Non-Contact Respiration Measurement Using Ultra-wideband Array Radar with Adaptive Beamforming Technique for Cancer Radiotherapy

Tumor position estimation is the key to effective radiation therapy. In clinical situations, tumor position is predicted from the breath phase. However, the accuracy of the prediction deteriorates when the breathing type is not known. To solve this problem, in this paper, we propose a new method to separate the signals from the chest and abdomen using ultra-wideband (UWB) radar based on an adaptive beamforming technique. First, we estimate the positions of the targets using the Capon method, and modify the method for near-field two-dimensional scanning. Next, we separate the signals using the directionally constrained minimization of power method that suppresses the off-axis signals by calculating the optimal weighting vector. We estimate the displacement at each scattering point using the separated signal, and evaluate the performance of the proposed method via realistic numerical simulation. The conventional non-adaptive beamforming technique failed to estimate the three scattering positions. However, the proposed method succeeded in estimating the scattering positions. The root mean squared errors of the displacement estimated using the proposed method at three scatters were 0.138 mm, 0.123 mm, and 0.259 mm.