Sakura Kadowaki

Unsupervised segmentation for MR brain images

As described herein, we propose an unsupervised method for segmentation of magnetic resonance (MR) brain images by hybridizing the self-mapping characteristics of 1-D Self-Organizing Maps (SOMs) and using incremental learning functions of fuzzy Adaptive Resonance Theory (ART). As the proposed method requires the appropriate parameters to segment tissues (such as cerebrospinal fluid, gray matter and white matter) that are necessary for brain atrophy diagnosis, first we derive the optimal parameter set through the preliminary experiments. The main contribution of this work is to evaluate the effectiveness of the proposed method, considering the conventional methods that are highly accurate in terms of usefulness as classification techniques. We focus on Fuzzy C-means (FCM) and Expectation Maximization Gaussian Mixture (EM-GM) with previous setting of the number of clusters, and then Mean Shift (MS) without previous setting of the number of clusters. Through the comparative experiments on the two metrics, we confirmed that our method could achieve higher accuracy than these conventional methods. Additionally, we propose a Computer-Aided Diagnosis (CAD) system for use with brain dock examinations based on case analyses of diagnostic reading. We construct a prototype system for reducing loads on diagnosticians during quantitative analysis of the degree of brain atrophy. Field tests of 193 examples of brain dock medical examinees reveal that the system efficiently supports diagnostic work in the clinical field: the alteration of brain atrophy attributable to aging can be quantified easily, irrespective of the diagnostician.

Facial expression spatial charts for representing of dynamic diversity of facial expressions

This paper presents a method to generate individual Facial Expression Spatial Charts (FESC) using Self-Organizing Maps (SOM) and Fuzzy Adaptive Resonance Theory (ART) networks. We specifically examine the dynamic diversity of facial expressions in time-series facial images after conversion using Gabor wavelet filters. The proposed method consists of three steps: the first step is to extract topological features from time-series facial image datasets using SOMs; the second step is to integrate weights of SOM into categories using Fuzzy ART networks; the third step is to create FESCs integrated by all arousal levels produced from categories of facial expressions in each basic facial expression. For considering the influence that stress gives an expression, we measured the psychological emphasis that a subject feels at that time. The result shows a negative correlation for psychological stress and the expanse of FESC, which means that the expression became poor during feelings of stress.

Analysis of psychological stress factors by using bayesian network

This paper presents a gender-specific stress model to analyze the psychological stress factors on intentional facial expressions. We have focused on the relationship between facial expression intensity and Stress Response Scale (SRS-18). In this paper, we extract three facial expressions (i.e., happiness, anger, and sadness) from the basic six facial expressions defined by Ekman, and then represent a graphical model of the relationship between these three facial expressions and the psychological stress factors (“i.e., depression and anxiety”, “displeasure and anger”, and “lassitude”). In the experiment, we created an original facial expression dataset consisting of three facial expressions and a psychological stress dataset by SRS-18 obtained from 10 subjects during 7-20 weeks at one week interval. As the results of probabilistic reasoning based on the observed values of each facial expression, such trends are obtained as follows. BNs shows trends of different stress factor between men and women in relations of expression levels and psychological stress. Stress models appeared on happiness faces of “lassitude” factor in men, the anger faces of “displeasure and anger” were affected with stress factors in women.

Experimental studies with a hybrid model of unsupervised neural networks

This paper presents an unsupervised clustering method to classify the optimal number of clusters from a given dataset based solely on the image characteristics. The proposed method contains a feature based on the hybridization of two unsupervised neural networks, Self-Organizing Maps (SOMs) and Fuzzy Adaptive Resonance Theory (ART), which has a seamless mapping procedure comprising the following two steps. First, based on the similarity of the spatial topological structure of images, we will form a local neighborhood region holding the order of topological changes. Then the region is mapped to one-dimensional space equivalent to more than the optimal number of clusters. Furthermore, by additional learning in accordance with the order of the one-dimensional maps formed in the neighborhood region, we must generate suitable labels that match the optimal number of clusters. We use it as a target problem for which the number of categories or clusters is unknown. We emphasize the effectiveness of the proposed method for resolving the target problem for which the number of categories and clusters is unknown, and we anticipate its use for the categorization of facial expression patterns for time-series datasets and for the segmentation of brain tissues shown in Magnetic Resonance (MR) images.

Objective Segmentation Based on Characteristics of Single Channel MR Images

We propose an objective segmentation method for magnetic resonance (MR) images of the brain using self-mapping characteristics of one-dimensional self-organizing maps (SOM). The proposed method requires no operators to specify the representative points, but can segment tissues (such as cerebrospinal fluid, gray matter and white matter) necessary for brain atrophy diagnosis. Doing clinical image experiments, we demonstrate the effectiveness of our method. As a result, we can obtain segmentation results that agree with anatomical structures such as continuities and boundaries of brain tissues. In addition, we propose a computer-aided diagnosis (CAD) system for brain-dock examinations based on the use case analysis of diagnostic reading, and construct a prototype system for reducing loads to diagnosticians that occur in quantitative analyses of the extent of brain atrophy. Through field tests of 193 examples of brain dock medical examinees at Akita Kumiai General Hospital, we also present the prospect of efficient support of diagnostic reading in the clinical field because the aging situation of brain atrophy is readily quantifiable irrespective of a diagnosticians expertise.

Unsupervised segmentation of MR images for brain dock examinations

As described herein, we propose an unsupervised method for segmentation of magnetic resonance (MR) brain images by hybridizing the self-mapping characteristics of 1-D Self-Organizing Maps (SOMs) and using incremental learning functions of fuzzy Adaptive Resonance Theory (ART). The proposed method requires no operator to specify the representative points. Nevertheless, it can segment tissues (such as cerebrospinal fluid, gray matter and white matter) that are necessary for brain atrophy diagnosis. Additionally, we propose a Computer-Aided Diagnosis (CAD) system for use with brain dock examinations based on case analyses of diagnostic reading. We construct a prototype system for reducing loads on diagnosticians during quantitative analysis of the degree of brain atrophy. Field tests of 193 examples of brain dock medical examinees reveal that the system efficiently supports diagnostic work in the clinical field: the alteration of brain atrophy attributable to aging can be quantified easily, irrespective of the diagnostician.