BioImagingUK BIS Capital Infrastructure Submission: Biological and Medical Imaging

From BioImagingUKWiki

(Difference between revisions)
Jump to: navigation, search
(Q12. Are the major international projects identified in the consultation the right priorities for this scale of investment at the international level? Are there other opportunities for UK involvement in major global collaborations?)
 
(5 intermediate revisions not shown)
Line 1: Line 1:
== Introduction ==
== Introduction ==
-
This is an ongoing collaborative process to submit a response to the BIS consultation from the Medical Imaging and the Light and Electron Microscopy communities as decided at the June 11th BioImagingUK [[LRI_LondonMtgAgenda|Meeting]], at LRI London.  
+
This is BioImagingUK's response to the Department for Business, Innovation & Skills ''[http://www.gov.uk/government/consultations/science-and-research-proposals-for-long-term-capital-investment Open consultation on Science and research: proposals for long-term capital investment]''.
-
'''NOTE: This is currently a DRAFT document and subject to change by members of the BioImagingUK Community. The document is being rapidly updated-- see the History tab aboveIf you want to contribute to this process, see the link for the BioImagingUK Mailing List on the [http://bioimaginguk.org home page].'''
+
The document has been prepared on the BioImagingUK Network's Wiki (http://bioimaginguk.org).  Contributions from across the biological and medical imaging domains have been receivedAuthor contributions are listed after each question.
 +
 
 +
The final text, submitted to BIS on 4 July 2014, is available at http://bioimaginguk.org/index.php/BIS_Submission_Biological_and_Medical_Imaging
== Q4. What balance should we strike between meeting capital requirements at the individual research project and institution level, relative to the need for large-scale investments at national and international levels? ==
== Q4. What balance should we strike between meeting capital requirements at the individual research project and institution level, relative to the need for large-scale investments at national and international levels? ==
Line 174: Line 176:
In the following, we present three UK-based open source imaging software projects, which are examples of imaging in plants, cells and tissues and humans to illustrate the power and importance of these projects.
In the following, we present three UK-based open source imaging software projects, which are examples of imaging in plants, cells and tissues and humans to illustrate the power and importance of these projects.
-
''Plant Phenotyping''.
+
* ''Plant Phenotyping''. At the University of Nottingham’s Centre for Plant Integrative Biology (CPIB; http://www.cpib.ac.uk/), we have developed a number of imaging platforms and associated image analysis tools for plant phenotyping. Interest in automatic image analysis has increased significantly within the plant sciences in recent years. This is due to the emergence of the systems approach to biological research and an increasing awareness that quantitative measurement of the phenotype has fallen behind understanding of the genotype. The ability to link genotype to phenotype is now seen as the major bottleneck facing efforts to ensure global food security. Our work and software tools address this key problem, providing the quantitative data on plant structure and function required by plant and crop scientists and breeders. <br /> The bulk of software development at CPIB has been concerned with root phenotyping, with tools created for the recovery of quantitative data from images at the cellular scale (e.g. Cellset, http://www.cellset.net), organ scale (e.g. Roottrace, http://www.roottrace.net) and whole plant scale (e.g. RooTrak, http://www.rootrak.net). Each software resource has an associated imaging platform, the most notable being the Hounsfield Facility for rhizosphere research. This contains a set of three X-ray micro-computed tomography scanners and sample automation which, along with RooTrak will allow high-throughput phenotyping of plant roots grown in soil. More recent work has addressed the detection of growth events in image sequences of seedling germination, and the development of multi-view stereo methods capable of providing rich 3D descriptions of the structure of plant shoots. In addition to the tools already developed, we have the capability to address new plant image analysis problems, and develop high quality software tools for their solution.  Our software is a set of standlone tools, written in JAVA or C family languages.   
-
At the University of Nottingham’s Centre for Plant Integrative Biology (CPIB; http://www.cpib.ac.uk/), we have developed a number of imaging platforms and associated image analysis tools for plant phenotyping. Interest in automatic image analysis has increased significantly within the plant sciences in recent years. This is due to the emergence of the systems approach to biological research and an increasing awareness that quantitative measurement of the phenotype has fallen behind understanding of the genotype. The ability to link genotype to phenotype is now seen as the major bottleneck facing efforts to ensure global food security. Our work and software tools address this key problem, providing the quantitative data on plant structure and function required by plant and crop scientists and breeders.
+
-
 
+
-
The bulk of software development at CPIB has been concerned with root phenotyping, with tools created for the recovery of quantitative data from images at the cellular scale (e.g. Cellset, http://www.cellset.net), organ scale (e.g. Roottrace, http://www.roottrace.net) and whole plant scale (e.g. RooTrak, www.rootrak.net). Each software resource has an associated imaging platform, the most notable being the Hounsfield Facility for rhizosphere research. This contains a set of three X-ray micro-computed tomography scanners and sample automation which, along with RooTrak will allow high-throughput phenotyping of plant roots grown in soil. More recent work has addressed the detection of growth events in image sequences of seedling germination, and the development of multi-view stereo methods capable of providing rich 3D descriptions of the structure of plant shoots. In addition to the tools already developed, we have the capability to address new plant image analysis problems, and develop high quality software tools for their solution.  Our software is a set of standlone tools, written in JAVA or C family languages.   
+
-
 
+
-
''Image Data Management, Sharing and Publication''.
+
-
The University of Dundee is the home of the Open Microscopy Environment (OME; http://openmicroscopy.org), and consortium of software development teams  based in UK, France, Germany, Italy, and USA) that builds and releases tools for accessing, sharing large, heterogeneous, multi-dimensional image datasets.  The project has grown from a small academic collaboration to an international consortium, with worldwide usage and recognition.  OME’s tools-- OME-TIFF, Bio-Formats, and OMERO—are used at academic research institutions, most biotechnology and pharmaceutical companies, and several commercial imaging vendors, and by many open source and commercial projects.
+
-
 
