Akademik Veri Yönetim Sistemi
JOURNAL OF PHYSICS D-APPLIED PHYSICS, cilt.58, sa.49, 2025 (SCI-Expanded, Scopus)
Terahertz (THz) frequency radiation offers significant potential for safe, non-ionizing, non-destructive, and multimodal biomedical imaging. However, the strong absorption of THz radiation by water-rich biological tissues severely limits the acquisition of high-quality images, especially for in vivo applications. This study investigates the efficacy of independent component analysis (ICA), a robust blind source separation technique and a learning algorithm, in enhancing THz biomedical imaging. Following the initial case study, ICA was applied to THz spectroscopic images of phantom skin models containing tumor-like inclusions in order to suppress background interference and isolate diagnostically relevant signals. Experimental data were collected using conventional THz time-domain spectroscopy. Various spectroscopic image reconstruction methods were compared to optimize image contrast and highlight regions of interest. Our results show that ICA notably improves image contrast and enhances tumor delineation, demonstrating strong frequency-dependent sensitivity. Consequently, ICA-based decomposition presents a highly promising strategy to achieve clearer and more accurate THz images, thereby substantially enhancing the capability of THz imaging in biomedical diagnostics.