Interdisciplinary research success stories
We have pump-primed a number of project to date, all of which have lead to on-going multidisciplinary collaborations, new research streams, publications and grants. You can download a brochure summarising these projects on the left hand pane.
Personalised Research Dashboard (Laurissa Tokarchuk, Athen Ma, Nick Bryan-Kinns, Pat Healey, Heiko Grossmann). The aim of this project is to develop a Personalised Research Dashboard providing a dynamic self-updating environment in which researchers get most up-to-date highly-specialised information. The two main goals of the project are to develop self-organising network and group infrastructure, and the mechanism for personalised intelligent news feed by referrence to the network and group definition. This project is inter-disciplinary with maths and computer science. Has led to a First Grant Proposal by Tokarchuk on “Social and Temporal Recommendation”, supported by BT and Last.FM. Also internal ESD funding for research paper recommendation, and a collaboration with the Food, Consumer Behaviour and Health Research Group at the University of Surrey.
Contact: Laurissa Tokarchuk. Download project presentation here.
To view a short video presentation, click here.
Terahertz imaging for molecular dynamics. (Robert Donnan and Tina Chowdhury). A long-standing interest in the biological community is the need for automated, high throughput diagnostics, which rapidly characterises the interplay of physiological cellular protein dynamics in response to mechano-electrochemical forces with non-invasive dynamic control. Potential 
applications include detecting single base mutations in genetic sequences or monitoring gene sequences as efficient diagnostic devices. A major challenge will be the distinction of the DNA strands within a sample volume without the need for functionalised/labelled probes. This project applies Dispersive Fourier Transform Spectroscopy (DFTS) to biological ‘targets’. The project seeks to make the first DFTS measurements on appropriate proteins in order to correlate absorption spectra with functional behaviour of the protein. As part of the project a THz imaging apparatus will be developed to enable imaging of samples in liquid phase. Part of this work has been in collaboration with the National Physics Laboratory.
Contact: Rob Donnan. To view a short video presentation, click here.
Statistical physical analysis of the dynamics of foraging bumblebees. (Reiner Klages, Tom Ings, Lars Chittka). 
Foraging bumblebees need to visit hundreds or even thousands of flowers every day to collect food, nectar and pollen. But, predatory crab spiders may be hiding on some of these flowers waiting to catch unsuspecting bees. Fortunately for bees, most attacks by spiders fail giving bees the opportunity to learn to recognise and avoid these dangerous predator. However, some spiders can potentially make this task even more difficult by changing their colour to match that of the flower they are sitting on. Therefore, in this experiment we set out to test whether spider camouflage does in deed hinder bees' ability to learn to avoid these dangerous predators. This project is developing probabilistic models of bee foraging that it is foreseen will have wider applicability, e.g. in network science.
Contact: Tom Ings or Reiner Klages. Download presentation here.
To view a short video presentation, click here.
Accurate landmark localization and registration of 3D facial scans for the evaluation of orthodontic treatments in maxillofacial and oral surgery. (Andrea Cavallaro, Prathap Nair and Lifong Zou) This project is improving an image analysis tool developed in the SEECS for the localisation of facial landmarks on 3D scans. Automated localisation eases the burden on clinicians who analyse the data and more importantly improve the accuracy of scan comparison. The project is half-way through and our achievements so far are: improved the analysis tool to detect facial landmarks with better accuracy; made the tool more user-friendly with the inclusion of user feedback.
Our current work (second half of the project) involves clinical evaluation of the landmark detection accuracy. Finally, we will use the improved landmark detection process to improve the accuracy of 3D facial scan registration for the analysis of patient data. The figure shows a comparison of the proposed registration results with a baseline algorithm (ICP). Our registration provides results consistent with the morphological knowledge of the person under study, as the cheek region consists of soft tissue which has changed over time, while the upper regions of the face consist of hard tissue which is known to be invariant. A change is seen in the eye region as the eyes are closed in one scan, and open in the other.
Contact: Andrea Cavallaro. Download presentation here.
