Machine Learning and Signal RepresentationUp one level
The following CMIV projects conducts research related to the Machine Learning and Signal Representation.
The aim of this medical image science project is to increase patient safety in terms of improved image quality and reduced exposure to ionizing radiation in CT. The means to achieve these goals is to develop and evaluate an efficient adaptive filtering (denoising/image enhancement) method that fully explores true 3D and 4D image acquisition modes. Four-dimensional (4D) medical image data are captured as a time sequence of image volumes. During 4D image acquisition, a 3D image of the patient is recorded at regular time intervals. The resulting data will consequently have three spatial dimensions and one temporal dimension. Increasing the dimensionality of the data impose a major increase the computational demands. The initial linear filtering which is the cornerstone in all adaptive image enhancement algorithms increase exponentially with the dimensionality. On the other hand the potential gain in Signal to Noise Ratio (SNR) also increase exponentially with the dimensionality. This means that the same gain in noise reduction that can be attained by performing the adaptive filtering in 3D as opposed to 2D can be expected to occur once more by moving from 3D to 4D.
Most signal processing tools, for feature extraction, image enhancement and visualization are still limited to 2D, while multi-dimensional imaging of the human body is clinical routine. The computational complexity significantly increases, when extending dimensionality beyond 2D. The need for efficient filtering on volumes and volume sequences is therefore increasing. This project focuses on efficient methods for local feature extraction and enhancement of multi-dimensional images. The fundamental approach is to find a methodology to achieve efficient implementations of filter networks for medical image processing of volumes and volume sequences.
This project aims to elucidate the pathophysiological mechanisms of visceral hypersensitivity in patients with Irritable Bowel Syndrome and compare the findings with those in healthy controls. Much time will be spent on method development.
A manifold is a mathematical concept which generalizes surfaces to higher dimensions. Examples of 2-dimensional manifolds are for instance the surface of a sphere and the the surface of a torus, both being examples of non-linear manifolds. Locally however, manifolds are flat and equivalent to the an Euclidean space. Features found in signals can often be described using manifolds. This is often not stated explicitly, but instead various parameterizations of manifolds are used. In order to describe a quantity which can be seen as a point on a sphere, spherical coordinates are commonly used. This has some drawbacks however which we wish to avoid if possible. We see a need for manifolds in the field of medical image analysis. Medical doctors express a wish to objectively quantify various features in medical images, such as local texture, shape and orientation of organs. We know from previous research that manifolds can do the job, but we lack a generic framework for dealing with manifold-valued signals in signal processing. In fact, we believe that such a framework will be useful in other areas signal processing too. The goal of this project is to explore a specific flavour of signal processing and continue the development of methods 1) to learn or identify manifold-valued representations from examples and 2) apply signal processing on manifold-valued signals which is analogous to filtering and interpolation using convolution operators in classic signal processing.
As the potentials for treating neurological disorders have increased tremendously the last decades, there is also a growing need for safe, reliable and cost-effective diagnostic tools. fMRI is valuable both for an improved description of normal brain function and for assessment of patients with neurological disorders. The core theoretical idea in the project is that by including/developing tools for reconstruction of the brains cortical surface new and highly signiﬁcant local spatial priors can be included in the fMRI data analysis and in this way signiﬁcantly improve detection performance.
Despite the enormous complexity of the human mind, fMRI techniques are able to partially observe the state of a brain in action. In this paper we describe an experimental setup for real-time fMRI in a bio-feedback loop. One of the main challenges in the project is to reach a detection speed, accuracy and spatial resolution necessary to attain sufficient bandwidth of communication to close the bio-feedback loop. To this end we have banked on our previous work on real-time filtering for fMRI and system identification, which has been tailored for use in the experiment setup. In the experiments presented the system is trained to estimate where a person in the MRI scanner is looking from signals derived from the visual cortex only. We have been able to demonstrate that the user can induce an action and perform simple tasks with her mind sensed using real-time fMRI.The technique may have several clinical applications, for instance to allow paralyzed and "locked in" people to communicate with the outside world. In the meanwhile, the need for improved fMRI performance and brain state detection poses a challenge to the signal processing community. We also expect that the setup will serve as an invaluable tool for neuro science research in general.
This is a proposal for long-term collaboration between Warwick, Norrkoping and Linköping. The goal is to gain a deeper understanding as to how humans perceive their multi-sensory world In particular we want to determine: 1. How much of a scene is actually perceived in any point in time 2. What precision of any sense is perceived at any point in time 3. How is this perception related to any individual 4. How can this understanding be used to create perceptually equivalent multi-sensory virtual environments and manipulate senses in order to influence a person’s perception for example pain perception. There are a number of steps to be undertaken. Many of these will evolve as the work progresses.
We will combine economic experiments with brain imaging to study the neural processes that govern economic decision-making in a broad sense. The first study in this project will focus on comparing the neural processes related to hypothetical and real purchase decisions. There is a sizeable literature in economics using hypothetical willingness to pay questions to value things like better environmental quality. These questions typically overestimates the willingness to pay, but the neural mechanisms behind this overestimation is unclear. The fMRI-paradigm builds on a study by Knutson et al (Neuron 2007) investigating the neural correlates of purchase decisions.
Image processing of medical image volumes (3D/4D data) requires a completely new approach compared to standard images (2D). Research on methods for high speed processing as well as for high image quality output is required. In a research project within the VINST programme a novel approach for solving the speed challenge was developed. Building on this “pre-study” project, this new project covers the remaining research for reaching a new clinical quality output level, while maintaining the speed. The goal is to have the research results from both projects verified in a prototype. The project is a cooperation between the company ContextVision AB and its research partner Center for Medical Image and Visualization (CMIV) at Linköping University.
SIMILAR - The European Taskforce Creating Human-Machine Interfaces Similar to Human-Human Communication - WP10 Medical Applications
* SIMILAR will create an integrated task force on multimodal interfaces that respond efficiently to speech, gestures, vision, haptics and direct brain connections by merging into a single research group excellent European laboratories in Human-Computer Interaction (HCI) and Signal Processing. * SIMILAR will develop a common theoretical framework for fusion and fission of multimodal information using the most advanced Signal Processing tools constrained by Human Computer Interaction rules. * SIMILAR will develop a network of usability test facilities and establish an assessment methodology. * SIMILAR will develop a common distributed software platform available for researchers and the public at large through www.openinterface.org. * SIMILAR will establish a scientific foundation which will manage an International Journal, Special Sessions in existing conferences, organise summer schools, interact with key European industrial partners and promote new research activities at the European level. * SIMILAR will address a series of great challenges in the field of edutainment, interfaces for disabled people and interfaces for medical applications. Natural immersive interfaces for education purposes and interfaces for environments where the user is unable to use his hands and a keyboard (like Surgical Operation Rooms, or cars) will be dealt with a stronger focus.