+
-
Several on-line image repositories use OMERO and Bio-Formats to store and publish large multi-dimensional image datasets.  Examples include the JCB DataViewer (http://jcb-dataviewer.rupress.org), the EMDataBank (http://www.emdatabank.org), the  Stowers Digital Data Repository (http://srdr.stowers.org), the ASCB Cell Image Library (http://cellimagelibrary.org, and the Harvard Medical School LINCS Resource, http://lincs.hms.harvard.edu). OME’s technology can also support next generation image data repositories like that proposed by Euro-BioImaging and ELIXIR (see Q8).
+
-
''Centre for BioMedical Imaging''.
+
* ''Image Data Management, Sharing and Publication''. The University of Dundee is the home of the Open Microscopy Environment (OME; http://openmicroscopy.org), a consortium of software development teams  based in UK, France, Germany, Italy, and USA) that builds and releases tools for accessing, sharing large, heterogeneous, multi-dimensional image datasets.  The project has grown from a small academic collaboration to an international consortium, with worldwide usage and recognition. OME’s tools-- OME-TIFF, Bio-Formats, and OMERO—are used at academic research institutions, most biotechnology and pharmaceutical companies, and several commercial imaging vendors, and by many open source and commercial projects. <br />Several on-line image repositories use OMERO and Bio-Formats to store and publish large multi-dimensional image datasets.  Examples include the JCB DataViewer (http://jcb-dataviewer.rupress.org), the EMDataBank (http://www.emdatabank.org), the  Stowers Digital Data Repository (http://srdr.stowers.org), the ASCB Cell Image Library (http://cellimagelibrary.org, and the Harvard Medical School LINCS Resource, http://lincs.hms.harvard.edu). OME’s technology can also support next generation image data repositories like that proposed by Euro-BioImaging and ELIXIR (see Q8).  
-
The Centre for Medical Image Computing at University College London (CMIC; http://cmic.cs.ucl.ac.uk/) combines excellence in medical imaging sciences with innovative computational methodology. Our research finds application in biomedical research and in healthcare with a particular emphasis on translation of new computational methods in imaging sciences to the clinic. The research of the group focuses on detailed structural and functional analysis in neurosciences, imaging to guide interventions, image analysis in drug discovery, imaging in cardiology and imaging in oncology.
+
-
CMIC is developing several software tools that will be of interest to the wider medical imaging community (http://cmic.cs.ucl.ac.uk/home/software/).  All of our software is available as open source through Sourceforge.  NifTK is the name of our platform, combining NiftyReg, NiftySim, NiftyRec and NiftySeg via the viewer NiftyView.  The aim is to take the tools that CMIC develops, and apply them in a clinical context.  In addition, CMIC has developed TOAST, a Matlab-based software suite for image reconstruction in optical diffusion tomography, and Camino, an object-oriented software package for analysis and reconstruction of Diffusion MRI data, tractography and connectivity mapping.
+
* ''Centre for BioMedical Imaging''. The Centre for Medical Image Computing at University College London (CMIC; http://cmic.cs.ucl.ac.uk/) combines excellence in medical imaging sciences with innovative computational methodology. Our research finds application in biomedical research and in healthcare with a particular emphasis on translation of new computational methods in imaging sciences to the clinic. The research of the group focuses on detailed structural and functional analysis in neurosciences, imaging to guide interventions, image analysis in drug discovery, imaging in cardiology and imaging in oncology. <br /> CMIC is developing several software tools that will be of interest to the wider medical imaging community (http://cmic.cs.ucl.ac.uk/home/software/).  All of our software is available as open source through Sourceforge.  NifTK is the name of our platform, combining NiftyReg, NiftySim, NiftyRec and NiftySeg via the viewer NiftyView.  The aim is to take the tools that CMIC develops, and apply them in a clinical context.  In addition, CMIC has developed TOAST, a Matlab-based software suite for image reconstruction in optical diffusion tomography, and Camino, an object-oriented software package for analysis and reconstruction of Diffusion MRI data, tractography and connectivity mapping.
'''Software Development as Capital Investment'''
'''Software Development as Capital Investment'''
-
The development and maintenance of software is most often performed by trained software developers and thus the major cost of software development involves staff salaries. However, these costs should be considered to be “capital” as  they synergise with the use of capital investment in data storage, backup and delivery.  All “Big Data” has to be processed, analysed and delivered to scientists and the wider public using applications (“servers”) running software built for these tasks.  All of these applications must be “maintained”—kept current with updates to operating systems, security updates, and technologies for data access and visualization (e.g., web browsers). Meeting the infrastructure challenge of Big data and imaging for the life and biomedical sciences requires investments for the scientific community—and in particular, in open source software.
+
The development and maintenance of software is most often performed by trained software developers and thus the major cost of software development involves staff salaries. However, these costs should be considered to be “capital” as  they synergise with the use of capital investment in data storage, backup and delivery.  All “Big Data” has to be processed, analysed and delivered to scientists and the wider public using applications (“servers”) running software built for these tasks.  All of these applications must be “maintained”—kept current with updates to operating systems, security updates, and technologies for data access and visualization (e.g., web browsers). Meeting the infrastructure challenge of Big Data and imaging for the life and biomedical sciences requires investments for the scientific community— and in particular, in open source software.
Authors: [http://www.lifesci.dundee.ac.uk/people/jason-swedlow Jason Swedlow], [http://cmic.cs.ucl.ac.uk/staff/dave_hawkes/ Dave Hawkes], [http://www.nottingham.ac.uk/computerscience/people/tony.pridmore Tony Pridmore]
Authors: [http://www.lifesci.dundee.ac.uk/people/jason-swedlow Jason Swedlow], [http://cmic.cs.ucl.ac.uk/staff/dave_hawkes/ Dave Hawkes], [http://www.nottingham.ac.uk/computerscience/people/tony.pridmore Tony Pridmore]

Current revision as of 13:55, 4 July 2014

Contents

Introduction

This is BioImagingUK's response to the Department for Business, Innovation & Skills Open consultation on Science and research: proposals for long-term capital investment.

