A. Gacsadi

Directional features for automatic tumor classification of mammogram images

One way for breast cancer diagnosis is provided by taking radiographic (X-ray) images (termed mammograms) for suspect patients, images further used by physicians to identify potential abnormal areas thorough visual inspection. When digital mammograms are available, computer-aided based diagnostic may help the physician in having a more accurate decision. This implies automatic abnormal areas detection using segmentation, followed by tumor classification. This work aims at describing an approach to deal with the classification of digital mammograms. Patches around tumors are manually extracted to segment the abnormal areas from the remaining of the image, considered as background. The mammogram images are filtered using Gabor wavelets and directional features are extracted at different orientation and frequencies. Principal Component Analysis is employed to reduce the dimension of filtered and unfiltered high-dimensional data. Proximal Support Vector Machines are used to final classify the data. Superior mammogram image classification performance is attained when Gabor features are extracted instead of using original mammogram images. The robustness of Gabor features for digital mammogram images distorted by Poisson noise with different intensity levels is also addressed.

Gabor wavelet based features for medical image analysis and classification

Identifying relevant, representative and more important, discriminant image features for analysis and proper image classification purpose is one of the main tasks in image processing and pattern recognition field. In this paper, Gabor wavelets based features are extracted from medical mammogram images representing normal tissues, or benign and malign tumors. Once features are detected, Principal Component Analysis (PCA) is further employed to reduce data dimensionality. To an end, directional properties and frequency spectrum of those features are analyzed with respect to the classification performance by employing multiclass support vector machines as classifier. For comparison, as baseline, PCA was also applied directly to the data set with no feature filtering. The experimental results indicate that directional properties of Gabor wavelets provided by their orientation are important issues to accurately discriminate mammogram tumor types.

An analogic CNN algorithm for following continuously moving objects

A potential application of cellular neural networks (CNN) in adaptive control of a robot based on visual information is considered. The high processing speed of the network is used to provide real time processing. In this contribution an analogic CNN algorithm for following a moving object is shown. The algorithm was tested with the CNN infrastructure (CADETWin and CCPS).

Vision based algorithm for path planning of a mobile robot by using cellular neural networks

The paper presents a new vision based algorithm for mobile robots path planning in an environment with obstacles. Cellular neural networks (CNNs) processing techniques are used here for real time motion planning to reach a fixed target. The CNN methods have been considered a solution for image processing in autonomous mobile robots guidance. The choice of CNNs for the visual processing is based on the possibility of their hardware implementation in large networks on a single VLSI chip (cellular neural networks -universal machine, CNN-UM (Roska and Chua, 1993 and Kim et al., 2002))

Medical image enhancement by using cellular neural networks

The paper presents a medical image enhancement method taking the noise reduction and the contrast enhancement into consideration, as well as the possibility of implementation on an existing cellular neural network universal chip (CNN-UC), in a single step, by using only linear templates of 3times3 dimensions. Due to complete parallel processing, computing-time reduction is achieved

Path Planning of Mobile Robots by Using Cellular Neural Networks

The paper presents some vision-based algorithms for mobile robots guidance in an environment with obstacles. Cellular neural networks (CNNs) processing techniques are used here for real time motion planning to reach a fixed target. The CNN methods are considered an advantageous solution for image processing in autonomous mobile robots guidance

Variational computing based segmentation methods for medical imaging by using CNN

The paper presents a new variational computing based medical image segmentation method by using Cellular Neural Networks (CNN). By implementing the proposed algorithm on FPGA (Field Programmable Gate Array) with an emulated digital CNN-UM (CNN-Universal Machine) there is the possibility to meet the requirements for medical image segmentation.

Motion planning for two mobile robots in an environment with obstacles by using cellular neural networks

The paper presents a visual control algorithm based on images, for two mobile robots in an environment with obstacles. Cellular neural networks (CNNs) processing techniques are used here for motion planning in real time of two mobile robots moving to the same target. The algorithm can be extended for three or more robots.

Image inpainting methods by using cellular neural networks

Some CNN methods are presented that can be used for the reconstruction of damaged or partially known images. The proposed methods take the possibility of direct implementation on an existing CNN chip into account, in a single step, by using 3*3 dimensional linear reaction templates. Due to complete parallel processing, computational time reduction is achieved. Efficiency of these methods can be increased by combining them with nonlinear template that ensures the growth of the local properties spreading area along with regional ones.

Variational computing based images denoising methods by using cellular neural networks

The paper presents a new variation based image denoising method by using Cellular Neural Networks (CNN). Our proposed method offers better efficiency in terms of image denoising and edge preservation, comparing to other previous CNN methods based on variational model.