Schlüsselwörter
(Englisch)
|
Unusual/abnormal event detection; telepathology; microscopic pathological images; image indexing and retrieval; feature extraction; classification; JPEG 2000; JPSearch; AIC; DICOM.
|
Forschungsprogramme
(Englisch)
|
COST-Action IC0604 - Anatomic Telepathology Network
|
Kurzbeschreibung
(Englisch)
|
The project aims at elaborating solutions for the automation of unusual events (e.g. breast, lung or colon cancers) detection in telepathology applications, by using advanced pattern recognition techniques on medical images. The approach consists in studying and developing new algorithms that will replicate pathologists’ behavior when observing medical images to detect unusual events. We propose the development of an innovative unusual event detection framework that first learns patterns from a set of procured microscopic slides with the help of subjective tests by experts in pathology and by the use of an eye tracker, coupled with an EEG signal recorder, to identify which part of the image pathologists find special (in other words, unusual, and containing anomalies). Then we extract all necessary features that would recognize and classify similar objects in new medical images. To achieve this goal, various techniques, such as image indexing and retrieval, unusual events modeling, feature extraction and classification will be considered. One substantial objective of the project is to enable a core contribution to WG3 of COST Action IC0604 (and also to further WGs), and so to International Standardization groups covering JPEG 2000, AIC, JPSearch and also DICOM, CEN and HL7 through the Action’s WG2.
|
Partner und Internationale Organisationen
(Englisch)
|
CH, CY, DE, ES, FI, FR, GR, HR, IT, LT, NL, NO, PL, PT, RS, UK
|
Abstract
(Englisch)
|
Curiosity is an emotion in human and animals, which motivates them to explore their environment for the purpose of learning, investigation and gaining information. This emotion can be replicated in artificial intelligent systems so that artificial agents such as robots or specific programs can show the similar behavior when they observe a phenomenon that is new and unexpected to them. If one or several physiological markers can be proven to have strong correlation with curiosity arousal, they can be used for the purpose of elaboration of an artificial curiosity module in machines. For instance, in the area of telepathology, the computers can benefit from such module and can select the scientifically relevant images among the immense image datasets that is collected using several high-definition sensors. the scientifically interesting images ( images with pathological cues) will be then further analyzed using image processing tools so that features corresponding to the saliency can be extracted. One of the physiological signals that is known to alter during some emotions such as surprise and curiosity is the brain electrical. In this research, electroencephalogram signal is analyzed to study whether it can be beneficial for the purpose of curiosity detection. To this end, several experiments have been performed. It has been shown that for most of the subjects, a strong correlation exists between the curiosity arousal and amplitude of P300 component of the EEG signal. The following contributions were made to the scientific and technical communities as the results of this project: 1- Contribution to JPEG committee call for medical image content in July 2011, on behalf of COST IC0604 and in collaboration with some of its partners including the Chair of the COST Action IC0604. 2- Contribution to the conference 'Advances in Telemedicine Applications', organized in Warsaw on June 2011, sponsored by COST IC0604, in which EPFL participated and reported its activities and achievements of this project in a contribution entitled: Recent Advanced in Medical Image Compression 3- Contribution to conferences 10th European Conference on Telepathology, organized in Vilnius on June 2010, co-sponsored by COST IC0604, in which EPFL paticipated and reported its activities and achievements of this project in a contribution entitled: Towards second generation expert systems in telepathology for aid in diagnosis. 4- Organization, contribution and co-charing of WG3 - Images: Analysis, Processing, Retrieval and Management in COST Action IC0604
|
Datenbankreferenzen
(Englisch)
|
Swiss Database: COST-DB of the State Secretariat for Education and Research Hallwylstrasse 4 CH-3003 Berne, Switzerland Tel. +41 31 322 74 82 Swiss Project-Number: C08.0129
|
