Adapted binning method for assessing pulmonary ventilation and perfusion distribution in children using lung matrix pencil decomposition (MP)-MRI

<bold>Introduction</bold>: The binning method introduced by He et al. <italic>Acad Radiol. 2016,1521–1531</italic> for hyperpolarized ¹²⁹Xe MRI classifies the pulmonary distribution into four categories: defect, low, normal and high, extracting more information than the commonly used defect percentage. <bold>Method</bold>: We developed a binning method adapted for lung MP-MRI ventilation and perfusion images on a reference dataset of 157 healthy children. We normalized the images to exclude abnormally high intensity voxels due to blood vessels for perfusion and accumulated tissue for ventilation. We then established the reference thresholds by categorizing the pulmonary distribution into 6 bins of 1 standard deviation (SD) width around the peak of the distribution: the normal bins (±1 SD from the peak), the high bins (>1 SD), the low bin (between -1 SD and -2 SD) and the defect bin (<- 2 SD). To evaluate the accuracy of the classification and the discriminatory power of the method, we normalized and binned a validation dataset of healthy controls (HC) and cystic fibrosis (CF) children. <bold>Results:</bold> The adapted binning method identified defected and low regions in expected areas upon visual examination for both HC and CF children. It also significantly distinguished CF children from HC in perfusion and ventilation images for both the defect percentage and the low percentage (all p’s<0.0001) with an excellent discrimination (all AUCs > 0.85). <bold>Discussion</bold>: The adapted binning method for MP-MRI might be a well-suited tool for monitoring disease progression and subtle changes in milder diseases.