- News
- Subject and Goals
- Council and Annual Reports
- Contacts and Working Groups
- Lectures and Presentations
- Conferences and Seminars
- Publications
- Application Demos
- International Travels
- Visitors
- Past Events
- E-Synergy Login
SEARCH
MEMBERS OF RESEARCH CENTER
Research Centre
Data - Algorithms - Decision Making
Data - Algorithms - Decision Making
Established in 2005 under support of MŠMT ČR (project 1M0572)
Lectures and Presetations
Probabilistic Estimation of Radiation Dose with Few Noisy Measurements.
Lecturer:
From:
Mar. 6 2007 2:00PM
To:
Mar. 6 2007 3:00PM
Place:
room 474, ÚTIA AS ČR
Description:
The topic will illustrate a case when nearly insufficient number of data is compensated by expert knowledge and external information in Bayesian way.
As a real application, treatment of thyroid cancer (the nuclear medicine) will be described. For its purpose, time sequence of thyroid activity (number of decays per second) is measured. The task is to identify the model of the time-activity curve with the patient-specific data and to estimate distribution of its time integral, which is proportional to absorbed dose.
The Bayesian identification of a linear regression model for time dependence of thyroid gland activity will be presented. The inclusion of prior knowledge will be discussed. The obtained posterior distribution is represented via a Langevin diffusion algorithm. Its optimization will be explained. The posterior inference of the total patient radiation dose is computed, allowing the uncertainty of the dose to be quantified in a consistent form. The relevance of this work in clinical practice will be explained.
As a real application, treatment of thyroid cancer (the nuclear medicine) will be described. For its purpose, time sequence of thyroid activity (number of decays per second) is measured. The task is to identify the model of the time-activity curve with the patient-specific data and to estimate distribution of its time integral, which is proportional to absorbed dose.
The Bayesian identification of a linear regression model for time dependence of thyroid gland activity will be presented. The inclusion of prior knowledge will be discussed. The obtained posterior distribution is represented via a Langevin diffusion algorithm. Its optimization will be explained. The posterior inference of the total patient radiation dose is computed, allowing the uncertainty of the dose to be quantified in a consistent form. The relevance of this work in clinical practice will be explained.
Contact person:
Within the event:
Pravidelný seminář Rozhodování a řízení za neurčitosti. (Contact person: Miroslav Kárný)