Reduction of Processing times for Temporal Subtraction on Lung Ct Image Employing Octree Algorithms

The temporal subtraction image, which can be obtained by subtracting previous image from current one, is useful for visual screening in clinical field. The temporal subtraction technique removes normal structures, e.g., blood vessel. Hence, it can enhance interval changes such as the new lesions and the changes of existing abnormalities on medical images. Recently, several temporal subtraction methods have been proposed for thoracic medical images. In temporal subtraction, image registration technique is required for correcting displacement between current image and previous one. However, efficient image registration technique of temporal subtraction for MDCT (Multi Detector-row CT) has not been proposed because of the complication of deformation in 3 dimensional region. In this paper, we propose a new efficient computer aided diagnosis (CAD) algorithms for detection of lung nodules which are obtained by temporal subtraction for thoracic MDCT images. We have tried to reduce the computational time for the temporal subtraction image by use of octree algorithms on 3-dimensional image space. To evaluate our method, we have applied the method to 4 MDCT dataset and confirmed its efficiency.