Pattern-Matching Unit for Medical Applications

We explore the application of concepts developed in High Energy Physics (HEP) for advanced medical data analysis. Our study case is a problem with high social impact: clinically-feasible Magnetic Resonance Fingerprinting (MRF). MRF new, quantitative, imaging technique that replaces multiple qualitative Imaging (MRI) exams single, reproducible measurement increased sensitivity and efficiency. A fast acquisition followed by pattern matching (PM) task, where signal responses are matched to entries from dictionary simulated, physically-feasible responses, yielding tissue parameters simultaneously. Each pixel response volume compared through scalar products all choose best reproduction. limited PM processing time, which scales exponentially dimensionality, i.e. number be reconstructed. HEP powerful, compact, embedded system, optimized extremely PM. This system executes real-time tracking online event selection experiments, exploiting maximum parallelism pipelining. Track reconstruction executed two steps. The Associative Memory (AM) ASIC first implements algorithm recognizing track candidates at low resolution. second step, implemented into FPGAs (Field Programmable Gate Arrays), refines AM output finding full propose use this perform MRF, achieve clinically reasonable time. paper proposes an adaptation shows some preliminary results.