Click here to

Session: New and Emerging Technology [Return to Session]

A Novel Super-Voxel Method for Generating Robust Lung Ventilation Image From CT

Z Chen*, Y Huang, G Ren, J Cai, The Hong Kong Polytechnic University, Hung Hom, Kowloon


SU-H330-IePD-F8-5 (Sunday, 7/10/2022) 3:30 PM - 4:00 PM [Eastern Time (GMT-4)]

Exhibit Hall | Forum 8

Purpose: To develop a novel CTVI-SVHU method for deriving lung ventilation images (VI) from CT images using a super-voxel approach, i.e., dividing the lung volume into hundreds of super-voxel which is a cluster of the voxels with similar characteristics, and using its mean CT value to generate robust VI without using deformable image registration (DIR).

Methods: A total number of 21 patients with 4DCT, the single-photon emission computed tomography (SPECT) and their corresponding lung masks from the Ventilation And Medical Pulmonary Image Registration Evaluation study were included in this study. For each patient, the lung volume of the exhale CT (ExCT) was segmented into 134 - 416 super-voxel volumes by using the Simple Linear Iterative Clustering (SLIC) method. The mean HU value of the super-voxel volume was used as the relative ventilation values and to calculate each voxel’s relative ventilation value based on the distance between the voxel and the center of all the super-voxels. The proposed CTVI-SVHU method was compared with two conventional DIR-derived VI
methods: HU value changes (CTVIDIR-HU) and volume changes (CTVIDIR-Jac) between the exhale/inhale CTs.

Results: The correlation between the mean HU value from ExCT and the mean ventilation value from SPECT of the super-voxel volume was 0.63 ± 0.08. Qualitative analysis showed that the CT value of the defect lung region is relatively lower than the normal lung region except the tumor-obstructed regions. On average, the CTVI-SVHU method showed better accuracy (0.63) than CTVIDIR-HU (0.32) and CTVIDIR-Jac (0.22) in terms of correlation values with SPECT.

Conclusion: The correlation between the mean HU value and the mean ventilation value shows that the mean HU value can be used to generate the VI. The correlations between the CTVI-SVHU image and SPECT indicate that super-voxel-based VI is consistent with SPECT VI.

Funding Support, Disclosures, and Conflict of Interest: This work was supported in part by General Research Fund (GRF15103520) from The University Grants Committee, and Health and Medical Research Fund (HMRF07183266) from The Food and Health Bureau, The Government of the Hong Kong Special Administrative Regions.


Ventilation/perfusion, Lung


Not Applicable / None Entered.

Contact Email