Cardiac C-arm CT: a unified framework for motion estimation and dynamic CT

Area of Science:

• Medical Imaging
• Cardiovascular Technology
• Image Reconstruction

Background:

• Generating high-quality 3-D/4-D cardiac CT images during interventional procedures is challenging.
• Current methods like ECG-gated Feldkamp (FDK) reconstruction can be limited by residual cardiac motion.
• Improving temporal and spatial resolution, as well as signal-to-noise ratio (SNR), is crucial for better diagnostics.

Purpose of the Study:

• To develop a unified framework for heart motion estimation and dynamic cone-beam reconstruction with motion correction.
• To introduce novel signal-enhanced reconstruction methods utilizing all acquired projection data.
• To evaluate the effectiveness of motion correction and new reconstruction techniques in improving cardiac CT image quality.

Main Methods:

• A unified framework for motion estimation and dynamic cone-beam reconstruction was developed.
• Three signal-enhanced reconstruction methods were introduced: simple averaging, intensity-weighted averaging, and temporal-distance-weighted averaging of motion-corrected back-projections.
• Seven reconstruction methods were compared, including non-gated FDK, ECG-gated FDK, ECG-gated and motion-corrected FDK, the three signal-enhanced methods, and temporally aligned/averaged ECG-gated FDK.

Main Results:

• Motion correction substantially reduced motion-related blurring, enhancing both temporal and spatial resolution.
• Signal-to-noise ratio (SNR) was increased by up to 70% compared to standard ECG-gated FDK.
• The proposed framework maintained spatial resolution comparable to ECG-gated FDK while significantly boosting SNR.

Conclusions:

• Data-driven, subject-specific motion estimation and correction are effective in improving cardiac C-arm CT image quality.
• The developed signal-enhanced reconstruction methods provide superior SNR and resolution.
• The framework offers excellent image quality for cardiac C-arm CT, aiding interventional procedures.