CW-fDOT with a standard optical brain model: improving hemodynamic sensitivity and functional localization

Continuous-wave functional diffuse optical tomography (CW-fDOT) enables three-dimensional (3D) functional brain imaging using portable functional near-infrared spectroscopy systems. However, its reconstruction accuracy is often limited by simplified background models and empirical initialization of optical properties. Although subject-specific anatomical structures and optical properties can provide accurate prior information, the additional data-acquisition burden and workflow complexity limit their practical translation. Therefore, a reusable standardized model is urgently needed to improve CW-fDOT reconstruction performance while preserving its portability and practical applicability.

This study aimed to develop and validate a CW-fDOT reconstruction framework based on a standard optical brain (SOB) model to improve cerebral hemodynamic reconstruction accuracy, activation detection, and functional brain-state discrimination without requiring additional acquisition of subject-specific anatomical structures or optical properties.

Frontal anatomical statistics derived from magnetic resonance imaging and tissue-specific optical properties estimated by time-domain DOT were integrated to construct a frontal SOB model. The model can be scaled according to individual external head dimensions and incorporated into a source-detector distance-dependent CW-fDOT reconstruction strategy, thereby providing anatomical constraints and tissue-specific optical-property initialization for reconstruction. The framework was systematically validated through SOB model performance evaluation, numerical simulations, phantom experiments, and in vivo breath-holding experiments involving 48 subjects.

The SOB model preserved the major frontal tissue structures, physiologically reasonable optical properties, and the main photon-transport characteristics of subject-specific models. Numerical simulations showed that SOB initialization achieved the best reconstruction performance among the four initialization strategies. Phantom experiments further confirmed that, under realistic measurement conditions, the proposed method improved target localization, morphological recovery, and quantitative reconstruction. In vivo breath-holding experiments showed that, compared with homogeneous-background empirical initialization, SOB initialization identified more activated regions, with the number of significantly activated gray matter nodes increased by 79.52%. The extracted features also more effectively discriminated different functional brain states. Dynamic 3D imaging further showed that the main GM response region exhibited a 15.42% relative increase in mean oxyhemoglobin concentration change during the breath-holding period compared with the first 30-s resting period of one trial.

Incorporating a reusable SOB model into CW-fDOT reconstruction provides effective anatomical constraints and optimized optical initialization, thereby improving quantitative reconstruction accuracy, activation detection, and functional brain-state discrimination. This framework offers a feasible pathway toward high-precision portable functional brain imaging, although its generalizability and clinical applicability require further validation in larger and more diverse populations.