Print ISSN: 1681-6900

Online ISSN: 2412-0758

Keywords : Pattern Recognition


Lung Cancer Detection from X-ray images by combined Backpropagation Neural Network and PCA

Israa S. Abed

Engineering and Technology Journal, 2019, Volume 37, Issue 5A, Pages 166-171
DOI: 10.30684/etj.37.5A.3

The lungs are portion of a complex unit, enlarging and relaxing numerus times every day to supply oxygen and exude CO2. Lung disease might occur from troubles in any part of it. Carcinoma often called Cancer is the generally rising and it is the most harmful disease happened in humankind. Carcinoma occurs because of uncontrolled growth of malignant cells inside the tissues of the lungs. Earlier diagnosis of cancer can help save large numbers of lives, while any delay or fail in detection may cause additional serious problems leading to sudden fatal death. The objective of this study is to design an automated system with an ability to improve the detection process in order to perform advanced recognition of the disease. The diagnosis techniques include: X-rays, MRI, CT images etc. X-ray is the common and low-cost technique that is widely used and it is relatively available for everyone. Rather than new techniques like CT and MRI, X-ray is human dependable, meaning it needs a Doctor and X-ray specialist in order to determine lung cases, so developing a system which can enhance and aid in diagnosis, can help specialist to determine cases in easily.

Weight Sensors Based Human Walking Step Recognition System: Implementation and Statistical Evaluation

Sahar Salman Mahmood

Engineering and Technology Journal, 2015, Volume 33, Issue 8, Pages 1876-1889

It is well known that with the growing of the humanity and all the development in technologies, there is an increasing in need for recognition systems. These systems can recognize people from distinct characteristics in which these are unique for each one individually. The researchers went to the finger print and eye recognition methods to be adopted as the dominated approaches, yet, these methods suffers from numerous health risks due to diseases transferring. Therefore, the walking step recognition method has been adopted recently. This is because each person has different walking style from others.
This paper proposed a human walking step recognition system that adopts group of weight sensors distributed amongst carpet. The reading data from sensors has been transmitted to the information center for processing. The data is transmitted through out a wired sensor network that includes sensor nodes and sink node. The latest node is used to collect the reading data from the sensor nodes. At the information center, the received data is processed using the proposed recognition algorithm. This algorithm gives two decisions; either matching with full information about the intruder or no matching. On the other hand, the proposed system has been designed and implemented using MATLAB simulator. Throughout this simulator, a database matrix is generated randomly to cover all the probability of walking step patterns available for humans. This matrix consists of three dimensions; one for users, second for sensor readings (walking patterns), and third for tries. Each user records numerous walking patterns by passing over the designed carpet several times at different modes just to cover the slightly changes in walking style in terms of modes. It is important to note, that the carpet include the sensors in between of two layers.
The simulation results show the successful performance of the proposed system with high efficiency and recognition accuracy. In addition, statistical analysis has been obtained using sampling theorem by adopting sample of 100 employees at University of Technology. Thisis done by distributing a questioner form over the employees to evaluate the acceptation of the proposed system by people in terms of health issues and ease of use. The outcome results show high ratio of accepted people in comparison with rejected.

Neuro-Snake Pattern Recognition And Classification Using Gradiant Vector Flow (Gvf And Hnn)

Wissam Hassan Ali

Engineering and Technology Journal, 2009, Volume 27, Issue 5, Pages 973-982

The most popular applications of Hopfield neural network algorithm (HNN) are
pattern recognition and classification. But the HNN has some limitation like the local
minima (oscillation) problem. In this paper a novel method of combining an active
contour (snake) and an artificial neural network to behave together as pattern recognition
and classification is presented. The approach used the technique of the gradient vector
flow (GVF) that locate the boundary of target pattern (image) then pass it to a classifier
built by Hopfield algorithm to classify it according to one of the storage pattern. The
snakes can find the boundaries of objects so it is very accurate to take the shape of the
object wanted, that will eliminate the noise from the original image and reduce the bit
error rate of the Hopfield network to 0.215 and overcome the oscillation state in
recognition of the entered pattern. MATLAB 7 program have been used for the
simulation of the active contour and the pattern classification.