Accurate Prediction of Lung Transplant Rejection by Combining Functional Respiratory Imaging (FRI) with Machine Learning
Chronic rejection or Bronchiolitis Obliterans Syndrome (BOS) remains the most important cause of death in lung transplant patients. Diagnosing BOS today is challenging due to the lack of sensitive lung function measurements. Spirometry only offers an overall assessment of the lung function without revealing any regional information. FLUIDDA’s FRI technology has proven to yield accurate regional information related to lung structure and function.
In a new study FLUIDDA’s researcher in collaboration with the University of Pennsylvania demonstrated the capability of FRI in combination with Machine Learning to accurately predict the onset of BOS. The investigators used CT images taken shortly after transplant in a cohort with, at that time, no signs of BOS based on spirometry. About half of the patients developed BOS in the years after transplant while the other half remained BOS free. FRI in combination with Machine Learning was able to predict with an accuracy of 85% whether a patient would go on to develop BOS or not using the images taken shortly after the surgery. These results are of significant importance for physicians treating lung transplant patients and for pharmaceutical companies developing medication for BOS.
The study was accepted for presentation at the ERS, next September in Milan, as a late breaking abstract.
