MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel
  1. Home
  2. Meeting Program
  3. All Monday - 7/30/2018
  4. Enhancement of Super-Resolution Time-Resolved 4DMRI Reconstruction in Low-Contrast Tissue by Two-Step Deformable Image Registration

Enhancement of Super-Resolution Time-Resolved 4DMRI Reconstruction in Low-Contrast Tissue by Two-Step Deformable Image Registration

X Nie1*, K Huang2 , S Sharma3 , J Deasy4 , A Rimner5 , G Li6 , (1) Memorial Slona Kettering Cancer Center, New York, NY, (2) Memorial Slona Kettering Cancer Center, New York, NY, (3) Memorial Slona Kettering Cancer Center, New York, NY, (4) Memorial Sloan Kettering Cancer Center, New York, NY, (5) Memorial Sloan Kettering Cancer center, New York, NY, (6) Memorial Sloan Kettering Cancer Center, New York, NY

Presentations

(Monday, 7/30/2018) 9:30 AM - 10:00 AM

Room: Exhibit Hall | Forum 5

Purpose: Super-resolution time-resolved (TR)-4DMRI with adequate spatiotemporal resolution provides the clinical opportunity of multi-breath motion simulation. This study aims to enhance and quantify low-contrast soft-tissue alignment in deformable image registration (DIR) for TR-4DMRI reconstruction.

Methods: We introduced a two-step DIR method to enhance low-contrast tissue alignment. Via DIR, TR-4DMRI is reconstructed by combining a low-resolution (5x5x5mm³) 3D cine in free breathing (FB) at 2Hz and a high-resolution (2x2x2mm³) breath hold (BH). After conventional DIR to deform the whole image, we added a piecewise DIR (3x3x2=18 sub-volumes), followed by automatic lung segmentation for DIR refinement of both the lungs and remaining body shell. For each sub-volume, the voxel intensity was renormalized. Four volunteers and one patient were scanned under an IRB protocol. Voxel intensity correlation (VIC) and structure similarity (SSIM) indices were used to evaluate the similarity between two images. Based on anatomic landmarks, such as a tumor or inferior lung vessel bifurcation points (6-13 points/subjects), image alignment was quantitatively evaluated. We simulated a 3D cine image by down-sampling a BH image at full inhalation (BHI) and deformed a BH image at full exhalation (BHE) to it, and then compared the deformed BHE with the original BHI, and vice versa.

Results: DIR accuracy is substantially enhanced by the 2-step DIR method. The VIC increases from 0.96 to 0.98 and SSIM increases from 0.87 to 0.91. The accuracy at the lung-diaphragm interface is 0.2±1.9mm. The accuracy within the lungs is 3.0±1.0 mm (BHI<->BHE), improved from 6.5±1.3mm (BHI->BHE) and 12.4±6.0 (BHI<-BHE) by conventional DIR, given the original misalignment of 27±5mm. A lung tumor alignment accuracy is <2mm. The computation time is ~15 minutes/volume, ~50% longer than the conventional DIR.

Conclusion: TR-4DMRI reconstruction is substantially improved by the 2-step DIR approach, promising to provide a clinically-acceptable accuracy for lung tumor delineation and motion assessment.

Keywords

Respiration, MRI, Radiation Therapy

Taxonomy

IM- MRI : 4D MRI

Contact Email

Additional Links