Search

Early detection of all kinds of lung nodules with different characters in patient’s medical modality images is the best acceptable remedy to save the life of lung cancer sufferers. Even though day by day the prominence of Computer-Aided Detection/Diagnosis (CADe/x) systems have been increasing as a part of medical routine in detection of different types of lung nodules, but detection rate performance depends on accuracy of lung parenchyma and nodule segmentation procedures. Segmentation of Juxta-Vascular nodules attached very complex. In this paper new fully automated CAD system is developed to detect and classify Juxta-Vascular nodules. In proposed methodology, lung parenchyma is segmented using iterative thresholding algorithm and lung nodules are segmented using proposed modified region growing algorithm. Since in vascular nodules, separation of blood vessel from nodule is difficult as intensity feature of attached blood vessel and nodule is same. Two new methods nodule segmentation method and vessel removal based on multi features to separate the vascular nodule part from the attached blood vessels are developed. To achieve the higher nodule-vessel separation accuracy, nodule-vessel attached region is refined. Validation of proposed method is performed on LIDC-CT lung images. A fully automated method segments the vascular nodules with less computational time and high accuracy.

In the recent era, the research interest has increased among different computer and communication societies towards microarray gene expression detection and profiling. Despite of having a wide range of applications, the more emphasize has been kept towards cancer and its sub-type classifications. It has been seen in the past that the existing data mining approaches impose more cost of computation during pattern discovery and correlation establishment. Thereby, it is needed to address this shortcoming to strengthen the reliable cancer detection and classification process cost-effectively. An efficient machine learning tool has a better scope of optimization towards handling margin and error factors. Addressing this open research issue, the current study has come up with a novel method namely Optimal Framework for Microarray Data Classification (OFDMC) which incorporates Eigenvector decomposition to perform dimension reduction of gene expression data without compromising the complexity and accuracy aspects. The study also validates the performance of the proposed system by introducing a numerical analysis.

The significant contribution of the clustering algorithm for diagnosis of the clinical condition through medical data consideration is must in the healthcare sector. The currently existing techniques implement the Fuzzy Logic in clustering and have been found by research gap which describes that less focus on the medical data clustering. Thus, this paper introduced a novel algorithm where the enhancement of fuzzy logic is performed to achieve better computational ability in the processing of highly complex medical data such as microarray data. The introduced algorithm is implemented for disease diagnosis and classification. The outcomes of the proposed algorithm are compared with recent approaches like the genetic algorithm, support vector machine (SVM), and artificial neural network (ANN). On analyzing these comparative results found that the proposed clustering model achieved significant performance in response time and classification of disease with better accuracy.