reply, consisting of the results of the analysis request, and optionally a record of the received analysis request, returned to the location at where the image analysis task was specified,

ii) a separate analysis retrieval, consisting of the results of the analysis request, and optionally a record of the received analysis request, retrieved from an analysis database and returned to the location at which the retrieval was initiated, where optional inclusion of a record of the received analysis request is used to verify that there has been no corruption of the ROI(s) or image(s) via their electronic communication.

A further form of the present invention provides an analysis platform with means to conduct any of the embodiments of the aforementioned procedures through one or more computers.

Another form of the present invention provides a tele-analysis platform incorporating means to conduct on or more of the embodiments of the aforementioned procedures through a distributed client-server computer network, preferably where:

one or more clients access the graphical user interface of a client-side application for the construction and transmittal of one or more analysis requests, and

one or more servers perform automated analysis by one or more image analysis tools for each analysis request through one or more server-side applications, and

one or more system administration utilities operate on one or more servers to:

sequence the execution of the one or more analysis requests,

return an automated analysis reply back to the requesting client for each analysis request,

forward a separate analysis retrieval to the client initiating the retrieval from an analysis database.

The distributed client-server computer network may be connected through the internet and:

the client-side application for the construction and transmittal of one or more analysis requests may be a web-derived application, and

the one or more server-side applications to conduct the automated analysis by one or more image analysis tools may operate one or more application servers, and

the one or more system administration utilities may operate on one or more web servers. The tele-analysis platform may incorporate Java technologies, and preferably where:

the client-side web-derived application for the construction and transmittal of one or more analysis requests may be implemented as either a Java Applet or Java WebStart Application,

the one or more server-side applications may be implemented through Java Beans or Enterprise Java Beans deployed in separate packages of preprocessing functions, core-processing functions, and post-processing functions, to promote scalability and reusability of tool components,

the one or more system administration utilities may be implemented through Java Beans or Enterprise Java Beans with Persistence Entities to record the state of each image analysis task.

A tele-health system in combination with a web portal enabling pay-peruse access to the one or more image analysis tools, where the web portal may include:

web pages accessed by the one or more clients providing customer operations to one or more of the following:

register a user for secure access to the tele-analysis platform,

provide secure access the tele-analysis platform interface,

track the state of one or more analysis requests,

retrieve the results of one or more analysis requests,

enable payment for one or more analysis requests; and

an eCommerce system deployed on the one or more servers including service databases of user details, analysis results, and payments, and

one or more server-side applications operating on the one or more web servers enabling the customer operations between the web pages and the service databases.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 shows integration of an embodiment of the present invention into the medical imaging and diagnosis workflow.

Figure 2 shows the framework of the system according to an embodiment of the present invention as a tele-health system based on a distributed client- server architecture, including the platform as a tele-analysis platform and a web portal, and with the procedure integrated into the tele-analysis platform.

Figure 3 shows a UML activity diagram for a procedure according to an embodiment of the present invention, where the image analysis activities for the tools used in an image analysis task are segregated and sequenced into preprocessing, core-processing, and post-processing stages of analysis.

Figure 4 shows an embodiment of the present invention including a UML activity diagram of the subprocedure for the parallel execution of mutliple functions belonging to multiple tools, without duplicating the execution of common functions between tools that act on common input data.

Figure 5 shows an architecture diagram of a preferred embodiment of the present invention as a tele-health system, comprising an internet-accessible tele- analysis platform and a customer web portal.

DESCRIPTION OF PREFERRED EMBODIMENTS

Embodiments of the present invention are explained in detail below with reference to the accompanying drawings.

In broad terms, embodiments of the present invention provide a scalable procedure, platform, and system for the automated analysis of medical images by multiple image analysis tools, without duplication of common activities between tools.

The procedure, platform, and system according to embodiments of the present invention has the potential to add significant value to the medical imaging and diagnosis workflow (see Figure 1 ). This workflow starts with image acquisition 1 , where the resulting digital image or images 2 may be assessed with other clinical information 3 at the stage of image interpretation 5. The interpeted images and other interpreted data may then be sent for storage and distribution 8, and culminate with their use in treatment 9. The procedure, platform, and system can be readily integrated into the diagnostic workflow as an adjunct processing step 6 at the stage of image interpretation 5. The digital results 7 from the processing step 6 can be used to aid the interpretation 5.

