Final

Project Proposal Instructions

Due: Project proposals are due Wednesday 3/14/2012 by midnight in the box outside Dr. Bangerters office (or via email to me, if you hand it in between 5pm and midnight). J Description: A full 25% of your grade in this class will be based on your final project. Significant long-term learning can come from the process of researching a topic in depth, possibly performing simulations or derivations, and then preparing to present your results or what you have learned in a way that teaches others. What you get out of the final project (and ultimately this course) will be strongly correlated with how much thought and effort you put into the final project. I am leaving the topic and structure of the final project very open to encourage each of you to focus on an area of interest to you. I have provided a list of possible topics at the end of this document, which we will discuss in more detail in class today. The project should (obviously) relate to medical imaging and image reconstruction, and should require a minimum effort of approximately 20-25 hours per team member over the final 4 weeks of class. Some acceptable sample topics would include: In depth study of the physics, instrumentation, or engineering background/challenges associated with one of the imaging modalities. (This would be particularly useful if you are interested in one of the modalities that weve skimmed through or havent covered in class, e.g., everything but MRI, including ultrasound and nuclear imaging.) Implementation and comparison of some image reconstruction techniques. For example, a group might choose to implement a simple filtered back-projection technique for CT, a filtered back-projection technique that compensates for fan beam geometry, and a 2D Fourier transform based technique that grids the data onto a Cartesian grid for reconstruction, and then compare each technique for computational efficiency and reconstructed image quality. I can help provide sample data for these types of projects. A deep dive into a particular current research topic, usually focusing on a paper or group of papers in the literature. This deep-dive may or may not include simulations, depending on the topic and time constraints. I encourage you to work in groups of two, although single person projects are okay. (If you have a proposal idea that has well-delineated areas for 3 people, come talk to me and Ill consider it.) Project Proposals: Your proposal should include a title for your project, the names of students to be involved, and 2-3 paragraphs describing the work that will be performed (at a very high level). This project proposal is not binding, since as you learn more you may want to slightly change the focus of your project. However, by two weeks into the project you should have a very clear outline of exactly what you are

attempting to do. If the project involves a deep-dive into specific articles or papers, please provide a list of these references in the project proposal. If any simulations or experiments are to be performed, outline what they are, how they will be accomplished, and what results are expected. Please prepare these electronically, and limit them to one page or less. I am available to give you ideas. Please come to office hours if you would like to discuss. Final Deliverables: You will be required to prepare a 10-12 minute slide presentation to be presented at the end of the semester (at a time to be determined). A 3-5 minute question period will follow your presentation, when you will field questions from the class and me about your topic. These should be polished and well-prepared presentations, equivalent to what you might prepare for a scientific conference. In addition to the final presentation, you will need to hand in a 2-3 page final report on your project. This should include enough detail to help me ascertain the work that was performed. Any simulation or experimental results should be included in the report. If the project consisted of a deep-dive into a particular topic area, the final report should have more of a review paper feel to it. If working in a group, please indicate which sections of work were performed by whom. Possible Final Project Topics: In-depth overview of one of the medical imaging modalities we havent covered in detail (PET, SPECT, ultrasound). This could also include a simulation of the technique or implementation of image reconstruction from sample data. I can help you obtain sample data if needed. Implementation and comparison of CT reconstruction schemes (backprojection, filtered backprojection, filtered fan-beam backprojection, and Fourier-domain gridded reconstruction. Implementation and comparison/analysis of partial k-space acquisition schemes that exploit conjugate symmetry (Homodyne, POCS, and phase-corrected conjugate synthesis). Parallel imaging algorithms in MRI: There are a large number of parallel imaging algorithms in MRI that exploit data from multiple receive coils to allow data to be undersampled in k-space (thus speeding up the acquisition). Implement one of these (SMASH, GRAPPA, etc.) and compare to the gold standard SENSE. Non-cartesian parallel image reconstruction: Like the previous one, but using non-cartesian trajectories such as spiral or PR. The aliasing is now no longer coherent. Off-resonance correction: As you have learned in class, there are a variety of sources for off-resonance across our object. This can introduce artifacts, shifts, or blurring into our image, depending on pulse sequence and trajectory. In this project you could implement reconstruction of a B0 field map, do multi-frequency reconstructions, and maybe implement an autofocus reconstruction technique. Gridding for non-cartesian reconstruction, including pre-weighting density correction, k-

space oversampling, and deapodization. Compressed Sensing: Compressed sensing and reconstruction of images from vastly undersampled datasets is a very hot topic in MRI right now. In this project, you could implement a simple compressed sensing reconstruction, explore the tradeoffs between different acceleration factors and image quality, and/or explore different undersampling schemes. Fat suppression in MRI: There are a variety of different fat suppression reconstruction techniques. These include phase-based methods like 2-point and 3-point Dixon and IDEAL, as well as some more complicated reconstructions based on SSFP imaging. I can provide sample data for any of these if you would like to try implementing some fat suppression techniques. In-depth overview of diffusion-weighted and diffusion-tensor MRI Functional MRI and the BOLD contrast Overview of current research topic in hybrid CT/MRI systems (or PET/MRI systems) MR-guided high-intensity focused ultrasound Coronary CT angiography Imaging flow with MRI and non-contrast enhanced MR angiography Dynamic contrast-enhanced MRI MR spectroscopy (forming an MRI image, but acquiring NMR spectroscopic data for each voxel) Temperature imaging with MRI MR sequence optimization problem: Weve looked at simulating signal levels of several sequences in class and homework problems. You could expand on this work to optimize a certain sequence for a given application. (For example, optimizing a steady-state inversion- recovery sequence to null cerebrospinal fluid while maximizing white matter/tumor contrast)