Infant head subsurface imaging using high-density diffuse optical tomography and machine learning

Infant head injuries and damage can be caused by various factors such as tumors or physical trauma. The treatment for a head injury will depend on the severity of the damage. Nevertheless, the infant’s head should be imaged before any treatment. High-density diffuse optical tomography (HD-DOT) is a non-invasive imaging technology that can be employed for subsurface imaging of the infant brain. However, there are problems with HD-DOT, such as low resolution, ill-posedness of the inverse problem, and high computational costs. In this study, to improve subsurface imaging of the infant head, an extreme gradient boosting (XGBoost) algorithm is combined with HD-DOT. The proposed method is then used to detect subsurface anomalies in the infant head. The proposed method achieves a similarity index greater than 0.97 in terms of cosine similarity and less than 0.12 in terms of the root mean square error, demonstrating its effectiveness. Moreover, the proposed method requires a minimal dataset compared to conventional deep learning methods and consumes significantly less time to train. The results of this study suggest that the proposed method can provide a promising alternative for subsurface imaging of the infant head, which could significantly impact the medical imaging field in the future.