Flavio Luiz Seixas

A Bayesian network decision model for supporting the diagnosis of dementia, Alzheimer's disease and mild cognitive impairment

Population aging has been occurring as a global phenomenon with heterogeneous consequences in both developed and developing countries. Neurodegenerative diseases, such as Alzheimer׳s Disease (AD), have high prevalence in the elderly population. Early diagnosis of this type of disease allows early treatment and improves patient quality of life. This paper proposes a Bayesian network decision model for supporting diagnosis of dementia, AD and Mild Cognitive Impairment (MCI). Bayesian networks are well-suited for representing uncertainty and causality, which are both present in clinical domains. The proposed Bayesian network was modeled using a combination of expert knowledge and data-oriented modeling. The network structure was built based on current diagnostic criteria and input from physicians who are experts in this domain. The network parameters were estimated using a supervised learning algorithm from a dataset of real clinical cases. The dataset contains data from patients and normal controls from the Duke University Medical Center (Washington, USA) and the Center for Alzheimer׳s Disease and Related Disorders (at the Institute of Psychiatry of the Federal University of Rio de Janeiro, Brazil). The dataset attributes consist of predisposal factors, neuropsychological test results, patient demographic data, symptoms and signs. The decision model was evaluated using quantitative methods and a sensitivity analysis. In conclusion, the proposed Bayesian network showed better results for diagnosis of dementia, AD and MCI when compared to most of the other well-known classifiers. Moreover, it provides additional useful information to physicians, such as the contribution of certain factors to diagnosis.

Image registration using genetic algorithms

This paper addresses the image registration problem applying genetic algorithms. The image registrations objective is the definition of a mapping that best match two set of points or images. In this work the point matching problem was addressed employing a method based on nearest-neighbor. The mapping was handled by affine transformations. Experiments were conducted using three 2D synthetic point-sets with different affine transformations and noise. The results were compared against other optimization techniques. The similarity of two point-sets is measured using the Euclidean distance between matched points.

Automated Segmentation of the Corpus Callosum Midsagittal Surface Area

The non-invasive in vivo nature of magnetic resonance imaging (MRI) makes it the modality of choice of many neuroanatomical imaging studies. This paper discusses automatic brain structure segmentation based on previous knowledge on statistical models. The method is validated by an experiment involving magnetic resonance images acquired from 20 healthy adult individuals (10 men and 10 women). The results provide normative data of the midsagittal surface area of the corpus callosum from a 46-55 years old range group, splitting results by gender. Our results were also compared with data obtained from other authors, validating the correlation between brain volume and the area of this structure. The final goal of this work is computer-aided diagnosis for brain diseases.

Intelligent automated brain image segmentation

The availability of modern computational techniques and advanced medical imaging protocols has increased the development of computer-aided diagnosis systems. This paper presents a fully automated brain structures segmentation algorithm for magnetic resonance (MR) images. Automated mechanisms reduce the excessive time consumed on manual segmentation and standardise the volumetric acquisition method. The proposed computational image segmentation method is based on a voxel-wise morphometry method, named voxel-based morphometry (VBM). The brain structure of interest of this paper is the hippocampus, a medial temporal lobe structure, precociously affected in Alzheimers disease (AD), which represents the most common cause of dementia worldwide. We evaluated 371 subjects from OASIS database, including normal controls and probable Alzheimers patients, splitting them in different age groups. Segmentation results demonstrated that grey matter and hippocampus volumes decrease in both groups proportionally to aging and it is more evident in AD subjects.

Automatic Segmentation of Brain Structures Based on Anatomic Atlas

The non-invasive in vivo nature of magnetic resonance imaging (MRI) makes it the modality of choice of many neuroanatomical imaging studies. This paper discusses automatic brain structure segmentation based on anatomic atlas. Our goal is to use image-processing algorithms and previous knowledge statistical models for segmentation and labeling of brain regions in order to support radiologists to make clinical diagnosis. Practical experiments show the results of brain tissue classification process and automatic region labeling in order to segment accurately the hippocampus and measure its volume. Hippocampus volumetric information can be useful to evaluate patients with Alzheimers disease. The final goal of this work is computer-aided diagnosis for brain diseases.

Clinical dementia rating score prediction based on MR segmentation

This work aims at predicting the clinical dementia rating (CDR) score with a fully automated human brain volumetric segmentation method based on anatomical atlas using magnetic resonance (MR) images. The CDR prediction method uses a Bayesian classifier considering 371 individuals. Practical results were assessed using the classifier true-positive rate. CDR score prediction can indicate an underlying neurodegenerative disorder, such as Alzheimerpsilas disease. Its early detection allows precocious therapeutic intervention and better clinical results.

A clinical decision support system for aiding diagnosis of Alzheimer's disease and related disorders in mobile devices

The worldwide aging phenomenon is a growing concern. Alzheimers disease (AD) has a high prevalence in the elderly. In this paper, we present a clinical decision support system for aiding the diagnosis of AD and related disorders. We describe systems main components and architecture, which is based on a mobile web-based platform. Its predictive model is based on Bayesian networks designed considering AD diagnosis criteria, trained and tested with the patient database of the Center for Alzheimers Disease and Related Disorder at the Institute of Psychiatry of the Federal University of Rio de Janeiro, Brazil. Patient database attributes are composed by predisposal factors, demographic data, assessment scales, symptoms and signs. When the system indicates a patient diagnosis, it provides: the most probable diagnosis, health data that lead to such diagnosis and, in case of low certainty factor, unobserved health data that should be collected to confirm or refuse the initial diagnostic hypothesis. Preliminary usability tests indicate potential use of the system in clinical practice.

Automated segmentation and volumetric analysis of brain components on MR imaging

As the world population has been growing old, the incidence of several brain diseases, including dementia, has risen. As a consequence, a greater demand for automated segmentation methods and volumetric analysis of the brain structure and its components has increased. This work is based on the use of image processing algorithms and previous knowledge statistical models for automated segmentation over 386 magnetic resonance exams. We analyze brain components considering two different image subsets: non-demented and demented individuals. Our results demonstrate gray and white matter loss, and CSF enlargement with aging.