Abstract
Segregating the magnocellular (M) and parvocellular (M) layers of the human lateral geniculate nucleus (LGN) is difficult yet remains an important goal because the LGN is the location where the two streams are anatomically disjoint. Previous studies measured M and P response amplitudes to different tuned stimuli, but this method is confounded by the hilum region of the LGN that exhibits greater response amplitudes and can be mistaken for the M layers (DeSimone & Schneider, 2019). Here, we employed two independent methods that do not rely on the stimulus-tuned responses: a quantitative magnetic resonance imaging (qMRI) measuring T1 relaxation time (qT1) and two functional MRI (fMRI) procedures measuring ocular signals. Method. Three participants were scanned with a 3T MRI scanner over multiple days. Seventeen qMRI scans were acquired (MP2RAGE sequence, 0.7mm isotropic, TR = 5 s, TE = 3.6 ms, TI1/TI2 = 0.9/2.75 s, α1/α2 = 3/5°) and used to calculate qT1 maps. From the averaged qT1 maps, we outlined the six LGN. For the fMRI scans (multiband sequence, slice acceleration = 6, TR = 1.5 s, TE = 39ms, 1.5mm isotropic), two tasks were used to measure ocular signals: a monocular task where subjects closed each eye alternately for 15 s while viewing a checkboard flickering at 4Hz, and a dichoptic task where the stimulus was presented to each eye alternately while the opposite eye viewed a blank gray screen. Results. By fitting the qT1 maps to a 2-component Gaussian model, our qMRI results agreed with anatomy: identifying M regions on the ventromedial surface of each LGN. The two fMRI procedures weakly agreed with each other, revealed significant right eye bias, and could not reliably identify the contralateral M layer of the LGN. Conclusion. The qMRI method is promising whereas the functional identification of contralateral layers requires further refinement.