One embodiment of the present invention is that of a tele-health system (see Figure 2) which provides online access to a central repository of medical image analysis toolset or tools 10 (toolset and tools have the same meaning within this specification. A toolset is a set of one or more tools available manipulate and/or manage image data). The tele-health system provides paid access to the tools 10 through a web portal 20. The web portal links the user to the toolset by providing web page access 30 to a tele-analysis platform 40, within which the tools 10 are incorporated and through which the analysis is performed. The tele-analysis platform is thus a further embodiment of the invention. The tele-analysis platform is built on a distributed client-server architecture where the client-side of the platform 50 presents an interface 60 to the client with which to define an image analysis task and launch an analysis request 70. The analysis request is channelled into a pipeline of requests on the server-side of the platform 80 by system administration utilities 90, which lodge the requested image analysis task defined by the request for execution on an application server 100 by a procedure 1 10 that implements the tools for the task from the available toolset, and that also performs the task. The procedure to perform the image analysis task is scalable for the implementation of multiple image analysis tools, without duplicating common image analysis activities between tools. The procedure is also thus a further embodiment of the invention. Once the procedure has completed the image analysis task, an automated analysis reply 120 with the result of the task is communicated back to the client-side interface 60 of the teleanalysis platform 40 by the system administration utilities 90. If the interface for the teleanalysis platform has been closed prior the the completion of the image analysis task, the result of the analysis is stored in the server-side database 130 of the web portal's eCommerce system 140. The results may then be retrieved through the portal's web pages via eCommerce service utilities 150 which communicate with the service database. The interface for the teleanalysis platform may also be reopened to retrieve the analysis results, in which case the system administration utilities communicate with the service utilities to retrieve the results from the service database.

In the aforementioned procedure for performing an image analysis task using multiple image analysis tools, the image analysis activities for each tool are segregated and sequenced into defined pre-processing, core-processing, and post-processing stages of analysis, as illustrated in the UML activity diagram of Figure 3. A tool may contain any of these stages provided they are executed in the defined sequence as indicated in the diagram by way of the conditional forks. Pre-processing operations 210 remove scanner dependencies and artefacts from the image input data 200, providing improved image quality for the clearer display of features within the images and/or more consistent and reliable results from further processing. Core-processing operations 220 can then be performed, if desired, following pre-processing operations 210 or on the starting input data 200 for the delineation of structure within the images, and/or the quantitative measurement of medical image scanner behaviour within the imaged subject. Post-processing operations 230 can then be performed to provide advanced visualisation and/or quantification of the results from either the pre-processing or core-processing stages of analysis, or on the image input data directly. The output data 250 may thus derive from a pre-processing, core-processing, or postprocessing stage of analysis. Image processing activities are segregated in this manner, as tools typically share common pre-processing and post-processing functions, achieving their basic differences in operation through different core- processing functions. The procedure is thus scalable for the inclusion of additional tools with minimal coding effort, by focussing on inclusion of each new tool's particular core-processing functions, provided that their pre-processing and post-processing functions are already available. In addition, where multiple tools execute common functions on the same input data for an image analysis task, the common functions need only be executed once. Segregation of the image analysis tools' activities into component pre-processing, core-processing, and post-processing functions thus eliminates the duplication of activities that would occur if the tools were otherwise maintained as singular entities. The specified segregation and sharing of image analysis activities also helps simplify the process of automating operation of the image analysis tools.

Grouping of an image anlysis tool's activities into either pre-, core-, or postprocessing functions is performed on the basis that the activities grouped into a function provide a result that adheres to the definition of a pre-, core-, or postprocessing operation. A pre-processing function is thus a grouping of image analysis activities that peform an identifiable pre-processing operation, where the results of the operation can be directly passed on as an input to either core- processing or post-processing. Similarly, a core-processing function is a grouping of image analysis activities that peform an identifiable core-processing operation, where the results of the operation can be directly passed on as an input to post-processing. A post-processing function is a grouping of image analysis activities that peform an identifiable post-processing operation, where the results of the operation can be directly passed on as an analysis result that conforms to the expected configuration of output data for an image analysis task.

