In patients with non-lesional epilepsy it is challenging to identify the zone of epilepsy onset in standard imaging approaches. Recent work suggests that it may be possible to determine the zone of epilepsy onset by identifying areas with reduced density of u-fibers [1]. U-fibers, or subcortical arcuate fibers, connect neighboring cortical region and provide inhibitory control. It is hypothesized that lack of inhibition due to lower numbers of u-fibers may contribute to the onset of epilepsy in affected tissues. Supporting this hypothesis, recent work in epileptic baboons suggests that there is a lack of neurons in motor areas that may be explained by lack of u-fibers in these areas [2]. U-fibers are interesting as a bio-marker in epilepsy because seizure propagation often occurs from the site of seizure onset to neighboring cortical regions [3]. Abnormalities in u-fibers density could be due to underlying vulnerability to seizures or due to adaptation to chronic seizures. Here we develop a method to quantify u-fibers from 7T MRI data and compare populations of epilepsy patients with lesions and without lesions to control subjects.Subjects: 5 control subjects, 2 epilepsy patients with lesions (1 cavernoma, 1 dysembryoplastic neuroepithelial tumor), 6 epilepsy patients without lesions. MRI: Subjects underwent 7T MRI consisting of a T1-weighted MP2RAGE sequence (0.7 mm isotropic resolution) and high-angular-resolved diffusion-weighted dMRI (1.05 mm isotropic resolution, 68 directions). Diffusion-weighted images were corrected for distortions and registered to T1-weighted images. Processing: Fiber orientation distributions for tractography were obtained from the corrected diffusion-weighted images by spherical deconvolution. Tractograpy was performed using the iFOD2 algorithm implemented in MRTRIX3 to obtain 100 million fibers seeded from the grey-white matter border. U-fibers were identified based on their length being longer than 3 cm and the distance between the ends being shorter than their length over pi (Figure 1). Quantitative U-fiber density maps were obtained by dividing track density maps of the u-fibers by track density maps of all the tracks. The end result is a quantitative u-fiber fraction map for each subject and u-fiber fraction for each region of the brain (freesurfer’s aparc+aseg). Statistics: A z-score was computed for the u-fiber fraction within each region across all subjects. The z-scores for all regions were compared for the 3 groups using t-tests and with cohen’s d.A qualitative comparison of u-fiber fraction maps from an axial slice in a typical subject from each group is provided in Figure 2. It is apparent that the non-lesional epilepsy patient (Fig 2a) has a lower fraction of u-fibers especially in the posterior regions when compared to the healthy control subject (Fig 2b). The lesional epilepsy patient had higher apparent u-fiber fractions than the healthy control subject (Fig 2c). A quantitative view of this trend is provided in Figure 3 which provides box-plots of the z-scored u-fiber fraction in each of the regions for each of the subjects. Each circle represents a u-fiber fraction in a single region from a single subject. In general, u-fiber fractions in the epilepsy patients with lesions were greatest followed by controls and non-lesional epilepsy patients.U-fiber fractions in non-lesional epilepsy patients were found to be lower than controls. Future work will be focused on comparing the regions with low u-fiber fractions to epileptogenic zones in those patients. The higher u-fiber fractions found in lesional epilepsy subjects may represent adaptations to abnormal function caused by the lesions. More data is needed to establish this as a phenotype. Future work will be aimed at validation and exploring u-fiber quantification in other neurological and psychiatric disorders.