Bio-medical Image Retrieval using Various Statistical Methods
Vinutha N1, Sandeep S2, P DeepaShenoy3, Venugopal K R4

1Vinutha N *, University Visvesvaraya College of Engineering, Bengaluru, India.
2Sandeep S, Practo Technologies Private Limited, Bengaluru, India.
3P DeepaShenoy, University Visvesvaraya College of Engineering, Bengaluru, India.
4Venugopal K R,Bangalore University, Bengaluru, India
Manuscript received on November 22, 2019. | Revised Manuscript received on December 15, 2019. | Manuscript published on December 30, 2019. | PP: 2669-2676 | Volume-9 Issue-2, December, 2019. | Retrieval Number:  B4231129219/2019©BEIESP | DOI: 10.35940/ijeat.B4231.129219
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (

Abstract: In recent, the healthcare sectors rely more on imaging technologies for early detection and diagnosis of the disease. But, the abundant images obtained from these imaging technologies have complex disease patterns associated with them and thus an expert requires more time to analyze and arrive at the decision. Hence, the image retrieval techniques have a significant role to assist the experts by retrieving the most similar images existing in the database and also help them to compare a new scan of the patient with the top matched images and arrive at the quick decision during the diagnosis of a patient. So, we have performed our studies on the two-dimensional structural Magnetic Resonance Imaging of the Open Access Series of Imaging Studies dataset. The collected images are preprocessed and categorized into different groups based on the ventricular region of the brain. After the categorization, we employ second and higher-order statistical approaches to extract the textural features. Then the computed textural features of the images existing in the dataset are compared with the textural features of a query image to retrieve the top matched images using similarity distance as the metric. Then the image retrieval performances of the proposed hybrid based statistical methods are measured. The obtained results shows that the combined features of Gray Level Co-occurrence Matrix and Law’s Texture Energy Measure attains the highest precision across the categorized groups of a dataset and it achieves 80% precision for Group1, Group2 images and 60% precision for Group3 images.
Keywords: Alzheimer’s Disease, Content-based Image Retrieval, Magnetic Resonance Imaging, Statistical Methods, Textural Features, Ventricle.