Whilst all functions must be able to provide a result that can be directly passed on to the next stage of analysis, be that from pre-processing to core- proessing, core-processing to post-processing, or post-processing to output data, a function may also take as input the results of a function within the same stage of analysis, and not just that from a previous stage. A tool's image analysis activities may thus be comprised of multiple pre-processing, core-processing, and post-processing functions to be executed in a specific sequence. Multiple core- processing functions may be required where a particular sequence of core- processing functions is necessary for the quantification of medical image scanner signal behaviour, in addition to the quantification of pathology.

Where an image analysis task calls on a number of tools for the analysis of one or more images, the procedure to perform the image analysis task must be able to call on the multiple functions within the pre-, core-, and post-processing stages without duplicating execution of any common function between tools when the function acts on common input data. Where common input data is acted on by different functions, these functions should be executed in parallel. In addition, where the same function operates on different input data, the function should also be executed in parallel.

To achieve the parallel execution of functions without duplicating function execution for common functions operating on common input data, the sequence of functions for each tool to be used in the image analysis task must be compared at corresponding function levels (or processing steps) in each processing stage. The first level of functions to be executed for all tools in either a pre-, core-, or post-processing stage thus comprise a first processing step for that stage, the second level of functions to be executed comprise a second processing step, and so on. For each input data item to be processed at a given step, the tool functions to be executed on that data item are identified, and are added only once to a list of functions that are to be executed on that input data item. The tool functions to be executed on that data are then referenced from the function list, thereby ensuring that no function is executed twice on the same input data. Further, as each input data item has it's own assoiciated function list, it is not only the functions operating on a given input data item that can be executed in parallel, but also the functions operating on different input data items.

The subprocedure for the parallel execution of mutliple functions belonging to multiple tools, without duplicating the execution of common functions between tools that act on common input data, is illustrated in the UML activity diagram of Figure 4. This subprocedure is applied to the pre-processing, core-processing, and post-processing stages of analysis. For each stage, an input data list 300 is associated with each step in the analysis, comprised of an input data item 340 or a number of input data items. The input data list is initially organised from input data 290 prior to the stage. However, where there is more than one step in the analysis stage, the input data list for subsequent steps is organised from the output data list 530 of the previous step, as indicated by the merge diamond 310. For the pre-processing stage of analysis, the input data list is initially generated from the analysis request, for which each input data item is either a medical image or a region of interest on a medical image as well as any number of the image's acquisition settings. For core-processing, the input data list is initially generated from the pre-processing output data list or the non-processed input data list. An input data item for core-processing is thus initially a non processed or pre-processed medical image or a non-processed or pre-processed region of interest (ROI) on a medical image as well as any number of values derived from pre-processing or the image acquisition settings. An input data item for core- processing may also however be a series of pre-processed images or ROIs, if a single transformed image or ROI is obtained on output from the core-processing function. For post-processing, the input data list is initially generated from either the pre-processing or core-processing output data list or the non-processed input data list, depending on whether or not pre- or core-processing has been performed. An input data item for post-processing may thus initially be a non-, pre- or core-processed image or ROI as well as any number of values derived from pre-processing or core-processing or the image acquisition settings. An input data item for post-processing may also however be a series of pre- processed or core-processed images or ROIs, if a single transformed image or ROI is obtained on output from the post-processing function.

Figure 4 shows that for each step in the analysis, an input data list 300 of input data items 340 is organised based on the next group of functions to be executed, as indicated by the action box 320. The input data items are processed in parallel as indicated by the parallel expansion region 330. For each input data item 340, a function list is created that lists the next functions to be executed on that data item, as shown by the action box 350. The functions matching those on the function list are executed concurrently as indicated by the conditional fork 360. Any number of functions N may be executed on the input data item, as indicated by action boxes 370, 380, and 390, depending on the number of tools that have been requested. At most, the number of functions to be executed equals the number of tools, but this may be less where the same function is in use by different tools on the same data item. After execution of a function, the tools that utilised the function are checked to see if they have any following functions to be executed, as indicated by the conditional forks 400, 410, and 420. Where there is no following function for one or more tools, the output is written once to the output data list 550 for the stage, as indicated by the action boxes 430, 440, and 450, and the workflow for those tools terminates for the stage, as shown by the "flow final" markers 460, 470, and 480. Where there is a next function for one or more tools, the output is written once to the current output data list 530, which becomes the new input data list for the next step in the analysis, as indicated by the action boxes 490, 500, and 510. When all the outputs for further processing have been written to the new input data list, the concurrent execution paths on the data item join back into a single path, as shown by the join 520. Processing on the data item is then complete, as demonstrated by the output to the expansion region 330. When all input data items for an analysis step have been processed, the new input data list is checked to determine if there is any further data to be processed for the stage, as indicated by the decision point 540. If the new input data list is not empty, then the next analysis step for the stage is performed by returning to the beginning of the procedure, where the input data list is organised for the next functions to be executed, as shown previously by the action box 320. (Note that the input data for the next functions may take as input one or more function outputs from the previous analysis step.) If the new input data list is empty, then all analysis steps for the stage have been completed, and the procedure returns the output data 550 for the stage.