The document has been prepared on the BioImagingUK Network's Wiki (http://bioimaginguk.org). Contributions from across the biological and medical imaging domains have been received. Author contributions are listed after each question.

The final text, submitted to BIS on 4 July 2014, is available at http://bioimaginguk.org/index.php/BIS_Submission_Biological_and_Medical_Imaging

Q4. What balance should we strike between meeting capital requirements at the individual research project and institution level, relative to the need for large-scale investments at national and international levels?

Our world class research environment is underpinned by funding for the capital requirements of individual research projects and institutions. To complement this, strategic decision making at the national and international level is often required to coordinate investments in the national interest. This consultation seeks views on how to balance these complementary needs.

BioImagingUK (http://bioimaginguk.org) was founded in 2009 to address this exact question. BioImagingUK was formed as a grass roots consortium of scientists who depend on and develop imaging technologies for research in the life sciences, medical research and clinical practice. Its goal is to define priorities for delivering imaging infrastructure across all levels of the scientific landscape in the United Kingdom. BioImagingUK aims to:

  • drive the development of new imaging technologies
  • ensure that there is shared access to and specialized expertise available for these technologies
  • build the career models for the critical technical personnel that help to develop and run these technologies
  • ensure the development of new training programmes to support and enable these technologies and to develop the next generation of scientists who will use these technologies;
  • articulate the funding requirements and priorities to enable all of these activities to be built, delivered and maintained.

Indeed through several years of collaboration and strategic prioritization we have established that there is no single definition for the correct distribution of resources across the various scientific entities in the UK—this will differ depending on the scientific domain and technology. Here’s a concrete, community-derived example:in 2012 with funding from the Wellcome Trust, BBSRC, MRC and EPSRC we held the first national meeting of UK biological and medical imaging scientists who came together for two days to define the priorities for the development and delivery of imaging infrastructure for life sciences and biomedical research. This community (perhaps surprisingly, as the implications for funding and delivery of imaging technology mean significant change from the current status quo) unanimously agreed on the principles of the coordinated delivery of infrastructure at the institutional, regional and national level. A detailed analysis of these priorities and conclusions were published in a publically-written meeting report (http://bioimaginguk.org/images/0/04/BioImagingUK_Meeting_Summary_v5.pdf). Groups of imaging scientists representing many different imaging domains and applications,from Electron Microscopy, Light Microscopy, Pre-Clinical/Model System Imaging, Medical Imaging, Data and Software Tools, independently concluded that future investment in BioImaging infrastructure should be targeted at three different levels: National Facilities, Centres of Excellence and Department/Institute level facilities. The definitions, and most critically, the suggested technologies that should be delivered at each of these level was agreed by the community and published (http://www.bioimaginguk.org/images/e/ee/BioImagingUK_Technology_Priorities_v4.pdf). BioImagingUK aims to renew this list in early 2015.

How should the UK balance the commitments to these different entities? As noted in the consultation, there are always competing interests for funding, and never enough resources to fuel all the possible projects and direction developed in the dynamic, thriving environment of UK science. Moreover, there are several national and international projects that should also be supported, as they directly benefit and drive UK science. In addition, BioImagingUK believes it is essential to include substantial investment to upgrade or replace aging institutional capital infrastructure that serves as the foundation for much of the ongoing research in the life, biomedical and clinical sciences. Of the illustrative scenarios presented in paragraph 40 of the consultation, Scenario 1 and 2 will probably deliver the most impact and best value, but it is essential to include substantial investment in aging institutional capital infrastructure that must be renewed and/or replaced, while maintaining substantial contributions for Centres of Excellence, National Facilities and critical international projects that will benefit UK science.

Given inevitable competition for these funds, all research infrastructure must be maintained at an advanced, world-competitive level. In particular, National Facilities and Centres of Excellence are sites providing access to advanced, world-class technology critical for cutting edge scientific research. These sites should be identified, funded and reviewed using the established, proven, peer-review mechanisms that have built UK science. The BioImagingUK Strategy Meeting highlighted the need for periodic review of these resources where utility, impact and strategic importance are evaluated. No Facility or Centre should be a permanent institution and, as technology matures, diffuses or is superseded, Facilities and Centres would either evolve or be wound down in favour of new sites providing the next generation of imaging technology.

Finally, the BioImagingUK community has demonstrated an effective, community-based method for defining technologies that should be delivered via Centres of Excellence and National Facilities, namely the assembly of the community into open meetings, and open drafting of recommendations for funding priorities. There is no question that the established, powerful and productive peer review system used in UK science should be the final arbiter and source of funding, but open community recommendations, based on a broad sampling of the scientific priorities and goals of the community can serve as a powerful tool for applicants, reviewers, and funders. They may reveal consensus or even discord, but the rationale and community priorities will be known to all, and can contribute to strategic decisions.

Two current examples of these types of efforts:

  • Medical Imaging: A good current example (July 2014) of inter-institutional cooperation is the coordination of bids to the MRC infrastructure calls for medical imaging, centred around cross-institutional multi-facility cooperation and UK-wide networking agreements.
  • Biological Imaging: the national coordination of the Euro-BioImaging Node Expression of Interest applications in 2013 has created a network of facilities offering a broad, complementary range of imaging technologies for the UK scientific community [[1]].

Authors: Dave Clarke, Jason Swedlow, Dave Hawkes, Xavier Golay

Q5 How can we maximise collaboration, equipment sharing, and access to industry to ensure we make the most of this investment?

In BioImagingUK this is the question that we have been trying to answer and address since 2009. In mid-2014 we have had significant success in this direction so we address this question not from the view of a wish list but indeed from the view of members of our Network with real experience in running a wide variety of different types of capabilities and facilities that provide access to the academic and industrial research sectors.

