
|
Contents
The
World Map
The
ANNIMAB Society
Individual
Membership
Datasets
The
ANNIMAB-1 Conference
About
this Page |
The ANNIMAB Society
The Artificial Neural Networks in Medicine and Biology Society (ANNIMAB-S) is based at Göteborg University (GU) and is open for individual membership to anyone with an active interest in artificial neural networks. ANNIMAB-S is associated with several other Swedish groups working with biological or medical applications of neural networks.
One of its tasks is the coordination of educational and research activities about artificial neural networks; another one was the organisation of the ANNIMAB-1 conference in Göteborg, Saturday May 13, to Tuesday May 16, 2000.
|
Individual
membership
The ANNIMAB Society, ANNIMAB-S, is open for membership for everyone with an active interest in the use of artificial neural networks and related techniques in medicine and biology. So far, the Society is mainly a discussion forum, and the main channels for this discussion are our Web pages and the ANNIMAB Mailing List.
Membership is free and carries no obligations whatsoever. To become a member, just send us an e-mail with a short description of yourself and your background.
|
Datasets
Some
investigators have decided to make their data freely available,
so that different techniques can be compared. We have so far put
two datasets on the Web, both from Dr. Simon Cross at the University
of Sheffield:
Fine
Needle Aspirate of Breast Lesions
Endoscopic
Biopsy of Inflammatory Bowel Disease Dataset
See
also the GCS visualisation
toolbox (from the same author).
We
encourage other researchers to share their data with the world.
|
The ANNIMAB-1 Conference
The
first international conference on Artificial Neural Networks in
Medicine and Biology, ANNIMAB-1, took place in Göteborg on
May 13-16. The program and abstracts from the conference can be
found here.
Artificial
neural network (ANN) techniques are currently being used for many
data analysis and modelling tasks in clinical medicine as well
as in theoretical biology, and the possible applications of ANNs
in these fields are countless. The aim of the ANNIMAB-1 conference
was to summarise the state of the art, analyse the relations between
ANN techniques and other available methods, and point to possible
future biological and medical uses of ANNs. It had three main
themes:
- Medical
applications of artificial neural networks: for better diagnoses
and predictions from clinical and laboratory data, in the analysis
of ECG and EEG signals, in medical image analysis, for the handling
of medical databases, etc.
- Uses
of ANNs in biology outside clinical medicine: for example, in
models of ecology and evolution, for data analysis in molecular
biology and in models of animal and human nervous systems and
their capabilities.
-
Theoretical aspects: recent developments in learning algorithms,
ANNs in relation to expert systems and traditional statistical
procedures, possible roles of ANNs in the medical decision process,
etc.
ANNIMAB-1
featured forty-two presentations of submitted papers organised
into six oral sessions and a poster session. There were also nine
invited papers presented by highly renowned authorities in the
field, covering a broad range of issues in the theory of artificial
neural networks and their application to medicine and biology.
In
order to facilitate the participation of medical and biological
scientists with only limited knowledge of ANN techniques, but
also to introduce mathematicians and statisticians to the main
medical and biological applications of ANNs, the first half-day
of the conference offered a set of overview lectures, giving up-to-date
information about ongoing research in these fields. The full text
of the opening lecture can be found here
(pdf).
The
Proceedings of the ANNIMAB-1 conference are published in the series
Perspectives in Neural Computing (Springer-Verlag London)
and can be ordered separately.
ANNIMAB-1
was the first international conference having artificial neural
networks in medicine and biology as its exclusive topic. Participation
in the conference hence offered an unique opportunity to be informed
of recent developments in this rapidly expanding field.
|
About
this Page
Coming soon...
|
|
We
are collecting links to people and organisations working with medical
applications of artificial neural networks and related techniques.
|
|
Austria
Vienna University of Technology
Pattern Recognition and Image Processing Group
Active Appearance Models in Quantitative Musculo Skeletal Radiology
We propose a computer based method that performs the quantification by means of automated image analysis and pattern recognition. The goal is to fully automatically identify the bones of the hand/wrist and extract exact quantitative information about the extent of the erosions caused by rheumatoid arthritis based on a radiograph.
|
Canada
Carleton
University, Ottawa
Department
of Systems and Computer Engineering
Decision-Support Systems designed for Critical Care
A case-based reasoner tool has been developed, allowing users
to compare the ten-closest matching cases to the newest patient
admission, from a database of intensive care medical records.
A back-propagation, feed-forward artificial neural network has
been trained and tested to estimate patient outcomes: duration
of artificial ventilation, the length of stay , and mortality.
M. Frize,
H.C.E. Trigg, F.G. Solven, M. Stevenson. B.G. Nickerson.
|
Finland
Turku
University Central Hospital
Central
Laboratory
Artificial
Neural Networks and Fuzzy Logic in Medical Data Analysis
The aim of the project is to study non-linear techniques in medical
data analysis. The terms soft computing or computational intelligence
are used to cover non-linear techniques, such as neural networks,
fuzzy logic and neuro-fuzzy tools. Recently, it has been shown
that these techniques are very useful in pattern recognition from
images or curvilinear signals. In our project, we apply soft computing
to clinical problems, in which the model is expected to be non-linear
and, accordingly, conventional statistical techniques do not give
satisfactory results.