In performance of the subprocedure, common functions between tools acting on common input data will most likely occur at the stage of pre-processing, as most tools share a common sequence of pre-processing functions. However, it is also possible that common functions may operate on common input data during core-processing or post-processing. For core-processing, this situation may arise where a core-processing result is not only required as an output for post-processing, but is also required to undergo a further transformation by another core-processing function to provide an additional output for postprocessing. For post-processing, this situation may arise where a core-processing result undergoes the same initial post-processing, before being rendered by two different post-processing functions into two visually different outputs.

To ensure that core-processing activities to determine quantitative information on medical image scanner signal behaviour provide consistent results across images taken from any make and model of scanner for the same imaging modality, a particular sequence of pre-processing operations must be performed. Whilst not all operations may need to be performed, the functions to perform each operation must be executed in the specified sequence. The pre-processing activities to be performed include one or more of the following operations:

modelling the signal behaviour of the background noise in the images;

subtraction of any background signal level offset from the images; reduction of any scanner-induced artefacts within the images;

reduction of any subject-induced artefacts within the images;

modelling any non-uniform signal penetration throughout the images owing to the scanner and/or subject;

rescaling the signal intensities within the images relative to a reference signal within the images. It will be apparent to those skilled in the relevant arts that the procedure outlined above to facilitate the analysis of medical images by multiple image analysis tools, without duplication of common activities between tools, can be used to perform an image analysis task for the analysis of one or more regions of interest on one or more of the medical images, where each region of interest (ROI) may comprise the entire medical image, or any part or parts thereof. It will also be apparent that a ROI need not be explicitly specified, in which case the ROI is implicitly taken to cover the entire medical image. It will also be apparent to those skilled in the relevant arts that the core-processing activities performed on the ROI to delineate structure may be used to identify an organ, tissue, or area of pathology. In addition, it will also be apparent that core-processing activities performed on the ROI to quantify medical image scanner signal behaviour may be used to directly or indirectly measure the extent and degree of pathological or physiological change.

In a further embodiment of the invention, the image analysis task to be performed by the aforementioned procedure may be specified at a separate location from where the analysis by one or more tools will be performed, thus requiring the communication of the image analysis task between the two separate locations. Whilst a variety of different methods may be used to communicate the image analysis task, a preferred embodiment of the invention uses electronic communication of the image analysis task over a telecommunications network. In order to process the image analysis task between the two separate locations, the communication can involve the transmittal of an analysis request, and preferably either the automated reception of an associated analysis reply or the separate transmittal of an analysis retrieval. The analysis request consists of data defining one or more ROIs on one or more medical images as well as any number of the images' acquisition settings for analysis by one or more specified image analysis tools. The analysis reply consists of the results of the analysis request, and optionally a record of the transmitted analysis request, automatically communicated back to the location at where the image analysis task was specified following completion of the analysis. Similarly, the analysis retrieval consists of the results of the analysis request, and an optional record of the transmitted analysis request, but retrieved from an analysis database for communication back to the location at which the retrieval was initiated. Where a record of the received analysis request is returned with the analysis reply and/or analysis retrieval, the returned request is used to verify that there has been no corruption of the medical images and/or the ROIs via their electronic communication.