The first step towards achieving collaboration and sharing of technology and expertise is the development of an open community where capabilities are discussed, promoted, shared and prioritized. In BioImagingUK we use a series of open public meetings, an open Wiki (http://bioimaginguk.org) and an open mailing list to define our strategies and to discuss these priorities. While a relatively simple kind of activity we have shown that it can be enormously powerful. This approach produced the strategic prioritization described above, and indeed, this consultation response. Most recently this community has matured into the BioImagingUK Network funded by a combination of the BBSRC, MRC, EPSRC, and the Wellcome Trust. Only with this open community-minded approach can such a broad set of scientific domains, requirements and funding agencies be addressed.

The funding and the sustainability of capital investment are two issues which must be considered together. We favour a joint funding approach in which large equipment is placed in shared facilities and sustained through fees for use charged to research grants. In this funding model there are two issues which must be considered together. The first is to build sufficient capacity in shared facilities to meet the research needs of academia and industry. The second is development of a research funding policy which encourages the use of shared equipment. RCUK can choose to preferentially fund large equipment either in association with specific research programs or in the form of shared facilities. We support the latter approach. If there is sufficient capacity in equipment and staff, shared resources can support any level of collaboration, equipment sharing, and access to industry desired by BIS.

In addition, research institutions can help to encourage collaborative and sharing endeavours by ensuring that a career structure is in place that appropriately recognises and rewards all contributions to ‘team science’ endeavours.

We strongly endorse the concept that critical research infrastructures cannot be delivered physically to every research institution in the UK. There are not the resources nor the trained expert staff available. Thus the creation of a tiered, distributed infrastructure where facilities and expertise build alongside world-class scientific excellence and made available to the community is the overriding theme that defines our approach.

These concepts exist not only within BioImagingUK but within Euro-BioImaging the European analogue to BioImagingUK. Euro-BioImaging seeks to deliver a pan-European imaging infrastructure that will support the scientific needs and aspirations of all EU scientists. As such there is natural collaboration and coincidence between the goals of BioImagingUK and Euro-BioImaging, and many of the concepts described below derive from Euro-BioImaging.

Several major themes for ensuring maximal collaboration:

  1. There is no better way to promote equipment sharing than to place it in a shared facility. Expert staff within shared facilities increase the number of researchers who can make use of advanced equipment by lowering the barrier to access: even less experienced researchers can make use of advanced equipment with the help of expert facility staff.
  2. Access to infrastructure must be fully funded, so all relevant expenses —running and staff costs, maintenance, travel etc. are accounted for. In other words, all costs incurred by the user in preparing for and executing the imaging experiment, and all costs incurred by the resource, including all costs associated with instrumentation, personnel, and lab resources must be covered. In general, the user cannot directly bear the full costs of using a resource, so models that include full funding for the user and/or the resource must be considered. Perhaps a move towards establishing indicative charge rates would help harmonise approaches, accepting that different technologies and/or communities will require different approaches. Such an approach is being floated by one of Germany’s major funders (DFG).
  3. A resource must provide a “complete” solution, including advice on preparation of the sample, tissue, organism or subject to be imaged, the imaging process itself, and perhaps most importantly tools for processing and analyzing the data, so that each user can potentially achieve a publishable result. In the case of medical imaging this advice must also include help and assistance in preparing appropriate ethics and NHS R&D approvals, or coordinated help in the establishment of proper governance and Home Office approvals for small animal imaging. As most imaging resources routinely collect many GB to TB of data, delivering maximal value from these resources demands that imaging infrastructure resources interface and collaborate with “big data” resources, for storage, sharing, analysis and visualization. Again patient confidentiality and other regulatory issues must be addressed within the resource with the aim that all subjects’ data is available, suitably anonymised for other researchers to use. As noted in Q10, this represents a major opportunity for synergy and collaboration between investments in “big data” and investments in medical and biological imaging.
  4. Work to remove the inter-HEI VAT restrictions, ideally at a national level and not just through local Cost Sharing Groups. Ideally all instrument sharing between HEIs should be seen as collaborative research and not subject to VAT.
  5. Access for industry: Shared facilities which rely on fee-for-use to achieve sustainability are equally happy to support any researcher who has the means to pay for access. An excellent example for promoting industry access to academic facilities is the Australian Victoria Platform Technologies Network (http://www.platformtechnologies.org/).

As noted above, BioimagingUK actively seeks and can demonstrate the power of infrastructure sharing and collaboration between scientists, Centres and National Facilities. However, our primary goal is to support and deliver the world’s best science. We do not believe sharing and collaboration is an end in itself. For example, while we are actively building collaboration between the Biological Imaging and Medical/Clinical imaging communities, there are significant differences in the way the communities have developed, currently practise their research and how their funding is provided. While collaboration is strongly encouraged and synergy maximised, these different ways of working must be acknowledged and supported by whatever funding initiatives emerge in the future.

Authors: Jason Swedlow, Dave Hawkes, Xavier Golay, Peter O’Toole, Kurt Anderson

Q6. What factors should we consider when determining the research capital requirement of the higher education estate?

Higher Education Institutions (HEIs) are the engine of many of the current research discoveries. Despite the success and investment in many research institutes, much of the research output in the UK comes from research based at HEIs. Indeed a critical outcome of research based at HEIs is the unparalleled opportunity and access for training for the undergraduates and postgraduates who will be the imaging researchers of tomorrow. Using researchers as course lecturers and exposing undergraduate and postgraduate students to research in world competitive labs, using the best technology available is a strong driver for making university training exciting, in particular for the STEM subjects, and likely to be followed in a longer career. By linking the proximity of world-class research with HEIs there is an enormous capability and benefit for training and future discovery.

By no means does this statement require that all major research investments be placed at HEIs. Again the BioImagingUK community has repeatedly concluded that an environment of multiple different types of institutions can be used to provide the underlying research infrastructure. Indeed a diversity of such facilities as National Resource and Centres of Excellence has resulted in the world class research in both biological and medical imaging within the UK.

