Intra-Slice Motion Correction of Intravascular OCT Images Using Deep Features

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Biomedical Engineering

Background:

• Intravascular imaging, particularly Optical Coherence Tomography (OCT), is vital for assessing coronary artery disease.
• Analyzing 3-D tissue structures from 2-D OCT images is hindered by motion artifacts.
• Accurate motion correction is essential for reliable volumetric analysis and pathological detection.

Purpose of the Study:

• To develop a robust motion correction method for 2-D intracoronary OCT image sequences.
• To improve the analysis of longitudinal tissue changes in 3-D.
• To enhance the characterization of internal tissue structures in coronary arteries.

Main Methods:

• A novel tissue-matching approach based on deep features from a Convolutional Neural Network (CNN).
• Correlation of deep features between sequential 2-D OCT frames to identify motion.
• Optimization of frame transformation by maximizing inter-frame tissue similarity.

Main Results:

• Demonstrated effective alignment of tissues within OCT image sequences.
• Successfully determined transformation parameters for motion correction.
• Validated the robustness of the CNN-based feature matching for intravascular OCT data.

Conclusions:

• The proposed CNN-based motion correction technique significantly improves the alignment of intracoronary OCT images.
• This method facilitates more accurate 3-D reconstruction and analysis of coronary artery tissues.
• The approach is crucial for advancing the diagnostic capabilities of intravascular OCT in cardiology.