返回Google圖書搜尋

Maximum Likelihood-expectation Maximum Reconstruction with Limited Dataset for Emission Tomography

出版	University of Akron, 2007
URL	http://books.google.com.hk/books?id=TM7ZxQEACAAJ&hl=&source=gbs_api

註釋"Medical imaging provides a non-invasive technique to look at the structural and functional information of internal organs and structures. One of the most widely used medical imaging techniques is emission tomography, in which a radioisotope is given to a patient. Measurement of the radioactive distribution throughout the patient gives the physiological and patho-physiological information about the patient [2]. The two types of emission tomography are Positron Emission Tomography (PET) and Single Photon Emission Tomography (SPECT). The image is reconstructed using the data acquired, also known as the projection data, by a mathematical technique known as Filtered Back Projection (FBP). In emission tomography, it is difficult to locate the exact location from which an emission originated. If the detectors are placed apart, the probability of scatter being detected as signal decreases. The main risk with this method is limited projection data due to the limited number of total number of detectors. The main purpose of this study is to verify whether or not we can reconstruct images with a limited dataset. This method will provide us a better estimate of the exact location of the radioactivity. Traditional reconstruction algorithms like FBP can not reconnstruct an image with a limited dataset. Hence we work with an iterative algorithm, Maximum Likelihood-Expectation Maximum (ML-EM). This research uses the most commonly used Shepp-Logan head phantom. To test whether we can reconstruct the image using a limited dataset without any statistical difference, we use the Chi-Square Goodness of Fit test. Since it is an iterative approach, we also look at the line profile of the reconstructed images with a different number of iterations. The primary conclusion drawn from this testing was that no statistically significant differences exist between the images reconstructed from a limited dataset and the original image. We have proven that by using Modified ML-EM algorithm, we can reconstruct the image with limited dataset."--Abstract