Partial volume transfer (PVT) conversion of cerebral tissue volumes between different magnetic fields MRI

Abstract

In neurodegenerative clinical studies, patients are usually followed up for several years frequently examined in different MRI scanners and base magnetic fields. Technical specification differences in MRI scanners, e.g., acquisition protocols and spatial resolution, are two crucial limitations for longitudinal long-term cerebral atrophy investigations. Although widely known that MRI base magnetic field (B0) boldly affects brain tissue volume measurements, no systematic study has been proposed to maintain brain volume consistency through longitudinal exams. We propose a method to convert the measured volumes of segmented brain compartments from one MRI scanner with lower B0 and resolutions to volumes consistent with a higher resolution. The proposed partial volume transfer (PVT) method is consistent with partial volume effect to correct the volume measures. The PVT demands at least one pair of simultaneous acquisitions on both MRI scanners to estimate three unknown coefficients completing the PVT system. We selected ten healthy subjects from a standard dataset on the web, ten other healthy subjects, and ten patients with multiple sclerosis diseases scanned by two different scanners, i.e., 1.5 T and 3.0 T from local hospital dataset, to evaluate the PVT. We obtained the mean relative errors ranging from 0.28 to 1.14% of total brain volume using the PVT conversion for brain compartments, significantly less than interpolation methods which were more than 6.54%. By providing a real longitudinal sequence of images during 9-years follow-up of a patient, the PVT conversion was used to accurately convert cerebral tissues volumes acquired in different spatial resolutions.