For embodiments of the invention where the image analysis is performed at a separate location form where the image analysis task was specified, the procedure to implement multiple image analysis tools without the duplication of common activities between tools is incorporated into the aforementioned tele- analysis platform. The tele-analysis platform is deployed from one or more networked servers, through which clients gain access to a client-side application that provides a graphical user interface for the construction of an image analysis task, and transmittal of the task as an analysis request. Any number of analysis requests may thus be transmitted for processing on one or more of the networked servers. For each request, a server-side procedure is launched that instantiates the analysis tools for the task, and that also performs the task. Sequencing of the analysis requests is performed by system administration utilities that operate on one or more of the servers. The system administration utilities are also responsible for returning an automated analysis reply back to the requesting client for each analysis request, as well as responding to any analysis retrieval.

A preferred embodiment of the invention comprises the distributed client- server computer network of the tele-analysis platform with the communication between clients and servers conducted over the Internet. In this embodiment of the invention, the client-side application of the tele-analysis platform interface is a web-derived application, the system administration utilities operate on one or more web servers, and the server-side applications to perform the image analysis task operate one or more application servers.

Figure 5 shows in part a preferred embodiment of the invention as an internet-accessible tele-analysis platform 600, for which the platform software applications and utilities are built using Java technologies. For this specific embodiment, the client-side application of the tele-analysis platform interface is implemented as either a Java Applet or Java WebStart Application 610 (run on the client machine 605 within the application client Java Virtual Machine 608). The system administration utilities (e.g. run on a web server 615) can be implemented through Java Beans or Enterprise Java Beans 620 with Persistence Entities 630 (e.g. run on an entity manager 632 within the EJB container 634) to record the state of each image analysis task (run on an application server 638 within the EJB container 635). The server-side applications to perform the image analysis task are implemented through Java Beans or Enterprise Java Beans that are deployed in separately packaged classes of pre-processing functions 640, optional core-processing functions 650, and post-processing functions 660 through which to instantiate the image analysis tools.

In at least one preferred embodiment of the invention as the aforementioned tele-health system, which provides paid online access to a central repository of medical image analysis tools, the system includes the internet-accessible tele-analysis platform 600, and a customer web portal 700 which provides access to the platform. The customer web portal consists of web pages 680 (e.g. accessed through a client's web browser 682) and an eCommerce system 685 deployed on one or more servers. The web pages 680 provide access to server-side customer operations 690 (operating within a web container 692 on a web server 615) to: register a user for secure access to the tele-analysis platform; provide secure access to the tele-analysis platform interface; track the state of one or more analysis requests; retrieve the results of one or more analysis requests; and enable payment for one or more analysis requests. The eCommerce system includes service databases 700 (e.g. run on a database server 705) of user details 710, analysis results 720, and payments 730, as well as server-side applications 740 operating on one or more web servers 615 that enable customer operations 690 between the web pages 680 and the service databases 700. Figure 5 shows in total the preferred embodiment of the invention as a tele-health system, including the internet- accessible tele-analysis platform 600 and a customer web portal 670.

From the above description, it will be apparent to those skilled in the relevant arts that the present procedure, platform, and system has numerous advantages of prior art procedures, platforms and systems for the analysis of medical images. Segregation of a tool's image analysis activities into defined pre-, core-, and post-processing functions, and the subsequent implementation of a tool by reference to its list of component functions, enables tools to share common functions without duplicating equivalent code for each tool. In addition, where multiple tools are to be used in an image analysis task, the common functions need only be executed once on the same input data through comparison of each tool's function list. Further, the identification of different functions acting on the same input data from each tool's function list enables those functions to be executed in parallel. The segregation of a tools image analysis activities into pre-, core-, and post-processing functions also allows new tools to be readily integrated into the existing toolset by making use of preexisting functions, and only coding up those functions for the tool that are not yet available within the toolset. As most tools share common pre- and postprocessing functions, the functions for the new tool that will need to be coded are thus then typically its core-processing functions, which distinguish the tool from other tools. The segregation also facilitates automation of a tool's activities, by enabling automation activities to be focussed on clearly defined functions. Ultimately, the procedure of segregating each tool's image analysis activities into pre-, core-, and post-processing functions, in addition to the sub-procedure for comparing each tool's function list for the once only execution of common functions acting on common input data, gives the described tele-analysis platform and tele-health system flexible scalability for the deployment of multiple image analysis tools, without duplication of common activities between tools.

