Automatic non-rigid temporal alignment of intravascular ultrasound sequences: method and quantitative validation

Area of Science:

• Biomedical Engineering
• Medical Imaging Analysis

Background:

• Clinical studies on atherosclerosis progression require accurate intravascular ultrasound (IVUS) sequence alignment.
• Current alignment methods may lack precision for longitudinal IVUS data.

Purpose of the Study:

• To develop an automatic, non-rigid alignment methodology for IVUS sequences.
• To enhance the accuracy of comparing IVUS data before and after interventions and during follow-up.

Main Methods:

• Utilized dynamic time warping (DTW) adapted for multidimensional signals representing vessel morphology.
• Integrated DTW within a framework addressing partial data overlap and sequence regularization.
• Validated the method on synthetic and in vivo datasets.

Main Results:

• Achieved low alignment errors: approximately 0.43 mm for same-phase sequences.
• Demonstrated 0.77 mm alignment error for successive intervention stage sequences.
• The proposed method shows high accuracy in aligning complex IVUS data.

Conclusions:

• The proposed DTW-based methodology enables accurate automatic alignment of IVUS sequences.
• This technique can significantly benefit clinical research on atherosclerosis regression and progression.
• Improved alignment accuracy facilitates more reliable patient monitoring and treatment assessment.