Ashis Kumar Dhara
visiting researcher at Department of Information Technology, Division of Visual Information and Interaction
Keywords: biomedical image analysis
Also available at
Dr. Ashis Kumar Dhara is presently serving as a post-doctral researcher at Department of Information Technology, Division of Visual Information and Interaction in Uppsala University, Sweden. He has expertise in developing computer-aided diagnosis system and content-based image retrieval system for clinical assistance. He received Ph.D. from Indian Institute of Technology Kharagpur, India in 2016. He received the B.Tech. and M. Tech.degrees in electrical engineering from University of Calcutta, Calcutta, India, in 2005 and 2007 respectively. His area of interest is biomedical image analysis, Pattern recognition and Machine learning.
Current Research Work: Diabetic retinopathy (DR) is a major threat to public health across the world. It could lead to irreversible blindness if not treated at an early stage. An early diagnosis of DR and proper treatment can prevent blindness in more than 50% cases. However, the number of retinologists is very small in relation to the world population. In this situation, the use of an efficient screening tool is the only way for diagnosis of DR.
I have started working for the development of a DR screening tool. The purpose of this screening tool is to detect retinopathy, determine its severity, and to decide which patients require referral for further investigation and possible treatment. The development of an automated screening system for staging of DR could lead to a highly effective way of reducing the burden on screening services by filtering out normal retinal fundus images from DR images as well as by determining the stage of DR. Such a system could lead to the reduction of cost of treatment by early detection of DR. Different tasks associated with the development of DR screening tool are:
- Development of a robust preprocessing technique for improvement of image quality
- Detection of several anatomical structures such as optic nerve head, fovea, macular region, and retinal blood vessels etc.
- Detection of several pathology such as exudates, micro-aneurysms, and hemorrhages
- Feature-based representation of several pathological patterns and development of a robust feature set for characterization of different signs of DR
- Pattern classification and development of a scheme for staging of DR
Ph.D. Research Work: Lung cancer accounts for the highest number of cancer-related deaths as compared to other types of cancer in both men and women. Several studies show that screening of lung cancer can substantially reduce the mortality rate. Accurate interpretation of pulmonary nodules is essential for the diagnosis of lung cancer and subsequent plan of treatment. Trainee radiologists have to depend on experienced professionals for interpretation of pulmonary nodules, the potential manifestations of lung cancer. The lack of time of experienced radiologists is the major bottleneck for such traditional learning procedure.
During my Ph.D., I have worked for the development of a self-learning tool for radiologists using content-based image retrieval technique. The radiologists could find similar nodules using this tool. Different steps associated with the development of the self-learning tool are the segmentation of pulmonary nodule, representation of nodule using machine level features, computation of similarity, and retrieval of similar nodules. The pulmonary nodules are segmented using the semi-automated technique. The segmentation technique is applicable for all types of pulmonary nodules based on their internal texture (viz. solid, part-solid and non-solid) and external attachment (viz. juxta-pleural and juxta-vascular). The efficacy of several combinations of shape-based, margin-based and texture-based features are studied to improve the accuracy of retrieval. The retrieval system can be used with minimal user intervention.
Please contact the directory administrator for the organization (department or similar) to correct possible errors in the information.