Let \(y_{i,j}\) be the presence / absence of ICH for voxel \(j\) from person \(i\). \[ \text{logit}\left(y_{i, j}\right) = \beta_0 + \sum_{k= 1}^{p} x_{i, j, k}\beta_{k} \]

• Case-control sample voxels for a fixed percentage (25%) of outcome