Edinburgh Data-Intensive Research - Biological/Medical Image Processing

MoSGrid - Molecular Simulations in a Distributed Environment

Sandra.Gesing — Thu, 27 Jun 2013 11:33:02 +0000

Speaker(s):

Presentation Type:

talk

Molecular simulations are indispensable methods in areas like material science, structural biology, and drug design. These methods address data-intensive and compute-intensive problems, which demand high-performance computing to allow data analysis in an acceptable time. The project MoSGrid (Molecular Simulation Grid) offers a workflow-enabled grid portal allowing access to molecular simulation tools on distributed resources in an intuitive manner. Users are able to exchange workflows and data via repositories and, thus, to exchange knowledge about the specific application domain. The talk will give an overview on the current MoSGrid portal, the underlying infrastructure, and an outlook on the next steps.

Attachment	Size
osdc2013MoSGridSandraGesing.pdf	4.77 MB

Date and time:

Tuesday, 18 June, 2013 - 12:00

Location:

OSDC PIRE Workshop 2013, Edinburgh, UK

Web site:

OSDC PIRE Workshop

Research topics:

Biological/Medical Image Processing

Collaborative environments

Data-intensive computing

Intuitive interfaces

The BraINS databank

David.Rodriguez — Wed, 12 Dec 2012 14:55:39 +0000

NeSC Research Seminar Series

Speaker:

Dominic Job & David Dickie

Brain Images of Normal Subjects (BRAINS) bank and atlases are being developed with >1000 normal subjects from across the lifespan, to be expanded in the future to include subjects with disease. The images have been collected in centres across Scotland and are in a range of magnetic resonance (MR) sequences, including T1, T2, T2*, and fluid attenuated inversion recovery (FLAIR). When BRAINS is released these will be searchable by a wide range of metadata, e.g. blood pressure<140/90; age=85; MMSE>26.
BRAINS atlases are based on calculated distributions of brain structure rather than parametric estimates. These will be used to support image analysis research and clinical reporting of brain images.

Attachment	Size
dir_dickieJob.ppt	13.01 MB

Date and time:

Friday, 14 December, 2012 - 10:00

Length:

45 minutes

Location:

Turing Room (5.42), Informatics Forum

Projects:

SINAPSE

Research topics:

Biological/Medical Image Processing

Automatic Lesion Area Detection Using Source Image Correlation Coefficient for CT Perfusion Imaging

Fan.Zhu — Fri, 16 Nov 2012 15:18:50 +0000

Speaker(s):

Fan.Zhu

Presentation Type:

poster

Computer tomography (CT) perfusion imaging is widely used to calculate brain hemodynamic quantities such as Cerebral Blood Flow (CBF), Cerebral Blood Volume (CBV) and Mean Transit Time (MTT) that aid the diagnosis of acute stroke. Since perfusion source images contain more information than hemodynamic maps, good utilisation of the source images can lead to better understanding than the hemodynamic maps alone. Correlation-coefficient tests are used in our approach to measure the similarity between healthy tissue time-concentration curves and unknown curves. This information is then used to differentiate penumbra and dead tissues from healthy tissues. The goal of the segmentation is to fully utilize information in the perfusion source images. Our method directly identifies suspected abnormal areas from perfusion source images and then delivers a suggested segmentation of healthy, penumbra and dead tissue. This approach is designed to handle CT perfusion images, but it can also be used to detect lesion areas in MR perfusion images.

Attachment	Size
Fan_poster.pdf	1.49 MB

Date and time:

Tuesday, 6 November, 2012 - 16:00

Location:

SICSA DEMOfest 2012, Edinburgh, UK

Web site:

Research topics:

Biological/Medical Image Processing

DECIPHER Health

David.Rodriguez — Fri, 16 Nov 2012 14:39:38 +0000

There is increasing demand to develop a more personalised approach to diagnosis and treatment regimes for patients, such as those with cancer, so that treatment offered is based on the knowledge that it will be effective. The current “one size fits all” approach should not be applied to care and treatment when the tools that are now available can target the individual.

