Multimodal Contrast-Free Pulmonary Perfusion Imaging by Integrating CT and MRI for Enhanced Lung Function Assessment

Area of Science:

• Medical Imaging
• Radiotherapy Physics
• Computational Biology

Background:

• Contrast-free lung function imaging is crucial for thoracic radiotherapy.
• Existing unimodal methods have limitations in capturing comprehensive pulmonary perfusion.
• Integrating computed tomography (CT) and magnetic resonance imaging (MRI) offers complementary information.

Purpose of the Study:

• Develop a multimodal contrast-free pulmonary perfusion reconstruction framework (MCF-Q).
• Integrate CT and non-contrast MRI for enhanced functional lung imaging.
• Improve agreement with single-photon emission computed tomography perfusion (SPECT-Q) for radiotherapy planning.

Main Methods:

• Prospective analysis of 21 lung cancer patients undergoing SPECT-Q, CT, and 1H MRI.
• Utilized a dual-branch deep learning architecture for MCF-Q.
• Evaluated voxel-wise and function-wise agreement with SPECT-Q using Spearman's R and Dice Similarity Coefficient (DSC).

Main Results:

• MCF-Q achieved a voxel-wise R of 0.7831 and function-wise DSC of 0.8396 (high-function) and 0.7680 (low-function).
• Significant improvement over single-modality methods was observed.
• MCF-Q-guided functional lung avoidance radiotherapy (FLART) improved dose sparing in high-functional regions.

Conclusions:

• MCF-Q demonstrates feasibility for multimodal, contrast-free perfusion reconstruction from CT and MRI.
• The framework provides planning-relevant functional information for FLART.
• Integration of routine CT and MRI for perfusion estimation shows promise, warranting further validation.