AI-Augmented Quantitative MRI Predicts Spontaneous Intracranial Hypotension

Area of Science:

• Neuroradiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Spontaneous intracranial hypotension (SIH) is linked to spinal cerebrospinal fluid (CSF) leakage, causing orthostatic headaches and CSF hypovolemia.
• Spinal CSF flow alterations in SIH are not well understood, unlike cerebral CSF dynamics.
• Quantitative assessment of spinal CSF flow is needed for SIH diagnosis and treatment prediction.

Purpose of the Study:

• To quantitatively assess spinal CSF flow at C2 using AI-enhanced phase-contrast MRI (PC-MRI).
• To evaluate the diagnostic utility of these CSF flow metrics for SIH.
• To predict patient response to epidural blood patch (EBP) treatment.

Main Methods:

• ECG-gated cine PC-MRI at C2 and whole-spine MR myelography were performed on 31 SIH patients and 26 healthy volunteers (HVs).
• AI (YOLOv4) and pulsatility algorithms were used for quantitative CSF flow metric extraction.
• Mann-Whitney U tests and ROC analysis were employed for statistical comparisons and performance evaluation.

Main Results:

• SIH patients exhibited significantly reduced spinal CSF flow parameters compared to HVs (p < 0.001).
• Downward peak flow (AUC=0.844) and summation of peak flow (AUC=0.841) showed excellent diagnostic accuracy.
• Baseline CSF flow metrics accurately distinguished patients needing one versus multiple EBPs (p < 0.001), with some parameters achieving AUCs up to 1.0 for predicting EBP success.

Conclusions:

• AI-enhanced PC-MRI provides robust quantitative evaluation of spinal CSF dynamics in SIH.
• These metrics effectively differentiate SIH patients from HVs.
• Quantitative CSF flow analysis accurately predicts EBP treatment response, aiding diagnosis and personalized treatment planning.