Fangxu Xing, Ph.D.

Assistant Professor of Radiology
Harvard Medical School
Assistant Professor of Radiology
Harvard Medical School
Research Scientist
Radiology, Massachusetts General Hospital
Assistant Investigator
Gordon Center for Medical Imaging, Mass General Research Institute
M.S.E. Johns Hopkins University 2011
Ph.D. Johns Hopkins University 2015
Bachelor of Science Tsinghua University 2009
biomechanics; brain cancer; brain mri images; cardiac biomarkers; cardiac imaging; classification; deep learning; glioma; image generation; image segmentation; machine learning; machine learning for healthcare appplication; medical image analysis; medical imaging; mild traumatic brain injury; motion; sarcoma; speech; speech disorders; statistical analysis; time and motion studies; tongue; tongue diseases

Dr. Xing has been pursuing a career that aims in helping the research and development of medical science from the medical imaging perspective. He has a broad background in mathematics and physics which merged well with his doctoral research at Johns Hopkins University, during which he developed novel tissue motion analysis techniques to help processing and interpreting cardiac, speech, and traumatic brain injury image data, resulting in a sequence of effective motion analysis methods to reconstruct spatiotemporal tissue deformations from sparsely acquired MRI data and enable accurate quantitative clinical image processing while maintaining efficiency in data acquisition.

Since joining the Gordon Center for Medical Imaging at Harvard and MGH, his research has mostly focused on tissue motion analysis and dynamic atlas construction. Besides cardiac, brain, and speech data, he expanded the range of applications to the area of PET imaging, velopharyngeal tissue tracking, fast dynamic MRI, amyotrophic lateral sclerosis (ALS) diagnosis, sleep apnea studies, etc. He has 1) developed a fully dynamic anatomical and functional atlas of the human tongue using multimodal MRI, and 2) quantitatively revealed internal speech muscle’s cooperation pattern during tongue deformation using atlas-based correlation analyses. Since then, his group have been pioneering structural and functional atlasing of the internal tissue during motion, and a large number of publications and presentations in top journals and conferences have been produced. He was recognized and awarded in best paper competitions in a few conference occasions. Based on his work’s reputation, many invited talks were given in both national and international venues.

With the rapid surging of AI and machine learning technology, he has incorporated various deep learning techniques in his most recent research. He published novel methods to achieve brain tumor image inpainting using deep neural networks and a few proceeding papers on multi-modality dynamic MRI image synthesis using GANs (generative adversarial networks). Moreover, he has been contributing to the professional society by serving as peer reviewer for various journals and scientific conferences.

Since the beginning of his faculty career at Harvard Medical School, he has directed his daily routine more into the duty of education. His previous experience in college and graduate school teaching bridged naturally to his current teaching duties, hosting the HST medical imaging course as associate director. As part of his regular research activity, he was invited to a few occasions both nationally and internationally to teach his new research findings in hosted seminars. Besides, he kept mentoring students and fellow peers both within HMS and from other collaborating institutions such as Johns Hopkins University, University of British Columbia, East Carolina University, University of Illinois Urbana-Champaign, etc., resulting in multiple co-authored publications, proceedings, and international presentations.

Research interests
  • Deep learning: network and processing algorithm development for image analysis/reconstruction
  • Image segmentation/classification/registration/analysis
  • Cardiac study: myocardium motion and strain analysis from dynamic MRI
  • Speech and Language Imaging: motion analysis for human tongue and velopharyngeal muscle mechanics from real-time MRI
  • Brain biomechanics: motion analysis in traumatic brain injuries
  • Statistical image analysis and image label fusion
  • Various applications involving image data processing and analysis