It will be apparent to those skilled in the relevant arts that numerous modifications and variations may be made without departing from the basic inventive concepts. For example, the sub-procedure applied to the pre-, core-, and post-processing stages of analysis could be equally applied to the entire image analysis task without reference to pre-, core-, and post-processing stages, but simply by reference to each tools entire function list. The sub-procedure itself may also be modified such that not all functions in an image analysis step need to have finished execution before continuing on to the next functions. This may be achieved by maintaining one input data list for all of the steps, rather than one for each step, which is constantly updated with new input data, and for which already processed input data is deleted, to ensure that the input data list is empty when all functions have been executed, so that the sub-procedure can terminate. It will also be apparent that the procedure, platform and system may be coded in a programming language other than Java.

All such modifications and variations that would be obvious to a person of ordinary skill in the art are deemed to be within the scope of the present invention.

Claims

CLAIMS:

1 . A procedure to perform an image analysis task for the analysis of one or more regions of interest (ROI(s)) on at least one medical image by two or more image analysis tools, without the duplication of common image analysis activities between tools, where a function list of each tool is compared with a function list of each other tool to determine if there is at least one common image analysis function between tools that operates on common input data, such that the common image analysis function need then be only executed once during the image analysis task by a single instantiation of that function that acts on the common input data.

2. A procedure according to claim 1 , wherein the image analysis activities include one or more of a pre-processing function, post-processing function or preprocessing and post processing functions.

3. The procedure according to claim 2, including:

for each tool, the image analysis activities are segregated and sequenced into at least one said pre-processing function, post-processing function or preprocessing and post processing functions, whereby:

the pre-processing function involves the removal of medical image scanner dependencies and artefacts from the ROI(s) or image(s) to provide either or both: improved image quality within the ROI(s) or image(s),

the post-processing function provides for the enhanced visualisation or interpretation of the results from pre-processing activities or optional core- processing activities with either or both

multi-dimensional representation of the results,

quantitative information on the results.

4. The procedure according to claim 2 or 3, where the image analysis activities further including the segregation and sequencing of at least one core- processing function between the pre- and post-processing functions providing for transformation of the pre-processed ROI(s) or image(s) to give either or both: the delineation of structure within the pre-processed ROI(s) or image(s), and

quantitative information on the medical image scanner signal behaviour captured within the pre-processed ROI(s) or image(s).

5. The procedure according to any one of claims 2 to 4, including segregation of image analysis activities into one or more pre-processing functions, one or more core-processing functions, and one or more post-processing functions, enabling a variety of the tools to be deployed achieving differences in their operation primarily through differences in their core-processing functions, whilst optionally sharing any number of pre- and post-processing functions.

6. The procedure according to any one of claims 2 to 5, whereby segregation of activities of an image analysis tool into a sequence of any of the preprocessing, core-processing, and post-processing functions further facilitates automation of the activities of the image analysis tool.

7. The procedure according to any one of claims 2 to 6, whereby the comparison of function lists for multiple tools enables different functions acting on common input data to be identified and then executed in parallel.

8. The procedure according to any one of claims 2 to 7 whereby:

the output of a said pre-processing function is input into another preprocessing function, core-processing function, or post-processing function,

the output of a said core-processing function is input into another core- processing function or post-processing function, and

the output of a said post-processing function is input into another postprocessing function.

9. The procedure according to any one of claims 2 to 8, whereby one or more core-processing activities is combined for the quantification of specific parameters that characterise the medical image scanner signal behaviour captured within the pre-processed ROI(s) or image(s).

10. The procedure according to any one of the preceding claims, including, for an image analysis task employing multiple tools, any said at least one preprocessing function is executed during a defined pre-processing stage of the analysis, any said at least one core-processing function is executed during a defined core-processing stage of the analysis, and any said at least one postprocessing function is executed during a defined post-processing stage of the analysis.

1 1 . The procedure according to any one of the preceding claims, including, for any said analysis stage, the first function to be executed for each tool in the stage is executed during a defined first processing step for the stage, the second possible function to be executed for each tool in the stage is executed during a defined second processing step for the stage, and so on.

