Brain MR Image Segmentation Using Self Organizing Map

Medical Image processing has become an accelerating subject of interest these days. Technology is growing day by day to capture the accurate internal body images of human beings to diagnose the abnormalities efficiently. To aggravate the efficiency Artificial Neural Networks (ANNs) can be used as a very good tool. This paper is an approach to classify the processed MRI images of human brain for the presence or absence of a tumor. The Artificial Neural Network (ANN) technology developed here further locates the tumor in case of its presence. The technology is based on the concept of segmenting the MRI image of the human brain and then finding out the gradient of pixels in different segments. On the basis of the pixel densities at different segments the ANN cluster the image into an image with a tumor or without a tumor. Further the position of the tumor is calculated.

International Journal of Computational Intelligence Research, 2007

This paper presents the application of a self-organizing map (SOM) model for the reconstruction of cerebral cortex from MRI images. The cerebral cortex is an important tissue for many brain science or medicine related researches. Since it is difficult to extract the highly folded and buried cortical surface, we apply the SOM model to deform the easily extracted white matter surface on a layered distance map to obtain the cortical surface. The layered distance map is calculated according to the extracted white matter surface and segmented gray matter. The proposed method can reconstruct the proper cortical surface and thus make the measurement of cortical thickness easy. The simulations on T1-weighted MRI images show that the proposed algorithm is robust to reconstruct the cerebral cortex.

The implementation of self organizing map (SOM) and its application to the analysis of biomedical images are presented. The SOM algorithm was implemented in MATLAB program suite with various optional parameters enabling the adjustment of the model according to the user's requirements. For easier application of SOM the graphical user interface has been developed. The segmentation and edge detection procedures are critical steps in the analysis of biomedical images, enabling for instance the detection of the abnormal structure or the recognition of di�erent types of tissue. The self-organizing map provides a quick and easy approach for these tasks with satisfying quality of outputs, which has been veried using the high-resolution CT images capturing the expressions of the Granulomatosis with polyangiitis (GPA) disease.

Digital Signal Processing, 2009

A new image segmentation system is presented to automatically segment and label brain magnetic resonance (MR) images to show normal and abnormal brain tissues using selforganizing maps (SOM) and knowledge-based expert systems. Elements of a feature vector are formed by image intensities, first-order features, texture features extracted from graylevel co-occurrence matrix and multiscale features. This feature vector is used as an input to the SOM. SOM is used to over segment images and a knowledge-based expert system is used to join and label the segments. Spatial distributions of segments extracted from the SOM are also considered as well as gray level properties. Segments are labeled as background, skull, white matter, gray matter, cerebrospinal fluid (CSF) and suspicious regions.

2009

Abstract. Image segmentation has been and is likely to be an important component of the content-based image acquisition and retrieval systems. In this paper, we have proposed an image segmentation technique that uses Kohonen's self-organizing map (SOM) neural network for segmentation of color images. It has been observed that, SOM training if performed on the wavelet-transformed image, not only reduces the training time of the SOM but also make more compact segments.

—Robust brain magnetic resonance (MR) segmentation algorithms are critical to analyze tissues and diagnose tumor and edema in a quantitative way. In this study, we present a new tissue segmentation algorithm that segments brain MR images into tumor, edema, white matter (WM), gray matter (GM), and cere-brospinal fluid (CSF). The detection of the healthy tissues is performed simultaneously with the diseased tissues because examining the change caused by the spread of tumor and edema on healthy tissues is very important for treatment planning. We used T1, T2, and FLAIR MR images of 20 subjects suffering from glial tumor. We developed an algorithm for stripping the skull before the segmentation process. The segmentation is performed using self-organizing map (SOM) that is trained with unsupervised learning algorithm and fine-tuned with learning vector quantization (LVQ). Unlike other studies, we developed an algorithm for clustering the SOM instead of using an additional network. Input feature vector is constructed with the features obtained from stationary wavelet transform (SWT) coefficients. The results showed that average dice similarity indexes are 91% for WM, 87% for GM, 96% for CSF, 61% for tumor, and 77% for edema. Index Terms—Brain magnetic resonance (MR), image segmen-tation, learning vector quantization (LVQ), self-organizing feature map, stationary wavelet transform (SWT).

Medical image processing is considered the area a great challenge for researchers because of the complexity of this area and its importance to help doctors to diagnose patients effectively. Here we will base on magnetic resonance brain images processing trough the image processing technique is hybrid self-organizing map (SOM) with fuzzy k-means (FKM) algorithm, which gives successfully on tumor and excellent areas within the brain tissue, we propose improved algorithm is accurate and efficient in term of accuracy detection, n Jaccard coif, nDice (DOI), sensitivity, specificity, recall, precision, performing segment. At the same time, the improved algorithm proposed provides the best information and functions, as well as the efficiency and accuracy to handle high input brain MR images.

Artificial Neural Networks - Architectures and Applications, 2013

IFIP Advances in Information and Communication Technology, 2011

Color image processing systems are used for a variety of purposes including medical imaging. Basic image processing algorithms for enhancement, restoration, segmentation and classification are modified since color is represented as a vector instead of a scalar gray level variable. Color images are regarded as two-dimensional (2-D) vector fields defined on some color space (like for example the RGB space). In bibliography, operators utilizing several distance and similarity measures are adopted in order to quantify the common content of multidimensional color vectors. Self-Organizing Feature Maps (SOFMs) are extensively used for dimensionality reduction and rendering of inherent data structures. The prop osed windowbased SOFM uses as multidimensional inputs color vectors defined upon spatial windows in order to capture the correlation between color vectors in adjacent pixels. A 3x3 window is used for capturing color components in uniform color space (L * u * v *). The neuron featuring the smallest distance is activated during training. Neighboring nodes of the SOFM are clustered according to their statistical similarity (using the Mahalanobis distance). Segmentation results suggest that clustered nodes represent populations of pixels in rather compact segments of the images featuring similar texture.