Deformable Multi Template Matching with Application to Portal Images1

The exact positioning of patients during radiotherapy is essential for high precision treatment. The registration of portal image sequences (megavoltage X-ray images) can help to control the patient position. The particular problem of portal imaging devices is that they provide imagery with extremely low contrast. Therefore, feature extraction followed by binary feature matching is a difficult task. To circumvent the step of feature extraction, the algorithm presented in this paper relies on an area-based matching of the image signal. This strategy contrasts with most state of the art registration algorithms for portal imagery. The paper includes the mathematical formalism of the least squares template matching method, as well as the framework for automated quality control, together leading to a fast, robust and very accurate image matching procedure. Tests have been carried out on 17 portal image series with more than 100 images in total. The algorithm only failed in less than 5 % of the images, which is very satisfying for portal images. In addition all matching failures were reliably detected by the self-diagnostics included in the computational framework. In all successful matches, no visible misalignment has been found. Tests with artificially rotated and shifted images allow us to demonstrate the good performance of the method with respect to a ground truth.