When prioritizing funding for infrastructure at HEIs, several key factors should be considered.

  1. The central location and integration of core facilities with the main science labs in any new buildings. Orphan facilities, whose capabilities are not constantly presented with new scientific challenges will struggle to deliver the highest performing, advanced technology.
  2. The potential to promote cross-disciplinary collaboration and synergy.
  3. The provision of a range of technologies from mid-range “workhorse” technologies and expertise, to cutting edge, ground-breaking resources. Much of the UK’s scientific excellence and performance depends on access to mid-range workhorse equipment (e.g., in Biological Imaging, laser scanning confocal microscopes and 120kV transmission electron microscopes; in Medical Imaging, 3T MR whole body scanners). These technologies often support substantial of current scientific research and serve as great opportunities to train undergraduate and postgraduate students and prepare them for productive use of advanced, next-generation systems.
  4. Aging, older generation systems that must be upgraded or replaced. Many of these workhorse systems, in both biological and medical imaging, were purchased >10 years ago. Older designs and performance specs are now significantly superseded by newer instruments that are faster, more sensitive and enable more complex experiments and questions to be addressed. MOST IMPORTANTLY however, these older systems are beyond their service life, with the original vendors unable to maintain them. This means the wide range of scientific projects that depend on these workhorse instruments are stymied or even completely blocked. As just one example, BioImagingUK’s catalogue of UK imaging facilities lists at least 63 aging laser scanning confocal microscopes (July 2014; http://www.york.ac.uk/biology/technology-facility/imaging-cytometry/uk-lm-facilities/), the vast majority of which are housed in imaging facilities based in HEIs. There are many more that are not in established core facilities, but placed in departments or individual research labs (an informal survey of three major manufacturers of these systems suggests that >400 aging systems are installed in the UK and are candidates for upgrade or replacement). The systems cost between £150k and £500k each, with maintenance contracts around £10k and £30k per year. A major investment in these systems is critical to simply maintain the current research output in basic biological, biomedical and preclinical research.
  5. The provision of key staff who develop, run, and maintain these technologies and provide training for undergraduate, postgraduate, and postdoctoral scientists. As discussed below (Q10) delivering new technology infrastructure into HEIs without simultaneous funding of staff technologists guarantees the creation of “white elephant”, inefficiently used, poorly maintained systems that do not deliver sufficient return on the original capital investment.

Looking forward to next the 6 years, many HEIs are evolving a hybrid between conventional imaging core-facilities based on commercial instrumentation and in-house developments of hardware and software to provide functionality that is customized or beyond the state of the art. While this has long happened in specific laboratories dedicated to technology development (usually for their own use), we increasingly see the aspiration to provide custom modifications or novel capabilities to a wider scientific community, and the community’s desire to access these advances. Currently, it is difficult to fund access to these technologies, so flexibility in capital infrastructure funding will be very important.

Authors: Peter O’Toole, Andrew Hibbert, Jason Swedlow, Dave Hawkes, Paul French,

Q7. Should - subject to state aids and other considerations - science and research capital be extended to Research and Technology Organisations and Independent Research Organisations when there are wider benefits for doing so?

There is no question that the power and impact of the research community in the UK comes from a broad ecology of the different types of the research institutions, although it is anticipated that the majority of facilities will be based in major centres of biomedical research, and in particular the Academic Health Science Centres. RTOs and RIOs play an important in this ecology. In particular they allow specific themes and capabilities to be targeted and developed. Good examples are the RCUK-funded resources at Pirbright, John Innes Centre, etc. On a larger scale, national resources like the Francis Crick Institute will require significant investment. These should be included as targets for infrastructure investments but in concert with investments in the HEI estate. As mentioned before there will no be specific formula that defines the division of investments and infrastructure in research organisations and HEIs. As we have seen within BioImagingUK, the most effective way to define these priorities is to use the knowledge of the community in an open and public discussion to define overall priorities, executed through the word-class peer review funding system that has been built in the UK.

Authors: Jason Swedlow, Dave Hawkes, Paul French

Q8. What should be the UK's priorities for large scale capital investments in the national interest, including where appropriate collaborating in international projects?

The impressive strength and breadth of the UK research base means that we are presented with a huge range of potential investment opportunities. Demand inevitably outstrips funding. Therefore, there is a constant need to prioritise, and this consultation seeks your views to inform our approach. These strategic judgements require us to look first at what international competitors are investing in, and identifying where it is in the UK national interest to collaborate in international infrastructure projects. This may involve significant contributions to projects around the world or hosting them in the UK. We are seeking views on which of the important projects laid out in this consultation (pages 54-58) should be the highest priority. We are also welcoming suggestions of new potential high priority projects not identified here.

BioImagingUK sees several major opportunities for investment that align quite closely with the identified priorities in the Capital Equipment Consultation. Across the UK, we must develop coordinated, distributed Bio-Imaging infrastructure in which critical technologies are placed either at institutions, Centres for Excellence or National Facilities, depending on their overall expense, requirements for technical expertise and most importantly the wider UK community needs access to these technologies. Our concepts of this wider ecology have been articulated repeatedly by members of the community and indeed with many of the members of both our Research Council and non-governmental charity-based funders.

BioImagingUK sees several specific examples as high priorities over the next 3-5 years. In the following sections we provide examples specific to either Biological or medical Imaging, and then one final example, Big data in Imaging, which is common to both. Where appropriate, we explicitly link ongoing or future UK priorities with opportunities at the international scale, in particular Euro-BioImaging and ELIXIR.:

For Biological Imaging:

  1. a major priority should be placed on capabilities for super-resolution as these technologies become more common, but also more diverse and focussed on specific types of problems or applications;
  2. several emerging technologies for high throughput imaging where a larger number of samples can be rapidly imaged rapidly to attain quantitative, systems-level analysis of organism;
  3. emerging technologies that enable high resolution imaging not only of single cells but of tissues or whole animals. Combining high throughout automation with whole tissue or animal imaging can provide new models for understanding human development and disease.