DECIPHER Health is a unique electronic data handling system that to enable this personalised approach by delivering sophisticated management and analysis of vast amounts of clinical, patient, imaging and genomic data, and subsequently providing a comprehensive picture of the genetic and clinical factors which affect a patient’s outcomes.

DECIPHER was born out of pilot work started in late 2010 to build a breast cancer informatics demonstrator system with the University of Dundee, a project which was funded by Cancer Research UK. Over the next two years DECIPHER Health will expand on this initial work to create an anonymised cancer Research Safe Haven and two clinical applications, encompassing consented patient cohorts from Lothian and Tayside with genetic profiling across six cancer types: breast, skin, prostate, ovarian, colorectal and lung.

DECIPHER Health is part funded by the Technology Strategy Board through a competition under the Stratified Medicine Innovation Platform umbrella, an initiative that oversees an investment of over £50 million of government funding in innovative research and development.

Acronym:

DECIPHER

Web site:

DECIPHER Health Aridhia

Dates:

1 January 2012 to 1 January 2014

Project members:

David.Rodriguez

Partners:

Aridhia

NHS Lothian

Research topics:

Biological/Medical Image Processing

Project Status:

In progress

Congratulations to Fan Zhu

Malcolm.Atkinson — Fri, 19 Oct 2012 16:28:41 +0000

I am delighted to report that Fan Zhu's external examiners were pleased with his thesis and his spirited viva defense. They have recommended a pass with a few minor corrections. A big thank you to Professors Gary Green and Richard Baldock.

Title:
Brain Perfusion Imaging - Performance and Accuracy

Abstract:
Brain perfusion weighted images acquired using dynamic contrast studies have an important clinical role in acute stroke diagnosis and treatment decisions.
The purpose of my PhD research is to develop novel methodologies for improving the efficiency and quality of brain perfusion-imaging analysis so that clinical decisions can be made more accurately and in shorter time.
This thesis consists of three parts:

1. My research investigates the possibilities that parallel computing brings to make perfusion-imaging analysis faster in order to deliver results that are used in stroke diagnosis earlier. Brain perfusion analysis using local Arterial Input Functions (AIF) technique takes a long time to execute due to its heavy computational load. As time is vitally important in the case of acute stroke, reducing analysis time and therefore diagnosis time can reduce the number of brain cells damaged and improve the chances for patient recovery. We present the implementation of a deconvolution algorithm for brain perfusion quantification on GPGPU (General Purpose computing on Graphics Processing Units) using the CUDA programming model. Our method aims to accelerate the process without any quality loss.

2. Specific features of perfusion source images are also used to reduce noise impact, which consequently improves the accuracy of hemodynamic maps. The majority of existing approaches for denoising CT images are optimized for 3D (spatial) information, including spatial decimation (spatially weighted mean filters) and techniques based on wavelet and curvelet transforms. However, perfusion imaging data is 4D as it also contains temporal information. Our approach using Gaussian process regression (GPR) makes use of the temporal information in the perfusion source imges to reduce the noise level. Over the entire image, our noise reduction method based on Gaussian process regression gains a 99% contrast-to-noise ratio improvement over the raw image and also improves the quality of hemodynamic maps, allowing a better identification of edges and detailed information. At the level of individual voxels, GPR provides a stable baseline, helps identify key parameters from tissue time-concentration curves and reduces the oscillations in the curves. Furthermore, the results shows that GPR is superior to the alternative techniques compared in this study.

3. My research also explores automatic segmentation of perfusion images into potentially healthy areas and lesion areas which can be used as additional information that assists in clinical diagnosis. Since perfusion source images contain more information than hemodynamic maps, good utilisation of source images leads to better understanding than the hemodynamic maps alone. Correlation coefficient tests are used to measure the similarities between the expected tissue time-concentration curves (from (reference tissue)) and the measured time-concentration curves (from target tissue). This information is then used to distinguish tissues at risk and dead tissues from healthy tissues. A correlation coefficient based signal analysis method that directly spots suspected lesion areas from perfusion source images is presented. Our method delivers a clear automatic segmentation of healthy tissue, tissue at risk and dead tissue. From our segmentation maps, it is easier to identify lesion boundaries than using traditional hemodynamic maps.

