Here’s the quick, apples‑to‑apples view on how deep each method “sees” in the brain (usable signal, not just theoretical reach):

• Conventional fNIRS (continuous‑wave)

  Sensitive mainly to superficial cortex; typical usable depth is about 1.5–2 cm from the scalp, with sensitivity peaking in the first ~1 cm of cortex. Deeper than ~2 cm the cerebral contribution drops off fast.

• DOT‑fNIRS / HD‑DOT (high‑density diffuse optical tomography)

  Same photon physics as fNIRS, but dense arrays + tomographic reconstructions improve localization and SNR. Depth is still limited to superficial cortex (~1–1.5 cm below the pial surface; ~≤2 cm from scalp)—it does not reach deep brain, though cortical maps can approach fMRI‑like layouts.

• OPM‑MEG (on‑scalp MEG with optically‑pumped magnetometers)

  Magnetic fields aren’t attenuated by skull like light is, so there’s no fixed “penetration depth.” Sensitivity, however, falls with source distance; OPMs help by putting the sensitive volume ~3–4 mm from the scalp (vs centimeters for cryogenic MEG), boosting superficial cortical signals. Deep sources are detectable in favorable cases (e.g., hippocampus), but reliability drops with depth and geometry; MEG overall prefers superficial, tangential cortical currents.

• Laser speckle imaging (LSCI / “speckle imaging”)

  Practically sub‑millimeter in tissue: photons sample only the superficial few hundred microns (≲1 mm). In brain work it’s usually done intraoperatively or with a cranial window; transcranially it measures scalp flow, not cortex.

Rule‑of‑thumb ranking (deepest → most superficial)

OPM‑MEG (whole brain sensitivity but depth degrades) → DOT‑fNIRS ≈ fNIRS (~superficial cortex only) → Speckle imaging (cortical surface/vascular layer only).

If you want, tell me your exact targets (e.g., hippocampus vs. inferior frontal gyrus) and constraints (helmet vs. bedside cap vs. OR), and I’ll map which modality and montage gets you there.