As part of the Euro-BioImaging’s First Open Call for Nodes (http://www.eurobioimaging.eu/sites/default/files/Euro-BioImaging%201st%20Open%20Call%20for%20Nodes%20-%20Summary%20290713.pdf), eight UK imaging facilities submitted expressions of interest (“EoIs”) for consideration and evaluation by Euro-BioImaging’s Independent Review Board. Each proposed Node presented its technology, business model, and proposed several projects that required access to the its technology and expertise. The proposed technologies covered a range of super-resolution modalities, high throughput biological imaging and image processing and analysis. Seven of the eight proposals were recommended for funding and construction with the eighth requiring minor corrections—a very strong performance. These EoIs provided concrete examples of the diverse, world-class imaging technologies that are available in the UK and, critically, the demand for access to these technologies by the scientific community. The Euro-BioImaging EoIs have been submitted by BioImagingUK to BBSRC on their request as “Model Nodes”, examples of imaging technologies and expertise that are needed by the community and can be delivered, if funded appropriately.

For Medical Imaging:

  1. Image guided surgical and interventional technologies, in particular real-time imaging, therapy monitoring and control. An example is the new national proton therapy facilities being set up at Manchester (Christie) and London (UCL). There is a need for National or Regional Centres of Excellence in these technologies as a focus for large scale development and testing. Such a resource in other countries, most notably Germany and USA, has provided a focus for industrial investment and generation of successful spin-outs in Medical Technology.
  2. Development of multi-physics imaging technologies such as combined optical/MRI clinical scanners and ultrasound/MRI clinical scanners, building on the success of the combined PET/MRI facilities. These need to be located with the physics and computational expertise for development co-located with the chemistry expertise developing new probes and the biomedical expertise developing biomedical applications. These also need to be collocated with investment in novel probes, including the infrastructure to take probe development through to pre-clinical and first-in-man studies.

An example of this type of resource is MRC’s recent award of infrastructure investment in medical bioinformatics. One award is for a partnership between Queen Mary University of London, UCL, London School of Hygiene and Tropical Medicine, the Francis Crick Institute, Wellcome Trust Sanger Institute and the European Bioinformatics Institute, to improve capability, capacity, training and capital infrastructure in medical bioinformatics. Imaging forms a component of this award.

Big Data in Imaging

Another major priority for BioImagingUK is the development of ’big data’ resources designed for imaging applications. Imaging by its very nature generates large datasets and indeed new modes of automation make it routine for imaging scientists to generate many GBs to TB of data in a single experiment. There are now several biological imaging systems in production in the UK, that routinely collect several TB each week. Developing tools for handling these critical datasets is an absolute necessity. Without these solutions these instruments will rapidly become white elephants as they collect datasets that are simply too large or complex to understand or to interpret. Several kinds of technology must be developed. Among these:

  • Data compression or reduction tools, to make the collected datasets more manageable while retaining sufficient resolution and sensitivity.
  • Analysis, processing and visualization tools
  • Data management platforms, for sharing, storing and processing larger image datasets.
  • Standardised interfaces for plugins, modules and functions

Moreover the development of centralized repositories where specific types of image data ("reference images") can be deposited and made publicly accessible is an important next step for the community. These datasets can be linked to other existing on-line repositories (e.g., public genomic resources) to linking of genomic with functional phenotypic data. Moreover these reference data will serve as benchmark datasets that will drive the development of new analytic and processing capabilities by algorithm developers and other data scientists. As the size of datasets and repositories grow, it will be impractical to expect reference data to be downloaded for further processing and analysis. Thus public repositories must also provide compute resources linked to references datasets, using "virtual" or "cloud" compute where developers and users can test and validate new tools before using them on their own data.

In delivering these capabilities the UK must consider close and active participation with Euro-BioImaging (http://www.eurobioimaging.eu/) and ELIXIR (http://www.elixir-europe.org/). Both of these EU-scale projects aim to develop infrastructures that map quite closely to the needs of the UK life and biomedical sciences. There is now a unique opportunity to collaborate with Euro-BioImaging and ELIXIR to develop "reference" image data repositories, with the publication of a joint MoU by these projects (http://www.eurobioimaging.eu/content-news/euro-bioimaging-elixir-image-data-strategy). Given the scale of the infrastructures that must be built as well as the fact that these resources and the technology that run them are being developed by an international community, the UK should support ELIXIR and Euro-BioImaging's efforts to deliver image data repositories for use by the scientific community. The UK has already made major strategic investments into the UK-based European Bioinformatics Institute (http://www.ebi.ac.uk) which serves as the ELIXIR Hub, and is therefore well-positioned to lead this strategic effort.

Looking forward, the integration of physiological computational modelling with imaging at multiple scales, including integration with the international Virtual Physiological Human project and its successors, will require continuous strategic funding in the future. This idea is taken up in more detail in Q10.

Authors: Jason Swedlow, Dave Hawkes, Paul Verkade, Xavier Golay, Kurt Anderson

Q9. What should the criteria for prioritizing projects look like?

Those engaged in the research process with a known track record for conducting high quality research would be best placed to develop the criteria for assessing the prioritization of projects, in terms of feasibility, value for money, scientific excellent and contribution to the community. Currently scientific peer review is used to determine funding for responsive mode funding calls for research projects and this mechanism could continue for project prioritization.

The projects should be assessed on evidence based criteria, e.g. track record of producing excellent results in previous projects, knowledge in the subject area and appropriate support for the project in terms of intellectual and technical resources. There should be a well thought through project plan with clear aims and developmental milestones. There should be a benefit to UK society either economically (e.g. through industry), technologically (increasing the intellectual capital), or societal (for Health and social well being).

