Projects Before Ph.D.

      This is a digital pathology project directed by Dr. Michael Miller aimed to map the microstructural features of healthy and pathological brain tissue to understand the mechanism of AD. I was in charge of biomarker segmentation and quantitative characterization, advisered by Dr. Daniel Tward. Two kinds of biomarkers, tau tangles and amyloid plaques, were segmented from histopathology images using a convolutional neural network called UNET. Each segmented biomarker was characterized using their size, orientation and roundness. Feature map reconstruction and multiresolution analysis followed.

Eye movement is considered as significant reference for assessing the cognitive development of chidren. Given the drawback of traditional eye movement diagnosis, where data was not recorded and diagnosis highly depended on physicians' experience, high-frequency infrared cameras were used to capture subtle features of children's eye movement in clinical tests. Time-series eye position data was derived after a set of image processing and was analyzed by comparing two-eyes coordination and relations of eye-movement and vision stimulus.