|
Germany
University
of Bielefeld, Faculty of Technology
Neuroinformatics
Group
Neural
Network Based Classification of Microscope Cell Images
In this project we develop a system for classification of cells
in fluorescence microscope images with a minimum of interaction
by a user. The work started in December 1998 in a close collaboration
of the work group Neuroimmunology and Molecular Pattern Recognition
at the University of Magdeburg. The aim is to develop and analyse
methods to score cells in certain multi-parameter images automatically.
|
Hungary
KFKI
Research Institute for Particle and Nuclear Physics, Budapest
The
Computational Neuroscience Group
Computational
approach to the functional organization of the hippocampus
Population activities as well as underlying single cell voltages
are simulated during normal and epileptiform activities in the
hippocampal CA3 slice. It is demonstrated that our model can reproduce
electrophysiological phenomena characteristic to both single cell
and population activities. Specifically, fully synchronized population
bursts, synchronized synaptic potentials, and low amplitude population
oscillation were obtained.
|
The Netherlands
Radboud University Nijmegen Nijmegen Institute for Cognition and Information
Goal-referenced control of verbal and nonverbal actions The project addresses the question: Given a goal and the motivation to reach it, what mechanism allows goal-directed responding? Goal-directed responding has perhaps been most intensively studied in the past by bringing people in situations where they had to resist temptation, thereby exploiting what is expressed by a famous quotation of Oscar Wilde (1893): "I can resist everything except temptation."
|
Sweden
Göteborg
University
ANNIMAB
Society
ANN
in information intensive health care
In the first phase of the project we will try to further knowledge
and competence concerning ANN and related analytical methods.
Later, these methods will be applied to selected medical problem
areas where the amount and kind of available data require unconventional
approaches. Examples are extraction of 3-D information from NMR
(nuclear magnetic resonance) images, prognostication in neurosurgical
intensive care, improved heart infarctation diagnosis with electrocardiography
(ECG) and methods to determine the cause of fever in patients
who are being treated for blood malignancies.
Department
of Computing Science, Umeå University
GeDeMeDeS
Lena
Kallin
A comparison of numerical risk computational techniques in screening
for Down's syndrome. In this paper we will compare three different
optimisation techniques in calculating the risk of a foetus having
Down's syndrome. Techniques described include statistical methods
(multiple Gaussian formula, (MGF)) and are compared to neural
network approaches using multi-layer perceptrons (MLP) and preprocessing
perceptrons (PPP).
|
UK
School
of Computing & Mathematical Sciences, Liverpool John Moores
University
Neural
Computing Group
Characterisation
of Nuclear Magnetic Resonance Spectra
The group offers expertise in statistics, neural computing and
mathematical modelling, with considerable experience in real-world
applications in the fields of medicine, finance, image processing
and process control. Personnel currently comprises four faculty
staff, a research assistant and five Ph.D. students, together
with visiting fellows. (Professor
P.J.G. Lisboa)
InferSpace
Richard
Dybowski
Decisions
and Discovery from Knowledge and Data
InferSpace has over 15-years experience in artificial intelligence,
data analysis, and information visualization. Our Mission is to
help you or your clients make better decisions from knowledge
and data through the use of state-of-the-art methods for analysis,
modelling and visualization.
|
USA
Univeristy
of Chicago
Department
of Radiology, Kurt Rossmann Laboratories
Neural
Network Approach for Differential Diagnosis of Interstitial Lung
Diseases
A neural network approach was applied for the differential diagnosis
of interstitial lung diseases, using both clinical and radiographic
information.
Colorado
State University
Computer
Science Department
Charles
W. Anderson
Neural network learning algorithms with applications to control
and signal processing problems. Current research topics include
efficient learning in hidden units, reinforcement learning, control
of alternative-fuel engines, recognition of EEG patterns, control
of HVAC systems, and parallel algorithms for reinforcement learning
(contains many papers).
Duke
University Medical Center, Department of Radiology
Duke Advanced Imaging Laboratories
Digital
Chest Imaging
We have a wide range of research projects in every
aspect of digital chest imaging. These include the use of Bayesian
statistical estimation to compensate for scattered radiation
in chest radiographs, as well as to improve the appearance
of solitary pulmonary nodules for lung cancer diagnosis. Several
computer-aided diagnosis projects are underway, including the
use of artificial neural network predictive models to predict
acute pulmonary embolism based on findings from the chest radiograph
and ventilation/perfusion scans.
Massachusetts
The
Lahey Clinic
Jeffery
E Arle (mail)
I am a Neurosurgeon at The Lahey Clinic in Massachusetts. Recent
publications: Neural
Network Analysis of Preoperative Variables and Outcome
from Epilepsy Surgery, Arle JE, Devinsky O, Perrine K,
and Doyle W, J Neurosurg, 90:998-1004, 1999. Posterior
Fossa Tumor Prediction in Children Using MR Image Properties,
Spectroscopy, and Neural Networks, Arle JE, Morriss C, Zimmerman
RA, Phillips P, and Sutton LN, J Neurosurg, 86:9-15, 1997
Children's
Hospital San Diego Research Center
Laboratory
for Research on the Neuroscience of Autism
Matthew
Belmonte
Two-layer and three-layer feed-forward artificial neural networks
were trained to predict behvioural performance from single-trial
EEG in autistic and normal subjects in a task involving response
to rare stimuli and shifting of attention between vision and audition.
Eyeblink artefacts were removed from the data using a frequency-domain
filter. Performances of the networks on separate test sets varied
across subjects but were usually at least 80%. |
|