The project ideally would be in line with current community initiatives or would be of benefit to them, e.g. invention of a new technique that would increase efficiency, knowledge, sharing of resources It should show how it makes use of investments / resources which have been previously made and what the outcomes of the investment would be: e.g. training of staff in a particular technology, support of a resource which many researchers use, development of a technology, use of an existing technology in a new way to generate new information which is of benefit to health, industry etc).

Authors: Ann Wheeler, Lucy Collinson, Paul French

Q10. Are there new potential high priority projects which are not identified in this document?

In BioImagingUK, we recognise the excellent work that funding from BIS / RCUK has done with regard to development institutional resources, centres of excellence, national facilities specialist training and careers for technology. We hope that looking forwards this will continue as it is essential to the continued world leading achievements of UK science.

One thing that is notably absent is the intersection and thus synergy of many of the proposed investments. In particular for both biological and medical imaging there is an obvious and critical opportunity to leverage investments in imaging with investments in big data. Imaging in the 21st century is by its very nature a big data science and desperately needs solutions to achieve and reach its potential impact. Big data is often discussed as an independent science of its own. Indeed the methodology technology development certainly is but the analytics management and processing of large datasets is intimately linked to the generation of those datasets in particular the metadata and the metadata that defines the binary data and the annotations and the linkages which makes that data rich and useful as possible.

From the viewpoint of BioImagingUK we see a strong and potentially world-changing opportunity to link the investments in big data with the investments of imaging to connect or "correlate" many different image types. Just one example:

  • an MR structural or metabolic (using new hyperpolarisation technology) image of a prostate gland within a patient;
  • the light microscopy image of the prostate gland after a biopsy;
  • the histopathology image of the biopsy from a slide scanner using conventional stains;
  • the fluorescence light microscopy image of cells from that biopsy stained with a range of molecular markers that characterize the molecular basis of that disease;
  • the super-resolution and electron microscopy images of a few of the cells in this biopsy which detail the structural changes in cells subject to specific molecular changes in a specific patient’s prostate cancer;
  • the linkage of these images with related images from cell- or animal-based models of prostate cancer.

Currently each of these imaging modalities is technically possible but the linkage between the different types of data across these multiple scales can only be imagined. A major ‘big data’ goal for imaging would be the linkage of these different imaging modalities in a way so that both clinicians and researchers could access these data sets and explore the basic mechanisms that underline disease and ultimately define the diagnosis and prognosis for a human patient. This multi scale access is a major defining vision for the scientists who use imaging and can be the basis of collaboration across multiple infrastructure modalities and imaging technologies in the United Kingdom.

Taking this concept one step further-- across the life and biomedical sciences the need to link multiple distinct datasets, e.g. genomic proteomic and phenotypic data is critical. This need exists in basic research where for example underlying mechanisms of disease or pathology are explored and continues al l the way up to modern veterinary and clinical medicine. The data collected in each of these domains demands “big data” solutions—linking them together is a wholly unmet, but scientifically and medically critical capability. Looking forward, the latest GWAS studies are mainly limited today by the poor quality of the phenotyping of the populations, and application of advanced biological and medical imaging technologies might help fill this gap.

To create and maintain these world class resources we must invest in training and career progression of those highly skilled professionals who will run, maintain and develop these resources. The training and retention through appropriate career progression of these people (engineers, computer scientists, physicists, chemist, biologists etc) is vital for the success of these facilities but often neglected. Indeed the full potential of these resources will not be achieved without investment in the staff to run them. The solution to this challenge involves a strategic, national commitment to funding these positons along with the development of career tracks within HEIs and research organisations. The scope of this challenge is large, but perhaps one of the most critical to actively address. Continuing to prioritise capital infrastructure investment without concomitant support for skilled technology professionals ensures that UK scientific infrastructure will underperform and not serve its critical roles in supporting research, training or societal benefit. In short we will waste our capital and social investments in science. We must not allow this to happen.

Authors: Ann Wheeler, Lucy Collinson, Pippa Hawes, Jemima Burden, Jason Swedlow, Paul French, Dave Hawkes, Peter O’Toole, Xavier Golay, Kurt I. Anderson, David Gadian

Q11. Should we maintain a proportion of unallocated capital funding to respond to emerging priorities in the second half of this decade?

It is often impossible to predict which directions or technologies will develop, and at what rate. The successful scientific research requires a landscape of both long term commitments to enable building significant research projects, programmes and institutes, but also a mechanism to rapidly redeploy some of its funding streams to take advantage of emerging fields. Of particular interest in this regard are fields such as imaging, genomics, proteomics and computational approaches which are all advancing at a rapid pace and will enable significant new research to approaches in the next 5 years. It is significant that the UK is a leader in all these fields with a publication record second only the US, and world-leading research happening in all of these fields.

The recent MRC Next Generation Optical Microscopy call is a case in point. This call was initially posted as funded at a level of £18M, based on the extreme interest and strength of the applications the initiative was eventually funded at £25M, a significant increase in funding. This initiative has funded a number of developments in super-resolution microscopy (SRM), e.g. Oxford NanO, Leeds, STFC/Harwell, as well as providing funding for providing new production super-resolution instruments for the community in Dundee and Edinburgh. As valuable as these investments are, they are based on technology that is evolving very quickly, often on timescales of 2-3 years. In the case of SRM, even the first round of commercial systems are on their way to obsolescence since new products offer much better performance, often at a lower price (e.g. STED microscopy).

Another example is work on in-vivo model-based imaging of tissue microstructure, distinguishing between tissue states based on modelling and classification using technologies such as diffusion MRI (recent award by EPSRC of the high gradient MR system at Cardiff), shear-wave elastography and photoacoustics. These all have a pathway to translation that will develop over the next 5-10 years.

Overall, very rapid and widespread investment seems to be to be inadvisable. It is desirable to have the ability to upgrade and sustain imaging technology and to expect demand for next generation technology every 2-3 years, e.g., two or three times over the period covered by this consultation.

