Biomedical Image Processing
Segmentation of Ultrasound Images
Supervised by Prof. Dr. Kazi Khairul Islam, Professor, Department of Electrical and Electronic Engineering, IUT.
Collaborated by S. Kaisar Alam, Research Scientist of Riverside Research’s Lizzi Center for Biomedical Engineering, USA.
- Segmentation is an important step of CAD systems. Both automation and accuracy of segmentation is crucial. Automation of segmentation is important because it facilitates the complete automation of the CAD system. A fully automatic CAD can minimize the effect of the operator-dependent nature inherent in ultrasound imaging and make the diagnosis process reproducible.
Our main objective is to propose an algorithm that called Projected Empirical Segmentation (PES) that locates boundary regions in ultrasound images. Beginning with an initial crude manual segmentation in an initial frame of the video, the method projects three regions (the internal region, the external region, and an intermediate region whose identity is to be determined). Using a set of statistical features, the method checks each point in the intermediate region to see if it is more like the interior or the exterior, and so locates the boundary of the tissue in the biomedical ultrasound images. The method then repeats automatically, operating throughout succeeding frames without manual intervention.
Cancer Diagnosis using Ultrasound Imaging
Supervised by S. Kaisar Alam, Research Scientist of Riverside Research’s Lizzi Center for Biomedical Engineering, USA.
- Currently, in the field of medical diagnosis, ultrasound is responsible for about one in five of all diagnostic images. Ultrasound elastography is emerging with enormous potential as a medical imaging tool for effective discrimination of pathological changes in soft tissue. It maps the tissue elasticity or strain due to a mechanical de-formation applied to it.
Our main objective is to propose an algorithm that people will find more economic and fruitful for regular check-ups of breast cancer, taking into account mammography, biopsy, and ultrasound imaging with computer aided decision making. It will focus more on ultrasound imaging because it holds great promise in terms of low cost equipments, null harmful radiation, and better possibility in finding malignancy in younger women. Furthermore, people do like non-invasive diagnosis a lot more than biopsy. Adding to that, biopsy is expensive and in developing countries not viable to be available at all places. So the trade-off of accurate diagnosis against more economy/comfortable diagnosis has to be studied. We will use thresholding of quantified acoustic features to suggest when it is safe to avoid biopsy, when to go for follow-up check up, and when to go for biopsy.