To assess the feasibility of line scanning BOLD fMRI on human subjects.Recently, Yu et al. have proposed line scanning fMRI as an experimental method to probe the blood oxygenation level dependent (BOLD) effect in the rodent brain with 50 ms temporal resolution [1]. Line canning fMRI trades spatial information for acquisition speed. It acquires BOLD signal from one “line” through the brain, which is defined by slice selection and additional saturation slices in plane (Fig. 1), without performing any phase encoding. Temporal resolution is therefore defined by the time required for acquisition of one line in k-space (i.e. TR), which is typically around 50 ms. Here, we have implemented line scanning fMRI on a clinical MR scanner, and compare the results to conventional EPI fMRI data and a fMRI protocol using a FLASH sequence from which the line scanning sequence was derived. All measurements were performed on a clinical 3 tesla Scanner (Siemens Prisma) using a 20 channel head coil and a TE = 30 ms. EPI and FLASH sequences were used as provided by the vendor. While the EPI sequence (N_slices = 30, TR = TR_total = 2000 ms) covered the whole brain, only one slice was acquired with the FLASH sequence (TR_total = 3.5 s/slice, TR = 40 ms, FA = 22°, matrix size 64 x 64). For line scanning, a modified FLASH sequence without phase encoding was used (TR = 100 ms, FA = 22°, slice thickness 10 mm, 128 frequency encoding steps). Frequency encoding was performed along the scanned “line”. Tissue outside this area of interest was saturated using the vendor provided saturation module (Fig. 1). For fMRI a visual stimulation paradigm consisting of 10 s stimulation and 15 s rest (black screen) was applied. In each stimulation block, 5 faces from the Ekman and Friesen collection of emotional faces [2] were presented. Conventional fMRI data acquired with EPI or FLASH were evaluated using Statistical Parametric Mapping (SPM8). To obtain the BOLD signal, the voxel cluster with maximum intensity was selected manually. For this activated region the BOLD signal time course was extracted without assuming any model. Line scanning data were Fourier transformed to yield projections of the line through the brain. Subsequently, signal intensities were summed over a region of interest, comprising the occipital cortex. Analysis was applied using matlab.The stimulation paradigm resulted in robust BOLD activation (2% signal change) of the occipital cortex and the amygdala in the data sets acquired with EPI (Fig. 2). Data acquired with FLASH showed robust BOLD in the occipital cortex only (5% signal change). From both data BOLD time courses could be extracted (Fig. 3 a, b). Line scanning data were summed over the occipital cortex, allowing to obtain a BOLD time course with 100 ms resolution (Fig. 3c). BOLD amplitude was 4-5%. From both EPI and line scanning data, BOLD responses were averaged over twelve stimulations to visualize the hemodynamic response. The response showed a delayed rise to a maximum that was reached only 13 seconds after stimulation onset (i. e. 3 s after end of stimulus), before showing a post stimulus undershoot and returning to base line (Fig. 4).Line scanning fMRI was successfully implemented on a clinical MR scanner and provided reliable BOLD signal of 4-5% amplitude in response to a visual stimulation paradigm. The observed BOLD time course perfectly matched the time course observed with a conventional EPI sequence. Currently the temporal resolution is limited to 100 ms, due to long saturation times required to achieve satisfying suppression of signal outside the “line”. In deeper brain regions (amygdala) line scanning fMRI failed to detect BOLD signal, presumably due to B₁ inhomogeneities, resulting in lower sensitivity or deviating saturation profiles. However, with optimized saturation methods line scanning fMRI may reach temporal resolutions below 50 ms and become a technique of high potential to boost temporal resolution in human fMRI.