Authors: Ian Dobbie, Jason Swedlow, Dave Hawkes, Paul French

Q12. Are the major international projects identified in the consultation the right priorities for this scale of investment at the international level? Are there other opportunities for UK involvement in major global collaborations?

As described above, Euro-BioImaging and ELIXIR are two international research infrastructures that will bring great value to UK life and biomedical sciences. These projects will make available world-class resources for imaging and data in the life and biomedical sciences.

The consultation document highlights a number of critical capital projects in its annexes. For example, “Big Data” and imaging projects in the life and biomedical sciences are prominently mentioned, but links and synergies between them are not considered. Big Data solutions cannot be constructed divorced from the data acquisition or even the scientific questions that these large data projects seek to deliver. In the experience of BioImagingUK, software development to support scientific discovery is most successful when it is associated with world-class science. These collaborations can occur by basing development software development teams in the same facilities as research labs. However, software tools that are released as open source resources can reach a much broader community, including users and developers anywhere in the world. Indeed, there are several UK open source software projects that are built though broad international collaborations and support worldwide user communities. These communities can extend across the academic and commercial sectors, so open source software, built through international collaborations, can have international scientific and commercial impact.

Examples of UK-based Open Source Software Projects for Biological and Biomedical Imaging

In the following, we present three UK-based open source imaging software projects, which are examples of imaging in plants, cells and tissues and humans to illustrate the power and importance of these projects.

  • Plant Phenotyping. At the University of Nottingham’s Centre for Plant Integrative Biology (CPIB; http://www.cpib.ac.uk/), we have developed a number of imaging platforms and associated image analysis tools for plant phenotyping. Interest in automatic image analysis has increased significantly within the plant sciences in recent years. This is due to the emergence of the systems approach to biological research and an increasing awareness that quantitative measurement of the phenotype has fallen behind understanding of the genotype. The ability to link genotype to phenotype is now seen as the major bottleneck facing efforts to ensure global food security. Our work and software tools address this key problem, providing the quantitative data on plant structure and function required by plant and crop scientists and breeders.
    The bulk of software development at CPIB has been concerned with root phenotyping, with tools created for the recovery of quantitative data from images at the cellular scale (e.g. Cellset, http://www.cellset.net), organ scale (e.g. Roottrace, http://www.roottrace.net) and whole plant scale (e.g. RooTrak, http://www.rootrak.net). Each software resource has an associated imaging platform, the most notable being the Hounsfield Facility for rhizosphere research. This contains a set of three X-ray micro-computed tomography scanners and sample automation which, along with RooTrak will allow high-throughput phenotyping of plant roots grown in soil. More recent work has addressed the detection of growth events in image sequences of seedling germination, and the development of multi-view stereo methods capable of providing rich 3D descriptions of the structure of plant shoots. In addition to the tools already developed, we have the capability to address new plant image analysis problems, and develop high quality software tools for their solution. Our software is a set of standlone tools, written in JAVA or C family languages.
  • Image Data Management, Sharing and Publication. The University of Dundee is the home of the Open Microscopy Environment (OME; http://openmicroscopy.org), a consortium of software development teams based in UK, France, Germany, Italy, and USA) that builds and releases tools for accessing, sharing large, heterogeneous, multi-dimensional image datasets. The project has grown from a small academic collaboration to an international consortium, with worldwide usage and recognition. OME’s tools-- OME-TIFF, Bio-Formats, and OMERO—are used at academic research institutions, most biotechnology and pharmaceutical companies, and several commercial imaging vendors, and by many open source and commercial projects.
    Several on-line image repositories use OMERO and Bio-Formats to store and publish large multi-dimensional image datasets. Examples include the JCB DataViewer (http://jcb-dataviewer.rupress.org), the EMDataBank (http://www.emdatabank.org), the Stowers Digital Data Repository (http://srdr.stowers.org), the ASCB Cell Image Library (http://cellimagelibrary.org, and the Harvard Medical School LINCS Resource, http://lincs.hms.harvard.edu). OME’s technology can also support next generation image data repositories like that proposed by Euro-BioImaging and ELIXIR (see Q8).
  • Centre for BioMedical Imaging. The Centre for Medical Image Computing at University College London (CMIC; http://cmic.cs.ucl.ac.uk/) combines excellence in medical imaging sciences with innovative computational methodology. Our research finds application in biomedical research and in healthcare with a particular emphasis on translation of new computational methods in imaging sciences to the clinic. The research of the group focuses on detailed structural and functional analysis in neurosciences, imaging to guide interventions, image analysis in drug discovery, imaging in cardiology and imaging in oncology.
    CMIC is developing several software tools that will be of interest to the wider medical imaging community (http://cmic.cs.ucl.ac.uk/home/software/). All of our software is available as open source through Sourceforge. NifTK is the name of our platform, combining NiftyReg, NiftySim, NiftyRec and NiftySeg via the viewer NiftyView. The aim is to take the tools that CMIC develops, and apply them in a clinical context. In addition, CMIC has developed TOAST, a Matlab-based software suite for image reconstruction in optical diffusion tomography, and Camino, an object-oriented software package for analysis and reconstruction of Diffusion MRI data, tractography and connectivity mapping.

Software Development as Capital Investment

The development and maintenance of software is most often performed by trained software developers and thus the major cost of software development involves staff salaries. However, these costs should be considered to be “capital” as they synergise with the use of capital investment in data storage, backup and delivery. All “Big Data” has to be processed, analysed and delivered to scientists and the wider public using applications (“servers”) running software built for these tasks. All of these applications must be “maintained”—kept current with updates to operating systems, security updates, and technologies for data access and visualization (e.g., web browsers). Meeting the infrastructure challenge of Big Data and imaging for the life and biomedical sciences requires investments for the scientific community— and in particular, in open source software.

Authors: Jason Swedlow, Dave Hawkes, Tony Pridmore

Personal tools