Multi-Scale Blood Vessel Detection and Segmentation in Breast MRIs

An algorithm is proposed to perform segmentation of blood vessels in 3D breast MRIs. The blood vessels play an essential role as an additional tool to detect tumors. Radiologists use a maximum-intensity projection for the exposure of vasculature. The breast is a challenging organ in detecting vascular structures, because of noise bias and presence of fat tissues. There are several existing algorithms for the detection of blood vessels in MRI images, but these usually prove insufficient when it comes to the breast.

Our algorithm provides a three-dimensional model of the blood vessels by utilizing texture enhancement followed by Hessian-based methods. In addition to this, we tackled blood vessel completion by employing center line tracking, where the seeds are the end points of detached blood vessels found through skeletonizing. The results were compared to the manually segmented golden models defined by radiologists in 24 different patients, which yielded an 86% Sensitivity to the ground truth and 88.3% specificity. It appears that with the application of mass detection as the last step, our algorithm provides a helpful tool for tumour enhancement and automated detection of breast cancer.

Computational Drug Design and Molecular Dynamic Studies-A Review

Drug designing and molecular dynamic studies were an intense, lengthy and an interdisciplinary venture. At present, a new approach towards the use of computational chemistry and molecular modeling for in-silico drug design. Computational in-silico drug design skills are used in bioinformatics, computational biology and molecular biology.

Drug designing using in-silico methods is cost effective in research and development of drugs. Currently, a vast number of software’s used in drug design. In-silico drug designing and molecular dynamic studies can be performed by using different methods namely homology modeling, molecular dynamic studies, energy minimization, docking and QSAR etc. By using in-silico drug designing we can produce an active lead molecule from the preclinical discovery stage to late stage clinical development. The lead molecules which are developed will help us in selection of only potent leads to cure particular diseases. Therefore in-silico methods have been of great importance in target identification and in prediction of novel drugs.

Multi-Scale Blood Vessel Detection and Segmentation in Breast MRIs

An algorithm is proposed to perform segmentation of blood vessels in 3D breast MRIs. The blood vessels play an essential role as an additional tool to detect tumors. Radiologists use a maximum-intensity projection for the exposure of vasculature. The breast is a challenging organ in detecting vascular structures, because of noise bias and presence of fat tissues. There are several existing algorithms for the detection of blood vessels in MRI images, but these usually prove insufficient when it comes to the breast. Our algorithm provides a three-dimensional model of the blood vessels by utilizing texture enhancement followed by Hessian-based methods. In addition to this, we tackled blood vessel completion by employing centerline tracking, where the seeds are the endpoints of detached blood vessels found through skeletonizing.

Download PDF

Decision Tree Eliminates the Computational Complexities of Big Data Processing

Hadoop is one of the reputed general purpose computing platforms used to process big data. Mapreduce is the Hadoop project’s main processing engine that provides a framework for distributed computing. This is generated from the combination of ‘Map’ and Reduce concepts in the functional programming. This functional programming model hides the entire complexities related to distributive computing nodes, so that the developer can focus on the implementation of the Map and Reduce functions.

Decision Tree is one of the most efficient ways of classification and decision making. Decision tree consist of root node and decision nodes and the data related to a training module for example can be divided based on the measurable, functional attribute. The entire data thus can be divided into number of partitions based on the attributes fixed. If the samples within a partition belong to a single class, then the algorithm gets terminated automatically. Otherwise the division process continues until it eliminates all the non-identical samples. The samples that don’t fit into the attributed partitions are labeled as unknown category.