Gyemin Lee, Electrical Engineering:Systems PhD student, received the Pascal2 Best Student Paper award at the 28th International Conference on Machine Learning (ICML 2011) Workshop on Unsupervised and Transfer Learning. The paper, "Transfer Learning for Automatic Gating of Flow Cytometry Data," was co-authored by Dr. Lloyd Stoolman (U-M Department of Pathology) and Lee's advisor, Prof. Clayton Scott.
Mr. Lee described the research:
"Flow cytometry is a technique for rapidly quantifying physical and chemical properties of large numbers of cells. It is widely used in many bio-medical and clinical laboratories to diagnose blood-related diseases such as leukemia and lymphoma. In clinical applications, flow cytometry data must be "gated'' to identify cell populations of interest. While this process is performed manually in most clinical settings, automation would greatly facilitate the process. In this work, we propose an automatic gating method that can leverage existing data sets previously gated manually by experts, while accounting for biological variation."
The authors' primary motivation is to apply their method to hematopathology, the study of blood-related diseases. They case flow cytometry auto-gating as a novel kind of transfer learning problem. By combining existing ideas from transfer learning, explains Lee, together with a low-density separation criterion for class separation, this new approach can leverage manually derived datasets to generate an automatically generated dataset.
Posted August 15, 2011 by
EECS/ECE Communications Coordinator
firstname.lastname@example.org or 734-936-2965
Related Topics: Health Machine Learning Medical diagnosis Scott, Clayton D. Signal and Image Processing