Midnight Scan Club dataset fsLR_32k format

Hi, does someone have experience with processing fsLR_32k representation of the MSC dataset?

The dimension of their resting-state data is 65,890 * 818. I’m confused, why the first dimension is 65,890? For human connectome project data, the first dimension of the whole cortex would be either 64,984 (if includes medial wall) or 59,412 (if excludes medial wall). If include both cortex and subcortical area, the number of vertices would be up to ~90k. However, their number (65,890) doesn’t match with any one of those.