Catálogo de publicaciones - libros

Multi-slice CT is a continuously evolving technology that is increasingly used for cardiovascular applications, in particular to assess the coronary arteries and CABGs. A comprehensive work-up of these patients, however, also includes assessment of the cardiac valves. Some of the important information about the valves can also be obtained from the same multi-slice CT scan obtained during imaging of the coronary arteries or CABGs, obviating the need of an additional scan.

Since the velocity of the blood and thus the pressure gradient along the aortic and mitral valve cannot be determined from multi-slice CT imaging, information provided by this approach about the cardiac valves is mostly restricted to morphological diagnosis. For example, multi-slice CT is particularly valuable for monitoring aortic valve calcium in patients on lipid-lowering therapy. Due to the radiation exposure inherent with the technique, however, the indications for repetitive multi-slice CT scans need to be restricted to patients who cannot be assessed thoroughly with echocardiography or MRI. Accurate quantification of aortic valve area, early detection of perivalvular abscess formation, and diagnosis of residual systolic anterior movement of the mitral valve and tendinous cords in patients with hypertrophic obstructive cardiomyopathy after percutaneous transluminal septal myocardial ablation may become other diagnostic niches of multi-slice CT in cardiac imaging.

Boolean network is one of the commonly used methods for building gene regulatory networks from time series microarray data. However, it has a major drawback that requires heavy computing times to infer large scale gene networks. This paper proposes a variable selection method to reduce Boolean network computing times using the chi-square statistics for testing independence in two way contingency tables. We compare the computing times and the accuracy of the estimated network structure by the proposed method with those of the original Boolean network method. For the comparative studies, we use simulated data and a real yeast cell-cycle gene expression data (Spellman , 1998). The comparative results show that the proposed variable selection method improves the computing time of Boolean network algorithm. We expect the proposed variable selection method to be more efficient for the large scale gene regulatory network studies.