Breaking the limitations with sparse inputs by variational frameworks (BLIss) in terahertz super-resolution 3D...

Area of Science:

• Medical Imaging
• Computational Imaging
• Terahertz Technology

Background:

• Terahertz (THz) 3D reconstructed imaging faces persistent challenges in data acquisition, image processing, and overall image quality.
• Current methods are often limited to 2D scenarios, struggling with super-resolution (SR) data acquisition and lacking efficient SR 3D reconstruction frameworks within conventional computed tomography (CT).

Purpose of the Study:

• To introduce BLIss, an innovative approach for achieving super-resolution (SR) 3D reconstruction in terahertz (THz) imaging using limited 2D data.
• To address the limitations of existing 2D-centric methods and enhance the quality of THz 3D reconstructed images.

Main Methods:

• BLIss integrates conventional computed tomography (CT) techniques with a variational framework, utilizing an adapted Euler-Elastica-based model.
• The approach processes sparse 2D data to achieve high-resolution 3D reconstructions.

Main Results:

• Quantitative evaluation using metrics such as standard deviation of Gaussian, mean curvatures, and multi-scale structural similarity index measure (MS-SSIM) demonstrated superior smoothness and fidelity.
• BLIss significantly outperformed conventional THz CT modalities in image quality.

Conclusions:

• BLIss represents a significant advancement in THz SR 3D reconstruction, offering improved image smoothness and fidelity from sparse 2D data.
• The framework shows potential for broad applicability in other imaging modalities, including X-ray and MRI, indicating extensive impacts on the imaging field.