Mohammad Niazy

تخصص الذكاء الاصطناعي · دفعة 2025

Validation of AI-Generated Cerebrospinal Fluid Flow from Resting-State fMRI Data: The Effect of Age

Cerebrospinal fluid (CSF) flow helps wash metabolic waste from the brain, which is directly related to brain health, and is known to slow with aging and in Alzheimer’s disease. Disruptions in CSF flow have been linked to impaired clearance of neurotoxins such as beta-amyloid, which build up in Alzheimer’s disease and may play a key role in early disease progression. Yet the current golden standard scan for measuring it, the phase-contrast MRI (PC-MRI), requires cost, time, and specialized hardware and personnel. A recent study solved this restriction by showing that the the inflow-based brightness changes seen in fast resting-state fMRI can be converted into quantitative, true CSF velocity curves using physics-informed neural network (PINN). Our project will put that method to the test on resting-state fMRI scans from the Human Connectome Project – Aging (HCP-A), covering mid-life through late adulthood, and compare its output to other flow measurements for validation. We will also examine the original model’s assumptions, such as fixed CSF T1, plug-flow shape, heart-rate near 1 Hz, constant proton density, and minimal respiratory influence, to learn which ones must be adjusted, and how, for data under different environments and conditions to get comprehensible results. After some pre-processing, which includes standard slice-timing, motion, and distortion correction, we will mask the bright CSF voxels, port the published Bloch formula + PINN code, and run targeted sensitivity checks by varying each assumption in turn. We expect to reproduce the key findings from the original study, including strong agreement with PC-MRI (correlation around 0.9 and low bias), and to replicate the reported decline in peak CSF speed of about 0.02 mm/s per decade. The outcome will be a validated, low-cost pipeline for large aging cohorts that shows how to adapt the model across sites; an important step toward using fast, routine fMRI for accessible CSF monitoring in Alzheimer’s research.

المشرف: Jean Chen

اقرأ التقرير النهائي (PDF)

اطّلع عليه في تقرير البرنامج (PDF)