Temperature modelling of human tissue subjected to ultrasound for therapeutic use is essential for an accurate instrumental assessment and calibration. The existence of accurate temperature models would enable a safe and efficient application of the thermal therapies. The main objective of this work is the comparison between the performance of non-linear models and linear models for punctual temperature estimation in a homogeneous medium. The final goal of the work hereby reported is the construction of neural models for “in-vivo” temperature estimation. The linear models employed were AutoRegressive with eXogenous inputs (ARX), and the non-linear models used were radial basis functions neural network (RBFNN). The best-performed RBFNN structures were selected using the Multi-objective Genetic Algorithm (MOGA). The best performed neural structure present a maximum absolute error of 0.2 oC, which is one order magnitude less than the one presented by the best ARX.
[1]
Carlos M. Fonseca,et al.
Genetic assisted selection of RBF model structures for greenhouse inside air temperature prediction
,
2003,
Proceedings of 2003 IEEE Conference on Control Applications, 2003. CCA 2003..
[2]
P. VanBaren,et al.
Two-dimensional temperature estimation using diagnostic ultrasound
,
1998,
IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control.
[3]
C. Fonseca,et al.
Non-Linear System Identification with Multiobjective Genetic Algorithms
,
1996
.