Archive for September, 2017

Seminar: Advancing PET Image Reconstruction with Machine Learning


Dr. Jing Tang received her PhD degree in Electrical Engineering from the University of Illinois at Urbana-Champaign and had her postdoctoral training in Radiology at the John Hopkins School of Medicine. She was employed as a principle imaging physicist at the Philips Healthcare before joining the Department of Electrical and Computer Engineering at Oakland University, where she currently holds an appointment as an associate professor. She has received a number of federal awards to support her research on the development and application of PET/MR imaging techniques. Dr. Tang was the guest speaker at a lecture organized by the MGH Gordon Center. Below is her presentation’s summary.

Machine learning has shown its promises to empower medical imaging, mainly in image analysis. The talk presented Dr. Tang’s investigation of integrating machine learning techniques into the other major component of medical imaging, image reconstruction. Dr. Tang proposes to advance maximum a posteriori (MAP) PET image reconstruction using two learning based methods.

The first one is through building a dictionary learning (DL) based prior and the second by designing an artificial neural network (ANN) to enhance smoothness MAP reconstructed images from different regularizing weights. The DL-based MAP reconstruction technique forms its regularization using the sparse representation of PET images based on dictionaries trained from various images. The ANN enhancement scheme solves the regression problem of mapping between multiple reconstructed image versions and an enhanced image through learning from examples.

According to Dr. Tang, both techniques have a potential for clinical applications due to their capacity to improve conventional MAP methods in reconstructing simulated and patient PET data.

Dr. Jing Tang delivering his presentation at the MGH Gordon Center

Improving PET Quantitation with Denoising, Motion Compensation, and Deblurring


This article was published in the Nuclear & Plasma Sciences Society newsletter of September 2017

Positron emission tomography (PET) enables 3D visualization of vital physiological information, e.g., metabolism, blood flow, and neuroreceptor concentration by using targeted radioisotope-labeled tracers. Quantitative interpretation of PET images is crucial both in diagnostic and therapeutic contexts. As a result of its unique functional capabilities, PET imaging plays a pivotal role in diagnostics and in therapeutic assessment in many areas of medicine, including oncology, neurology, and cardiology. Accurate quantitation requires correction of PET raw data and/or images for a number of physical effects. These include attenuation correction, randoms and scatter correction, subject motion correction, and partial volume correction. We have developed a range of techniques that address the PET denoising, motion compensation, deblurring problems. Several of these methods greatly enhance the quantitative capabilities of PET particularly by incorporating information from an anatomical imaging modality such as magnetic resonance imaging (MRI).

Faced with a fundamental tradeoff between radiation dose and image noise, PET data is inherently noisy. The high levels of noise in PET images pose a challenge to accurate quantitation. This issue is particularly well pronounced at the early time frames of dynamic PET images, which are usually short to capture rapid changes in tracer uptake patterns. In order to improve image quality and quantitative accuracy, statistical image reconstruction algorithms model the Poisson characteristics of PET data and employ numerical optimization algorithms to solve the corresponding optimization problem [1, 2]. Common reconstruction procedures, such as ordered subsets expectation maximization, are therefore routinely followed by a post-filtering step for denoising the reconstructed image. A range of strategies have been proposed for post-reconstruction denoising of both static and dynamic PET images [3, 4]. In recent years, image denoising based on non-local means (NLM) has become popular [5]. Unlike conventional neighborhood filters, which use local similarities, in this technique, the search for voxels similar to a given voxel is no longer restricted to its immediate vicinity.

Full article in PDF

From Academia to Industry and Back Again: Transitioning Technology & Changing Employers


Dr. Lee Josephson is a chemical biologist and an inventor on more than 20 patents in the fields of medical imaging. He has been an NIH grant Principal Investigator and a Vice-President and Chief Scientific Officer of a diagnostic drug pharmaceutical company. He was the guest speaker at a lecture organized by the MGH Gordon Center.

Dr. Josephson's discussed the transition from academic to industrial employers, including some of the different types of employers in the Boston area. His presentation also covered the difference between what companies and academic institutions look for in new hires. Dr. Lee used his personal case examples of success and failure to describe issues that arise during the transition from academic science to industry.

Dr. Lee Josephson delivering his presentation at the MGH Gordon Center

Seminar: Collaborative Innovation at Partners HealthCare


Dr. Glenn Miller is Innovation’s Market Sector Leader for Radiology, Anesthesiology, Neurology, Neurosurgery and Psychiatry at Partners Healthcare. He has nearly 30 years of experience in the clinical laboratory industry and personalized medicine.

The role of Partners Healthcare (PHC) Innovation is to assist investigators in the sustainable support for their work, intellectual property protection of their inventions and commercialization of their discoveries. Dr. Miller's lecture discussed how PHC Innovation can help investigators protect and commercialize their scientific inventions through different mechanisms such as patents, licenses, industry collaboration and venture investments.

Dr. Glenn Miller delivering his presentation at the MGH Gordon Center