A Methodological Review of 3D Reconstruction Techniques in Tomographic Imaging

Area of Science:

• Medical Imaging
• Computer Vision
• 3D Modeling

Background:

• Computer Vision significantly aids medical diagnostics, enabling root cause analysis via 3D medical imaging.
• 3D modeling in medicine utilizes surface rendering, volume rendering, and regularization methods.
• Tomographic reconstruction differs from camera-based scene reconstruction, facing unique challenges.

Purpose of the Study:

• To comprehensively study existing 3D modeling algorithms in medical imaging.
• To summarize state-of-the-art techniques developed over four decades.
• To establish a foundation for precise 3D human spine reconstruction.

Main Methods:

• Review of diverse 3D reconstruction techniques (minimal surfaces, level sets, snakes, graph cuts, etc.).
• Analysis of limitations in tomographic image acquisition (CT, X-ray, MRI) and calibration.
• Exploration of surface rendering, volume rendering, and regularization-based approaches.

Main Results:

• Identified limitations in image acquisition and calibration impacting tomographic reconstruction quality.
• Summarized a wide array of 3D modeling algorithms used in medical imaging.
• Established a foundational understanding of current 3D reconstruction methodologies.

Conclusions:

• Current tomographic reconstruction methods face inherent limitations affecting final image quality.
• A comprehensive review provides a basis for future advancements in medical 3D modeling.
• Future work will focus on precise 3D reconstruction of the human spine.