Research On CBCT Projection Data Denoising Algorithm

With the development of economy, the pressure which people need to be underis bigger and bigger, more and more patients with tumor are appeared, the methods oftreatment of tumor are in progress. At present, the popular approaches for cancerinclude operation, radiotherapy and chemotherapy, radiotherapy is more and morerecognized as one of the top three cancer treatments. Image guided radiation therapyis the most advanced radiotherapy techniques, it considers the influence of the motionof tissue and organ and the errors of the position on the distribution of dose, it uses theadvanced equipment to monitor the tissue and organ real-time during the treatment,adjust the conditions, locations and doses of treatment according to the imageinformation to acquire precise radiotherapy, it is the image guided radiation therapy.Cone-Beam Computed Tomography (CBCT) has been widely applied to IGRTdue to its good real-time performance, high sensitivity and feasibilities. However, lowcontrast of soft tissue and noises in CBCT have affected the quality of the image andprevented the acquisition of the positioning information seriously, especially at noises.The low quality of image not only affect acquiring the accurate positioninginformation, but also affect the clinical operation. The amount of noise is determinedby X-ray dose level in CBCT image, to increase the amount of X ray radiation canreduce the noises and obtain higher quality of CBCT image, but higher radiation dosewill also affect the health of the patients. So how to improve the quality of CBCTimages and obtain accurate radiotherapy positioning information in the lowerradiation dose has become a hot topic.First of all, this paper analyzes the principle, noises and denoising algorithm ofmedical images, then focus on CBCT image and proposes an adaptive non-local meandenoising algorithm based on the analysis of principle, the factors affecting the quality,the process of denoising and the denoising algorithm of CBCT image. The main innovative points of this paper are as follow:(1) Do research on the characteristics of noises of CBCT images. A formula for noiseestimation is put forward based on the analysis of noises. This formula can obtainestimation results effectively no matter in the test image or in the clinical image, thisnoise standard deviation estimation formula can also get better denoising effect.(2) According to the principle and characteristics of CBCT image, a denoisingalgorithm based on adaptive non-local mean for CBCT projection data is proposed.The results of comparison in the test images and the clinical images show that thisalgorithm can obtain better effect in denoising, preserve the detail information andenhance the contrast of CBCT image at the same time, it is good at acquiring accurateposition information and diagnosing in the clinical.