Presentation description

Lung functional imaging is used to guide radiation therapy by localizing functional tissue to spare functional portions of the lung during treatment. By directing radiation to the tumor through lung regions with poor gas exchange, you reduce risk to healthy lung tissue and minimize thoracic toxicity. CT ventilation mapping offers a cost-effective alternative to PET or SPECT by utilizing routinely acquired 4DCT data reducing additional radiation exposure and imaging, as well as avoiding additional costs to patients. We aimed to develop and validate accurate and reproducible CT-based ventilation maps using deformable image registration and voxel-wise computations through MIM software and MATLAB. I researched two common computational methods: HU-based image subtraction, the Jacobian method, and an experimental density-change-based computation examining local change in air content at the voxel level. Both the Hounsfield Unit (HU)-based subtraction and the Jacobian determinant used peak inhale and exhale 4DCT scans to create deformable image registrations (DIR), ensuring correct spatial alignment between voxels before applying methodology procedures. The experimental density-change computation took a different approach, moving away from MIM and into MATLAB, a high-level programming and numerical computing platform. Here, a script was coded to 'correct' the inhale CT scan to account for respiratory-induced changes in tissue perfusion. This 'corrected' CT scan would be used in MIM replacing the original inhale CT scan with the hopes of yielding better results, or a clearer ventilation map. Preliminary results showed the HU method to be sensitive to noise, artifacts, and tissue density changes but it produced a highly detailed ventilation map. In contrast, the Jacobian determinant was sensitive to DIR accuracy and often showed excessive expansion near diaphragm or large airways but yielded consistent results. The local density-change voxel methodology remained inconclusive as of 7/7 since it is still being worked through.