Purpose: Tracking and quantifying local tissue changes in the lungs is imperative in furthering the study of radiation-induced lung damage (RILD). Traditional intensity-based registration approaches fail in this task due to dramatic geometric changes between timepoints. This work aims to successfully register longitudinal pre- and post-radiotherapy (RT) lung CT scans with considerable radiological changes due to RILD. This is achieved by extracting and registering only consistent anatomical features (lung boundaries, main airways, vessels).
Methods: Fifteen pre-RT and 12-month post-RT CT pairs from a chemoradiation trial for lung cancer were available, all with varying degrees of RILD, ranging from mild parenchymal change to extensive consolidation and collapse. For each CT, signed distance transforms from delineations of the lungs and main airways were generated and the Frangi vesselness was calculated. These feature maps were concatenated into multi-channel images and diffeomorphic multichannel registration was performed for each image pair. For the evaluation, the pre- and post- registration landmark distance was calculated for all patients at a mean (range) of 44 (38-65) manually identified landmark pairs. Traditional intensity-based registrations were also performed for all patient datasets and evaluated in the same way as a comparison.
Results: The mean (standard deviation) distance for all datasets decreased from 15.95 (8.09) mm pre-registration to 4.56 (5.70) mm post-registration. Qualitative improvements in image alignment were observed for all patient datasets. In comparison, the post-registration error of the traditional intensity-based registrations was higher, at 7.90 (8.97) mm and registrations of patients with presence of atelectasis and extensive consolidation failed.
Conclusion: We have demonstrated that our novel registration method can successfully align 12-month follow up scans from RILD patients in the presence of large anatomical changes such as consolidation and atelectasis, outperforming the classical registration approach both quantitatively and through detailed visual inspection.
Funding Support, Disclosures, and Conflict of Interest: AS- EPSRC-funded UCL i4health CDT (EP/S021930/1). CV- Royal Academy of Engineering, Research Fellowship scheme (RF\201718\17140). JRM- Cancer Research UK Centres Network Accelerator Award Grant (A21993), ART-NET consortium. IDEAL CRT trial- Cancer Research UK, grant no. C13530/A10424 and C13530/A17007. AS, CV, EC, AS, JRM report support from a charitable donation.