RESUMEN
Lung size measurements play an important role in transplantation, as optimal donor-recipient size matching is necessary to ensure the best possible outcome. Although several strategies for size matching are currently used, all have limitations, and none has proven superior. In this pilot study, we leveraged deep learning and computer vision to develop an automated system for generating standardized lung size measurements using portable chest radiographs to improve accuracy, reduce variability, and streamline donor/recipient matching. We developed a 2-step framework involving lung mask extraction from chest radiographs followed by feature point detection to generate 6 distinct lung height and width measurements, which we validated against measurements reported by 2 radiologists (M.K.I. and R.R.) for 50 lung transplant recipients. Our system demonstrated <2.5% error (<7.0 mm) with robust interrater and intrarater agreement compared with an expert radiologist review. This is especially promising given that the radiographs used in this study were purposely chosen to include images with technical challenges such as consolidations, effusions, and patient rotation. Although validation in a larger cohort is necessary, this study highlights artificial intelligence's potential to both provide reproducible lung size assessment in real patients and enable studies on the effect of lung size matching on transplant outcomes in large data sets.
RESUMEN
BACKGROUND: Multiple listing (ML) is a practice used to increase the potential for transplant but is controversial due to concerns that it disproportionately benefits patients with greater access to health care resources. RESEARCH QUESTION: Is there disparity in ML practices based on social deprivation in the United States and does ML lead to quicker time to transplant? STUDY DESIGN AND METHODS: A retrospective cohort study of adult (≥ 18 years of age) lung transplant candidates listed for transplant (2005-2018) was conducted. Exclusion criteria included heart only or heart and lung transplant and patients relisted during the observation period. Data were obtained from the United Network for Organ Sharing Standard Transplant Analysis and Research File. The first exposure of interest was the Social Deprivation Index with a primary outcome of ML status, to assess disparities between ML and single listing (SL) participants. The second exposure of interest was ML status with a primary outcome of time to transplant, to assess whether implementation of ML leads to quicker time to transplant. RESULTS: A total of 35,890 patients were included in the final analysis, of whom 791 (2.2%) were ML and 35,099 (97.8%) were SL. ML participants had lower median level of social deprivation (5 units, more often female: 60.0% vs 42.3%) and lower median lung allocation score (35.3 vs 37.3). ML patients were more likely to be transplanted than SL patients (OR, 1.42; 95% CI, 1.17-1.73), but there was a significantly quicker time to transplant only for those whom ML was early (within 6 months of initial listing) (subdistribution hazard ratio, 1.17; 95% CI, 1.04-1.32). INTERPRETATION: ML is an uncommon practice with disparities existing between ML and SL patients based on several factors including social deprivation. ML patients are more likely to be transplanted, but only if they have ML status early in their transplant candidacy. With changing allocation guidelines, it is yet to be seen how ML will change with the implementation of continuous distribution.