X Wu¹*, H Yao¹, G Hugo¹, C Eldeniz¹, H An¹, H Gach¹, D Yang², (1) Washington University in Saint Louis, St. Louis, MO, (2) Duke University, Chapel Hill, NC

• 
  X Wu

• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  

SU-E-207-6 (Sunday, 7/10/2022) 1:00 PM - 2:00 PM [Eastern Time (GMT-4)]

Room 207

Purpose: Cardiac signal is essential in the analysis of cardiac imaging and cardiac motion models. We developed an approach, purely based on the image intensity, to directly extract the cardiac signal from different image modalities such as 2D cine MRIs and CBCT projections. It can be used as an alternative when cardiac signal such as ECG is not available.

Methods: For images with a consistent FOV such as the 2D cine MRIs, the main contributors of the image intensity changes are the cardiac and respiratory motions. The cardiac signal was generated by 1) calculating the average intensity of the whole FOV, and 2) filtering the intensity signal based on the common frequency of the cardiac motions. For CBCT projection images acquired with rotating gantry, the cardiac signal is extracted by 1) manually selecting cardiac boxes in several frames with obviously different heart position such as the 90 180 360 in a 360 rotation; 2) smooth-fitting then interpolating the sizes of the box for the frames in between; 3) calculating the average image intensity in the moving box; 4) using smoothing-spline-fitting to remove the baseline in the signal causing by the rotation; and 5) generating the cardiac signal based on the band-pass filter.

Results: The proposed cardiac signal extraction approach was tested on 2D cine MRIs and CBCT projections from 15 different subjects. The extracted cardiac signals were consistent and synchronized with the images. A comparison with the landmark-based cardiac signals from ICD lead locations in three subjects showed that the two cardiac signals were consistent.

Conclusion: A novel approach was developed to extract the cardiac signal from the image itself. The results were consistent with landmark-based cardiac signals.

Keywords

Heart, MRI, CT

Taxonomy

IM/TH- Image Analysis (Single Modality or Multi-Modality): Image processing

Contact Email