A Patch Based Approach for the Segmentation of Pathologies: Application to Glioma Labelling
Journal Title: Elysium Journal of Engineering Research and Management - Year 2016, Vol 4, Issue 2
Abstract
This paper proposes a novel and generic approach with multi-atlas patch-based voting techniques for addressing the fullyautomatic segmentation of brain tumors. To avoid the local search window assumption, conventional patch-based framework is applied through many procedures like, improved training dataset in both purity of label and statistics of intensity, augmented features for guiding the nearest-neighbor-search indirectly, patches of multiscale, invariance to cube isometries, votes stratified with respect to labels and cases. A probabilistic method allocates the interest region enclosed with the volumes of high probability tumor. It can permits the algorithm for achieving highly best running time even it has less processing power and resources. The evaluation of this method was done by the Multimodal Brain Tumor Image Segmentation challenge datasets. The results of this proposed method are attained with limited learning stage even it avoids the risk of overfitting.
Authors and Affiliations
Balakumar B. , Raviraj P. , Nithu S. V.
HADOOP MAP REDUCE ENHANCEMENT THROUGH IMPROVED AUGMENTED ALGORITHM
The MapReduce model is implemented by using open-source software of Hadoop. A number of issues faced by Hadoop to achieve the best performance. A serialization barrier requires to achieve the best performance whi...
Sleep Apnea Detection using Heart Rate Variability and Classifiers
The Obstructive Sleep Apnea (OSA) or Obstructive Sleep Apnea Syndrome (OSAS) is the sleeping disorder causing the pause in breathing process or a very low breathing rate while sleeping. The standard technique for analyzi...
AN EFFICIENT AND SECURE POLICY UPDATE OUTSOURCING SCHEME FOR BIG DATA STORAGE IN CLOUD
Store the big data in cloud is a best and an effective option due to its high volume and velocity. Generally, the cloud has the capability to store the big data and to process the high volume of user access requests. In...
Enhancement of Security and Usability Standards with a Novel Storage Optimized Honeyword Generation Approach
Among the modern security threats on password based authentication techniques, the brute force computation is the one that performs the inversion of hash values. Several technologies have been developed for the computati...
AN INTEGRATED FRAMEWORK FOR PARALLEL TRAINING AND CLASSIFICATION OF ECG SIGNAL
Recently, various techniques are developed to analyze and classify the Electrocardiograph (ECG) signal to know the threaten results. Many real time development tools such as VLSI, DSP, etc., are availabl...