Lung allocation pipeline: Machine learning approach to optimized lung transplant

Lung donation is the most risky transplant procedures. With low survival rates, and poor acceptance of donated lungs, those in need of a lung transplant are at high risk of dying. One reason for poor outcomes is the lack of optimal match between donor and recipient when it comes to lung size and shape. Lungs that do not properly fit in the recipient's chest cavity can fail to inflate fully and quickly start to deteriorate. In such patients, lung contusions can form, edema occurs in healthy lung tissue, and overall lung function declines. To improve patient outcomes after lung transplant, we describe here a developed a computational pipeline which enables donor lungs to be properly matched to recipients. This tool uses CT scans from both the donor and potential recipients to calculate how anatomically different the sets of lungs are, and therefore provide improved matches in both size and shape for the donor lungs.