Catálogo de publicaciones - libros

Nuclear medicine (NM) images inherently suffer from large amounts of noise and blur. The purpose of this research is to reduce the noise and blur while maintaining image integrity for improved diagnosis. The proposed solution is to increase image quality after the standard pre- and post-processing undertaken by a gamma camera system. Mean Field Annealing (MFA) is the image processing technique used in this research. It is a computational iterative technique that makes use of the Point Spread Function (PSF) and the noise associated with the NM image. MFA is applied to NM images with the objective of reducing noise while not compromising edge integrity. Using a sharpening filter as a post-processing technique (after MFA) yields image enhancement of planar NM images.

We tried to evaluate the blood flow in left ventricle quantitatively by combining wall thickness obtained from cardiac magnetic resonance imaging (MRI) and myocardial perfusion from single-photon emission tomography (SPECT). Paired MRI and myocardial perfusion SPECT from 16 patients including apical hypertrophic cardiomyopathy (APH) and normal subjects were considered. Blood flow per unit myocardium volume was calculated by 3-D surface-based registration between MRI and SPECT images.

Evaluation was performed by creating bull’s eyes map of wall thickness, blood flow (cardiac perfusion), and blood flow per unit myocardium volume. In APH case, decrease of the blood flow was detected in the cardiac apex by integration of the two values though that could not be recognized from SPECT image alone.

Myocardium perfusion in left ventricle can be evaluated accurately by converting to the value per unit myocardium volume reflecting wall thickness. And it helps to distinguish APH case whose SPECT image is resemble to a distribution of normal case.

Even though an assumption of Gaussian-distributed data has been typically used for magnitude fMRI data analysis, magnitude fMRI data belongs to a Rician distribution rather than a Gaussian distribution because it is obtained by computing the magnitude of complex valued, Gaussian-distributed data. It has been known that the performance of hypothesis testing can be improved using a Rician distribution, but a main obstacle to its use is computational load. In statistical inference, maximum likelihood estimation of parameters and the likelihood ratio method of hypothesis testing are commonly used for the Rician and Gaussian distributions, especially generalized likelihood ratio tests (GLRTs) for the Ricain distribution. But, by considering Rician-distributed data, maximum likelihood estimators (MLEs) are not acquired as a solution of a closed form expression unlike ordinary least squares (OLSs) for Gaussian-distributed data. In this study, we apply an iterative optimization solution by means of the EM (expectation-maximization) algorithm to the Rician maximum likelihood estimation problem. Our simulation results suggest that the EM algorithm is an efficient way to find Rician MLEs, so that it makes hypothesis testing based on the Rician distribution practically applicable to fMRI data analysis.

Recently, with the increase in the number of the FFDM, these schemes have started on being applied to images stored in the DICOM standard. Thus, this work purpose was the development of a procedure intended to detect microcalcifications and suspicious breast masses from regions of interest previously selected from digital mammograms as part of a CAD scheme. A routine designed to be applied to a mammogram in DICOM standard was developed and the stages corresponding both to microcalcifications and masses segmentation and detection were integrated to the scheme. A previous developed scheme based on the Watershed Transform to masses detection was applied to 252 ROIs from 130 digitized conventional mammograms. The results have shown 92% of true positive and 10% of false positives. For clustered microcalcifications detection, another previous developed procedure was applied to 165 ROIs from 120 mammograms, resulting in 93% of true positive and 16% of false positive. By using the same procedures to 71 digital mammograms obtained from FFDM, the rates have shown a little decrease in the scheme performance: 86% of true positive and 19% of false positive for masses detection; 90% of true positive and 29% of false positive for clusters detection. Although the tests with digital mammograms have been carried with a smaller number of images and different cases compared to the digitized ones, including several dense breasts images, the results can be considered comparable, mainly for clustered microcalcifications detection with a difference of only 3% between the sensibility rates for the both images sets. Another important feature affecting these results is the contrast difference between the two images set. This implies the need of extensive investigations not only with a larger number of cases from FFDM but also on the parameters related to its image acquisition as well as to its corresponding processing.

Cross-section Tomography (CT) is the most commonly used technique to identify and measure calcium buildup in and around the coronary arteries. In this work, a novel diagnosis assist tool is presented for doctors to inspect conditions of patents’ coronary arteries. The input of the proposed tool is CT volume image taken by medical imaging instruments. The main objective of the proposed algorithm is to build up the 3-D model of coronary arteries, and it would be easier for doctors to diagnosis patents’ heart condition. The proposed tool also capable of straighten the selected coronary artery fragments, and provide slices of the straightened artery at any cross-section angle, which can provide doctors with more information about the tissues surrounding the interested coronary artery. The experiments are proceeded with hospital collaboration, and the result shows that the assistances provided by the proposed algorithm is clinical practical.

A simple method of threshold selection for medical image denoising has been proposed. A noise added image was transformed to wavelet domain, the histograms of wavelet coefficients were obtained for each resolution level and component. We found the shapes of histograms were almost the same, and have a common shape. The threshold value was determined by linear approximation of the portion of a histogram.

Multi-grid method is applied to optical flow estimation in ultrasound image sequences. The existing multigrid method is modified to accommodate the nonlinearity in the motion model. The nonlinear iteration, instead of the linear one, is proposed for the approximate solution in the finest grid. The proposed method was tested in both of panoramic imaging and the spatio-temporal noise reduction. The results showed that its performance was not distinguishable from the one-level method in spites of quite less computational burden.

We present an efficient clustering method for detecting the tumor in positron emission tomography(PET) of the tumor bearing small animal. We used iterative threshold method to remove the background noise and then we applied two clustering procedures in order. The one is clustering method based on intensity to segment the tumor region and the other is clustering based on connectivity to remove false positive region from the segmented region. The tumor tissue looks bright in the image compared to surrounding normal tissue because of glucose uptake. Therefore, based on volume intensity, we divided all elements of the image into several clusters, the tumor, living bodies, background using improved fuzzy c-means clustering(FCM). Using FCM with the sorted initial mean of each cluster gets out of the wrong optimization and reduces the amount of time-consumed. However, not only the tumor tissue, but also the other organs like heart, bladder can also have high intensity value because of glucose metabolism. So in order to separate the tumor and false positive region, we applied geometric clustering based on connectivity. Proposed segmentation method can lead a robust analysis of the tumor growth with the aid of the quantitative measurements such the tumor size or volume.

The purpose of this study is to develop a semiautomated system for human image processing with which tissues or organs from human images can be segmented and classified by people who have basic knowledge of image processing. Furthermore, the proposed system was designed to adopt an intuitive menu method which helped users to use and understand the program easily. In addition, the proposed system is independent on types of human tissues or images. In this paper, a new intuitive menu based and semi-automated image processing system with essential image processing functions for medical images is introduced.

The imaging system most widely used to obtain 3-D position information for an object is a stereo vision system using two cameras. The purpose of this research is to calculate 3-D distance information using the image differences acquired by making angle changes with one small-sized endoscopy camera. As a result of measuring increments of 1 mm while adding changes of 1°–7° in angle revolution for the distal tip for close range endoscopy (10–25 mm is the working distance for an endoscope) there was 0.25–0.18 mm of error on average. When the test section was 15–20 mm, which is the optimal working distance, it showed 0.15–0.05 mm of error on average. Therefore, the error ratio according to the measured distance is proven to be within 1 %. Also, it is estimated that 3-D short distance information shown on this research will be applicable to medical and industrial automation.