An Artificial Intelligence Method for Auto-Compressed Sensing and Blind Deconvolution in Magnetic Resonance Imaging Data of Shoulder Muscle Metabolism

该奖项是作为2021财年数学和物理科学上升博士后研究奖学金，MPS-Ascend计划的一部分。Talon约翰逊被授予该奖学金，在赞助科学家Jimin Ren和Anke Henning的指导下，在德克萨斯大学西南医学中心进行数学科学的研究和教育计划，包括其他学科的应用。这是一个旨在开发磁共振成像（MRI）的计算图像重建方法的研究项目。 约翰逊计划设计压缩感知方法，该方法能够从受受试者运动影响的MRI测量结果中重建图像数据，例如肩部肌肉代谢MRI成像中的运动相关运动。 沿着这项研究，约翰逊将参与活动，通过招募德克萨斯大学阿灵顿分校的学生，并与亚特兰大大学中心数据科学计划合作，扩大代表性不足的少数民族在数学科学领域的参与。这项研究旨在通过完成三个研究目标来提高图像精度和质量：1）基于用于压缩感测的l1-magic和用于固定核函数的去卷积的l2最小化，开发和改进用于使用l1-l2最小化方案重建用于模糊MRI成像的压缩感测数据的方法，2）开发用于从模糊压缩MRI数据集中自动识别任何未知核函数的AI方法，以实现任何模拟MRI数据的同时压缩感知-盲去模糊，3）完善自动驾驶的人工智能方法，压缩传感和盲去模糊的实验数据在MRI成像的肩部肌肉代谢和提高成像精度。该奖项反映了NSF的基金会的使命是履行其法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This award is made as part of the FY 2021 Mathematical and Physical Sciences Ascending Postdoctoral Research Fellowships, MPS-Ascend Program. Talon Johnson is awarded this fellowship to conduct a program of research and education in the mathematical sciences, including applications to other disciplines, at University of Texas Southwestern Medical Center under the mentorship of the sponsoring scientists Jimin Ren and Anke Henning. This is a research project aimed at development of computational image reconstruction methods for Magnetic Resonance Imaging (MRI). Johnson plans to design compressed sensing methods that are capable of reconstructing image data from MRI measurements that have been affected by the subject's motion, such as respiratory-related motion in MRI imaging of shoulder muscle metabolism. Along with this research, Johnson will be involved in activities to broaden participation of underrepresented minorities in mathematical sciences through the recruitment of students from the University of Texas at Arlington and a collaboration with the Atlanta University Center Data Science Initiative.This research is aimed at improving image precision and quality by completing three research objectives: 1) Develop and refine the methodology for reconstruction of compressed sensing data for blurred MRI imaging using l1-l2 minimization scheme, based on l1-magic for compressed sensing and l2 minimization for deconvolution of a fixed kernel function, 2) develop an AI method for auto-recognition of any unknown kernel functions from blurred compressed MRI data sets to achieve simultaneous compressed sensing - blind deblurring of any simulated MRI data, 3) refine the AI method for auto-compressed sensing and blind deblurring of experimental data in MRI imaging of shoulder muscle metabolism and improve imaging precision.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.