Coagulation and Communications.(Raul Mondragon and Karim Brohi). Trauma is one of the world's leading causes of death and disability. Some trauma patients develop a profound derangement of their ability to form blood clots, associated with a 4-fold increased likelihood of dying. The traditional description of the coagulation protein-protein interactions as a cascade model is incorrect. Now it is known that human coagulation network has typical properties of a complex network. The aim of this project is to apply complex networks science to analyse and describe the coagulation metabolic network and obtain a representation that captures the dynamics of haemostasis. Interdisciplinary with trauma surgeon – Prof. Karim Brohi.
Contact: Raul Mondragon.
Computational Flow Simulations in Cerebral Aneurysm Treatment. (Jens-Dominik Mueller) Aneurysms have a high rate of mortality in case they rupture. However there is no consensus about when to treat un-ruptured aneurysms. Recent studies have proposed that hemodynamic factors such a wall shear stress induced by the flow or intra-aneurysm pressure play a major role in aneurysm rupture. Recent interest has been focussed on the patient-specific approach in which the actual vessel geometry is extracted from a CT or MRI scan of a specific patient. This promises high accuracy and could be extended to clinical use on a patient-per-patient basis. The aim of this study is to extend research to a larger cohort of subjects in order to confirm the hypothesis that flow sensitivity can be linked to an increased likelihood of aneurysm rupture. The project will involve radiologists and neurologists at Barts who provide the patient data and medical expertise, as well as engineers at SEMS who perform the flow simulations.
Contact: Jens-Dominik Mueller.
Past success stories:
Electronic Engineering and Mathematics:
Recently another interdisciplinary project has been awarded by EPSRC: "Optimal design of performance measurement experiments for complex, large-scale networkx (DOENET)" (PI = John Schormans, QMUL and Andrew Moore, Cambridge. CI = Steven Gilmour and Jonathan Pitts).
-----------------------------------------
Interdisciplinary work on the application of mathematics of complexity to the behaviour of packet networks lead to 3 granted patents. Work on network modelling and analysis using non-linear dynamical (“chaotic”) models for traffic and networks has succeeded with long-term collaboration with Mathematical Sciences, being funded by EPSRC and industry in a series of closely related projects (e.g. the MANMADE project and EPSRC grant GR/T18615/01).
------------------------------------------
The Design Of Experiments (DOE) has been widely applied to research in the Biological Sciences and many aspects of Industrial Engineering. Prof Steven Gilmour heads The Statistics Group at Queen Mary, which has an international reputation for its work in DOE. When the networking research of John Schormans and Jonathan Pitts discovered serious limits to the accuracy of sampled packet level network measurements, it seemed a natural time to turn to statisticians for help.
In response, the college funded a joint PhD studentship for Ben Parker, in which the first supervisor is Steve (in Maths) and his second is John (in EE). Ben has an MA in Maths and a Diploma in Computer Science (both from Cambridge), and an MSc in Applied Statistics and Operational Research (from Birkbeck), and is an ideal researcher for this strongly interdisciplinary problem.
A key recent discovery in the field of network measurement by sampling is that traditional Poisson sampling is not always optimal; a more general class of distributions (gamma renewal processes) minimizes the mean-square error in the sampled estimator. This has left the optimal probing rate as the salient unsolved question. The unique aspect of our joint approach is that, for the first time, network measurements are treated as numerical experiments which can, at least in principle, be optimally designed to yield the most information from the fewest number of samples of network performance (packet loss and delay).
Ben’s work models a network buffer as a 2 state Markov Chain and has now demonstrated how to use this to analyze the overflow probability - by deducing the likelihood function, and hence the Fisher information matrix. On submitted a thesis Abstract to the ACM Student Thesis Panel, Ben’s work, “Optimal Active Probing Rate for Networks”, has been accepted for presentation at ACM SIGMETRICS 2008. The ACM received many excellent submissions and were able to accept only a small fraction of them.
Publications:
'Parker, B., Gilmour, S and Schormans, J.A (2008) Measurement of packet loss probability by optimal design of packet probing experiments. IET Communications. In print.