MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel
  1. Home
  2. Meeting Program
  3. All
  4. Cone Beam CT-Ventilation From Mass Conserving Point Cloud Density Functions

Cone Beam CT-Ventilation From Mass Conserving Point Cloud Density Functions

E Castillo1*, R Castillo2 , Y Vinogradskiy3 , D Solis4 , A Thompson5 , T Guerrero6 , (1) Beaumont Health, Royal Oak, MI, (2) Emory Univ, Atlanta, GA, (3) University of Colorado Denver, Aurora, CO, (4) Beaumont Health, Royal Oak, Michigan, (5) Beaumont Health, Royal Oak, Michigan, (6) Beaumont Health System, Royal Oak, MI

Presentations

(Sunday, 7/29/2018) 1:00 PM - 1:55 PM

Room: Karl Dean Ballroom B1

Purpose: To develop a numerical method for improving the robustness of transformation based CT-derived ventilation (CT-V) imaging. CT-V employs deformable image registration (DIR) to recover inhale/exhale lung motion, and estimates voxel volume changes by numerically approximating the Jacobian factor of the DIR transformation. However, errors associated with this approximation can cause uncertainties in CT-V. As such, current methods generate poor correlations between the CT-Vs derived from 4D cone beam CT (CBCT) and 4DCT. We propose computing a mass conserving point cloud density function from DIR mapped voxel positions and inhale/exhale lung segmentations. The Jacobian value approximations are then described, according to the Continuity equation, as one divided by the mapped density values.

Methods: The simulation (treatment planning) 4DCT and 4DCBCT images for five lung cancer patients were used to assess the proposed point cloud density Jacobian (DJ) method. Lung segmentation masks were processed with a semi-automated threshholding method. 4DCBCTs were acquired after simulation and before radiotherapy. The standard finite difference numerical Jacobian approximation (NJA) method and the DJ method were used to compute CT-V images on the 4DCTs and 4DCBCTs. CBCT CT-V images were aligned to 4DCT CT-V images via affine registration for direct comparison.

Results: The average (std) relative error in the NJA and DJ estimated global volume changes on CB were 0.1154 (0.0474) and 0.0019 (0.0010), respectively, while on 4DCT they were 0.0393 (0.0217) and 0.0057 (0.0051). The average Pearson correlation coefficient and average SSIM between CBCT and 4DCT NJA ventilation were 0.3397 (0.1799) and 0.6982 (0.0708) respectively, while for DJ they were 0.8411 (0.0459) and 0.9570 (0.0336).

Conclusion: Preliminary results indicate that the DJ method produces superior global volume change estimates than the standard NJA method, while the average SSIM and correlation values between CBCT and 4DCT DJ ventilation are significantly higher than those produced by NJA.

Keywords

Cone-beam CT, Image Processing, Ventilation/perfusion

Taxonomy

Not Applicable / None Entered.

Contact Email

Additional Links