Students: James Williams II
Faculty Advisor or Community Project Lead: Martin Cenek
Summary:
My project proposed object segmentation using level set methods and a three-dimensional model to improve cellular object classification accuracy and reliability in EM images. The publicly available dataset of 30 serial sections of Drosophila larva neural tissue will be used to develop the 3D model. A level set method can then be applied taking advantage of the information of all slides to improve object recognition and segmentation.
Quick diagnosis and treatment can be the difference in life or death of a patient. The leading cause of death from Malaria is the delay of a proper diagnosis. Traditional methods are problematic because recognizing patient signs and symptoms are difficult. Laboratory technologies to improve diagnosis are necessary. Prompt and accurate malaria diagnosis would alleviate suffering, but also decreases chances it spreads within a community. An expert can detect malaria parasites in blood samples rather successfully but it takes too much time. For instance, datasets produced by medical imaging techniques, such as electron microscopy (EM), can be over terabytes (1012) in size. Manual analytics of a dataset this size are infeasible and powerful computer analytics are not available worldwide. This project aims to explore the use of Level sets to improve object segmentation on medical images while maintaining quick operational runtime.
Document(s):
Comments are closed