頁籤選單縮合
題名 | Estimation of Power Intensity Distribution with Artificial Neural Network for Ultrasound Hyperthermia= |
---|---|
作者 | Chen,Yung-yaw; Lu,Da-ching; Lin,Win-li; Chen,Chi-hung; |
期刊 | 醫學工程 |
出版日期 | 19960200 |
卷期 | 8:1 1996.02[民85.02] |
頁次 | 頁43-48 |
分類號 | 410.1644 |
語文 | eng |
關鍵詞 | Power intensity distribution; PID; Transducer parameters; Artificial neural network; ANN; Average error; Ultrasound hyperthermia; |
英文摘要 | Varying the power intensity distribution (PID) is the conventional approach to obtaining a suitable temperature distribution during hyperthermia treatments. For a multi-element ultrasound transducer, the phase and amplitude of each element can be varied to earn a suitable PID to satisfy the requirements of an individual treatment. Control and optimization methods are the generally accepted means of searching for a set of parameters to satisfy what a treatment requires. Calculating the distributions of ultrasound intensity and resulting temperature is required when a new set of transducer parameters is applied during the searching process. However, too much computational time would be required to directly calculate these distributions during clinical patient therapy or even in treatment planning. This study employs an Artificial Neural Network (ANN) method to determine the relationship between the transducer parameters and the ultrasonic PID. This relationship can be then used to estimate the ultrasonic PID for a new set of transducer parameters within a relatively short period of time. A two-element annular array transducer is used in this preliminary study. Owing to the central axis symmetry of intensity distribution for this annular array, a two-dimensional (r-z) simulation program is implemented. During the training process the average error is used as the stop criterion. The simulation results demonstrate that the difference for the ultrasonic PIDs from ANN and direct calculation is reasonably small. The computational time for ANN estimation is short, i.e., about one-hundreds of that by direct calculation. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。