Topic of this submission:

News

Research topics:

Biological/Medical Image Processing

Projects:

SINAPSE

Brain Perfusion Imaging - Performance and Accuracy

Fan.Zhu — Fri, 12 Oct 2012 11:22:04 +0000

Speaker:

Fan Zhu

I aim to use computer science methodologies to solve brain perfusion imaging (medical imaging) problems in my PhD research.
I focus on three issues. The first issue addressed was the acceleration of the algorithms used to generate hemodynamic maps. It proved possible, using GPGPUs, to achieve substantial speedups while retaining the quality of generated hemodynamic maps. The next two investigations sought to make better use of the information latent in the time-domain of CT perfusion scans. The first of these showed that Gaussian process regression (GPR) could significantly enhance the quality of derived data at modest computational cost. The second of these then used that denoised data to improve the discrimination between healthy, penumbra and dead tissue.

Attachment	Size
201210Seminar.pdf	3.29 MB

Date and time:

Friday, 12 October, 2012 - 10:00

Length:

45 minutes

Location:

Turing Room (5.42), Informatics Forum

Research topics:

Biological/Medical Image Processing

Brain Perfusion Imaging - Performance and Accuracy

David.Rodriguez — Tue, 28 Aug 2012 14:54:37 +0000

NeSC Research Seminar Series

Speaker:

Fan Zhu

Date and time:

Friday, 12 October, 2012 - 10:30

Length:

90 minutes

Location:

Turing Room (5.42), Informatics Forum

Projects:

SINAPSE

Research topics:

Biological/Medical Image Processing

Intuitive Large-scale Image Processing for Biologists

Jano.van.Hemert — Mon, 31 May 2010 14:05:17 +0000

Modern cell and developmental biology and the now-established domain of systems biology use quantitative imaging methods to measure the location, dynamics and interaction of molecules in fixed and living cells, and at increasingly high spatial and temporal resolution. Quantitative imaging depends on the development, delivery, and use of sophisticated image processing and analysis algorithms. The availability of these data analysis tools is commonly cited as a major bottleneck in scientific discovery. Previously, the absence of common interfaces and defined standards for data structures hindered the sharing of new analysis methods and the open, shared access to image datasets. Moreover the sheer computational cost of running complex algorithms on large datasets demands access to compute facilities that, while existing, are not accessible via standardised, intuitive tools for most bench biologists.

This project combines developments in the OMERO application developed by the Open Microscopy Consortium led by Prof J Swedlow and the Rapid portlet development tool developed in Dr J I van Hemert's laboratory at the UK's National e-Science Centre. The resource generated will be a service with an intuitive user interface that enables bench biologists to access high performance computing resources for processing and analysing their multi-dimensional images of cells and tissues. We do not propose to develop a single stand-alone resource under this project but to provide a vital service for bench biologists, based on world-leading work performed in the UK that uses common, standardised interfaces and established principles in usability to provide access to cutting-edge image analysis methods for bench biologists. The resource will be released as a component of the open-source OMERO software suite that is currently either in testing or in daily use at most imaging sites in the UK and over 1200 sites worldwide. A stable version of Rapid will be bundled in these releases under the same license. We will build this service on top of the Edinburgh Compute Data Facility and the National Grid Service to provide the underlying e-Infrastructure.

Acronym:

Rapid-OMERO

Value:

£780956

Dates:

1 July 2010 to 31 December 2012

Partners:

Swedlow Lab, Wellcome Trust Centre for Gene Regulation & Expression, University of Dundee, UK

Projects:

Rapid-OMERO

Research topics:

Biological/Medical Image Processing