Diffusion MRI data are typically acquired with a series of diffusion encoding directions, corresponding to one or more b values, for the targeted analyses (e.g., diffusion tensor imaging, DSI, HARDI and q-ball reconstruction among others). For example, 1) MRI protocols with 30 diffusion-encoding directions at \(b=1000 \frac{s}{mm^2}\) have been used in many diffusion-tensor imaging projects; and 2) protocols with a higher angular-resolution (i.e., more diffusion-encoding directions) at multiple b values are generally needed for high-order reconstruction (e.g., DSI).

For research projects that analyze data obtained with different protocols, one might need to down-sample higher angular-resolution diffusion MRI data (with more diffusion directions) to lower angular-resolution (with fewer directions), in a way that the down-sampled encoding schemes are similar across data sets.

In this page we show a matlab script that down-samples a high angular-resolution bvec scheme defined by ‘more_directions.txt’ and produces a 30-direction scheme resembling bvec defined in ‘fewer_directions.txt’. Basically, for each of the diffusion-encoding bvec directions defined in ‘fewer_directions.txt’, we use matlab dot() function (inner product) to identify the most similar one among all diffusion-encoding directions defined in ‘more_directions.txt’. Note that two diffusion encoding directions are considered the same, when the inner product is either 1 or -1 (assuming that diffusion encoding schemes [1,0,0] and [-1,0,0] have the same effect on diffusion signals).

The provided matlab code has a few limitations. First, it is implemented with an assumption that the two bvec files have the same b value. For a more complicated scheme (e.g., only 1 b-value is used in low angular-resolution scan and multiple b values are used in high angular-resolution scan), then some additional manual processing (e.g., extracting diffusion-encoding directions with the same b value) would be needed. Second, this matlab code does not examine whether the down-sampled diffusion directions are equally distributed in 3D space.