12. The procedure according to claim 1 1 , including, for each analysis stage, an input data list created for each processing step that is comprised of a number of input data items to be processed by one or more of the functions to be executed in the analysis step, where an input data item is one or more of:

for the first processing step;

of the pre-processing stage, at least any one of:

• a medical image,

• a ROI defining a part of a medical image,

• a medical image that has been pre-processed or not pre-processed,

• a series of medical images that have been pre-processed or not pre-processed, • a series of ROIs defining parts of medical images that have been pre-processed or not pre-processed,

• and possibly any number of values derived from pre-processing if pre-processing has been performed;

of the post-processing stage, at least any one of:

13. The procedure according to claim 10 or 1 1 , where for any said analysis stage, the parallel execution of functions for any said processing step, without duplicating function execution for common functions between tools acting on a common input data item, is achieved by a sub-procedure applied to the stage that for each input data item on the input data list:

identifies the functions to be executed on the input data item from each tool's function list;

14. The procedure according to any one of claims 1 1 to 13, where for each analysis step in any given analysis stage, the sub-procedure executes the functions from different input data items' function lists in parallel.

15. The procedure according to any one of claims 2 to 14, wherein said preprocessing includes one or more of the following activities such that said core- processing to determine quantitative information on the medical image scanner signal behaviour provides consistent results across images taken from any make and model of scanner for the same imaging modality:

a) Modelling the signal behaviour of the background noise in the image(s),

b) Subtraction of any background signal level offset from the image(s), c) Reduction of any scanner-induced artefacts within the image(s), d) Reduction of any subject-induced artefacts within the image(s), e) Modelling any non-uniform signal penetration throughout the image(s) owing to the scanner and/or subject, and

f) Rescaling the signal intensities within the image(s) relative to a reference signal within the image(s).

16. The procedure according to any one of claims 2 to 15, where a core- processing activity for the delineation of structure within the ROI(s) or image(s) identifies an organ or tissue.

17. The procedure according to any one of claims 2 to 16, where a core- processing activity for the quantification of medical image scanner signal behaviour within the ROI(s) or images(s) is directly or indirectly used to quantify pathological or physiological change.

18. The procedure according to any one of the preceding claims, where the image analysis task is specified at a separate location from where the analysis by one or more image analysis tools is performed, thus requiring the electronic communication of the image analysis task over a telecommunications network, the communication including: a) an analysis request, consisting of one or more regions of interest on one or more medical images and possibly any number of the images' acquisition settings, for analysis by one or more specified image analysis tools, and

b) either or both of:

i) an automated analysis reply, consisting of the results of the analysis request, and optionally a record of the received analysis request, returned to the location at where the image analysis task was specified, ii) a separate analysis retrieval, consisting of the results of the analysis request, and optionally a record of the received analysis request, retrieved from a analysis database and returned to the location at which the retrieval was initiated,

where optional inclusion of a record of the received analysis request is used to verify that there has been no corruption of the ROI(s) or image(s) via their electronic communication.

19. An analysis platform including means to conduct any of the procedures of claims 1 to 18 through one or more computers.

20. A tele-analysis platform including means to conduct the procedure of claim 18 through a distributed client-server computer network, where:

one or more system administration utilities operate on one or more servers to:

sequence the execution of the one or more analysis requests,

21 . The tele-analysis platform of claim 20, wherein the distributed client-server computer network is connected through the Internet and:

the client-side application for the construction and transmittal of one or more analysis requests is a web-derived application,

the one or more server-side applications to conduct the automated analysis by one or more image analysis tools operate one or more application servers, and

the one or more system administration utilities operate on one or more web servers.

22. The tele-analysis platform of claim 21 incorporating Java technologies, where:

a) the client-side web-derived application for the construction and transmittal of one or more analysis requests is implemented as either a Java Applet or Java WebStart Application,

b) the one or more server-side applications to conduct the automated analysis by one or more image analysis tools are implemented through Java Beans or Enterprise Java Beans deployed in separate packages of preprocessing functions, core-processing functions, and post-processing functions, to promote scalability and reusability of tool components,

c) the one or more system administration utilities are implemented through Java Beans or Enterprise Java Beans with Persistence Entities to record the state of each image analysis task.

23. A tele-health system including the tele-analysis platform of claims 20 to 22 in combination with a web portal enabling pay-per-use access to the one or more image analysis tools, where the web portal includes:

register a user for secure access to the tele-analysis platform,

provide secure access to the tele-analysis platform interface,

track the state of one or more analysis requests,

retrieve the results of one or more analysis requests, enable payment for one or more analysis requests; and