In a study of 50 schizophrenia (SCZ) patients, 67 bipolar disorder (BD) patients, and 52 healthy controls, fNIRS during a verbal fluency task (VFT) showed reduced prefrontal oxyhemoglobin responses in both SCZ and BD vs controls—and lower activation in SCZ vs BD across multiple regions, including the left orbitofrontal cortex (lOFC). The authors conclude lOFC may be the most critical biomarker for differentiating SCZ from BD, with potential gains when paired with deep learning and interpretability methods.

Why It Matters To Your Practice

• SCZ and BD can overlap clinically; an objective, task-based signal (prefrontal hemodynamic response during VFT) could support differential diagnosis when history is limited or symptoms are mixed.

• fNIRS is comparatively accessible vs MRI and can be performed during cognitive tasks, making it a plausible “clinic-adjacent” neurophysiology tool if validated.

Clinical Implications

• Consider fNIRS-VFT as a potential adjunct (not a stand-alone test) in complex SCZ vs BD differentiation—especially where cognitive activation patterns may add information beyond symptom checklists.

• If adopting AI-assisted fNIRS tools, prioritize workflows that report interpretable features (e.g., which channels/regions drove the classification) rather than only a binary label.

Insights

• Both SCZ and BD showed reduced prefrontal activation vs healthy controls, reinforcing that “reduced PFC signal” is not specific—regional patterns (not just magnitude) matter.

• Channels mapping to rFPC, lFPC, lDLPFC, and particularly lOFC were highlighted as differentiators, suggesting region-focused models may outperform diffuse whole-PFC approaches.

The Bottom Line

• Task-based fNIRS plus interpretable deep learning is being positioned as an auxiliary diagnostic aid for SCZ vs BD, with lOFC emerging as a leading candidate feature—but clinical utility hinges on external validation, calibration, and clear reporting.