PhD Defense: "Probabilistic Graphical Models for Contour Tracking and Segmentation in Electron Microscopy Images"

Min-Chi Shih

December 1st (Monday), 11:00am
Harold Frank Hall (HFH), Room 1132

Automatic 3D reconstruction of neuronal circuitry in biological images is key to discerning cellular ultra structure. This talk covers my research work on the problem of 3D reconstruction of neuronal cells in electron microscopy images, using probabilistic graphical models. First I will cover a 1D hidden Markov model-based contour tracking algorithm for a single or a few neuronal processes involving topological changes. In this method, uncertain segments with lower likelihoods are detected, and then a few hypothetical arcs are created to perform contour refinement to enable the discovery and corresponding tracing of topology changes. Secondly I will cover a method, wherein a two-dimensional hidden Markov model is utilized for tracing a large number of cells by modeling the problem as a pixel labeling task. This method leverages the concept of spatially adaptive states, wherein the state-space at each pixel is locally extracted to be a small subset of the full state-space. This local adaptation of states, not only reduces the computational complexity significantly, but also improves the segmentation accuracy. While the first contour tracking algorithm precisely locates cell boundaries, the second pixel-labeling-based algorithm easily scales to a large number of cells, and hence represent two complimentary techniques that together offer significant advancement on the problem of 3D reconstruction of neuronal cells.

About Min-Chi Shih:

Min-Chi Shih received his B.S. degree in communication engineering from National Chiao Tong University, Hsinchu Taiwan in 2005, and received his M.S. degree in communication engineering from National Tsing Hua University, Hsinchu Taiwan in 2007. He entered UCSB in 2009 and joined the Signal Compression Lab in 2010. He is currently a Ph.D. candidate in the department of electrical and computer engineering. During summer 2013, he spent three months interning at Intel Labs, Hillsboro Oregon. His research interests include object tracking, video segmentation, and object detection and recognition, using various probabilistic graphical models.

Hosted by: Professor Kenneth Rose, Signal Compression Lab