Noninvasive black-box temperature simulation: precise spatial generalisation

Teixeira, C. A.; Ruano, Antonio; Ruano, M. Graça; Pereira, W. C. A.

http://hdl.handle.net/10400.1/2242

Use this identifier to reference this record.

Name:	Description:	Size:	Format:
CBEB2006_Cesar.pdf		254.26 KB	Adobe PDF	Download

Send Feedback

Authors

Abstract(s)

In this paper the performance of a blackbox methodology is accessed for non-invasive timespatial temperature estimation. A gel-based phantom was heated at different intensities with therapeutic ultrasound, while temperature and RF-lines were collected. The models were trained and its structure selected to estimate the temperature in three discrete points, and at the end validated in unseen data, in the trained points and in two additional intermediate untrained points, in order to test the model s spatial generalization capacity. The best model had low complexity and a high generalization capacity, presenting in both the points and intensities a maximum absolute error inferior to 0.5 ºC, as desired in hyperthermia/diathermia.

Keywords

Non-invasive temperature estimation Black-box models Rradial basis functions neural networks Multi-objective genetic algorithms

URI

http://hdl.handle.net/10400.1/2242

Citation

Teixeira, C. A.; Ruano, A. E.; Ruano, M. G.; Pereira, W. C. A. Noninvasive black-box temperature simulation: precise spatial generalisation, Trabalho apresentado em XX Brasilian Congress on Biomedical Engineering-XX CBEB, In Proceedings of the XX Brasilian Congress on Biomedical Engineering-XX CBEB, São Paulo, 2006.