Neural network models for non-invasive two-point temperature monitoring in a homogeneous medium irradiated by therapeutic ultrasound

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

http://hdl.handle.net/10400.1/2271

Use this identifier to reference this record.

Name:	Description:	Size:	Format:
2473_Teixeira_CA.pdf		278.85 KB	Adobe PDF	Download

Send Feedback

Authors

Abstract(s)

In this paper the performance of neural network models for non-invasive temperature prediction in two points of a glycerine medium, irradiated with therapeutic ultrasound is investigated. These points are located in the axial line of the therapeutic ultrasound transducer. It is assumed that the temperature in these points is non-linearly related with some spectral features and one temporal feature, extracted from the collected A-Scans. The neural networks used were Radial Basis Functions Neural Networks (RBFNN), where the best-fitted models structures for each point were selected in a genetic multi-objective fashion, due to the enormous number of possible model structures. The best-fitted models predicted temperature curves of two unseen data sequences during approximately 2 hours with maximum absolute errors less than 0.5 ºC.

URI

http://hdl.handle.net/10400.1/2271

Citation

Teixeira, C. A.; Ruano, A. E.; Ruano, M. G.; Pereira, W. C. A. Neural Network Models for Non-Invasive Two-Point Temperature Monitoring in a Homogeneous Medium Irradiated by Therapeutic Ultrasound , Trabalho apresentado em 3rd European Medical and Biological Engineering Conference (EMBEC’05), In Proceedings of the 3rd European Medical and Biological Engineering Conference (EMBEC’05), Prague, 2005.