pub fn nextqsm_tiled(
total_field: &[f64],
mask: &[u8],
grid: &Grid,
bdir: (f64, f64, f64),
bf_onnx: &[u8],
vjp_onnx: &[u8],
cfg: &TileConfig,
progress: impl FnMut(usize, usize),
) -> Result<Vec<f64>, OnnxError>Expand description
Memory-bounded NeXtQSM via whole-algorithm overlap-tiling — the full VarNet gradient-descent
loop runs on each patch sub-volume (each internally padded to /64). NeXtQSM’s forward model is
global, so tiling is strongly off-design and approximate; prefer a whole-volume run (e.g.
QSMxT). See crate::inversion::tiled::tiled_volume_algorithm.