PROJECT SUMMARY
Idiopathic pulmonary fibrosis (IPF) is a smoking-related disease that is end-stage at diagnosis, with a median
survival of 3.8 years. Current treatments slow the future progression of IPF but do not reverse the disease. Thus,
there is an important need to detect patients who are at risk of developing IPF and may benefit from earlier
initiation of anti-fibrotic medications. Recent work has validated changes in the lung parenchyma on chest
computed tomography (CT) scans of smokers that represent early pulmonary fibrosis. These parenchymal
changes, either detected visually and called interstitial lung abnormalities (ILA), or through an automated image
analysis tool developed by Dr. Choi’s lab called quantitative interstitial abnormalities (QIA), are associated with
poor lung function, exercise limitations, and increased mortality. However, QIA caught at a point in time likely
represents heterogeneous disease, encompassing both the non-progressive and transient processes that are
caught on CT, and the clinically meaningful early smoking-related disease that will eventually progress to IPF.
Radiomics may enable the characterization of, and increase specificity of, QIA phenotypes associated with IPF.
Radiomics analyses use high-throughput computing to measure many features that are already available but not
typically measured in CT scans, including measurements and statistics about the textures, shapes, gray levels
within regions of interest, and relationships amongst voxels. Radiomics may provide a novel, specific tool to
stratify disease severity and predict disease progression of early pulmonary fibrosis.
Dr. Choi will use radiomics features to distinguish heterogeneous phenotypes of smoking-related lung injury. In
Aim 1, she will characterize the radiomics signatures of smokers with early pulmonary fibrosis (QIA) at risk for
worse clinical outcomes. In Aim 2, she will move her focus to identifying the patients at the earliest stage of lung
injury. She will characterize the radiomics signatures of smokers with visually normal CTs at risk for progression
to early pulmonary fibrosis and worse clinical outcomes.
Dr. Choi will perform this work within the Division of Pulmonary and Critical Care Medicine, at Brigham and
Women’s Hospital (BWH), a core teaching hospital of the Harvard Medical School, under the mentorship of Dr.
George Washko, an expert in quantitative medical imaging analysis and co-director of the Applied Chest Imaging
Laboratory at BWH. With her mentors and Scientific Advisory Committee, Dr. Choi has developed a training plan
to gain proficiency in big data preparation and analysis, machine learning algorithms, advanced statistical
methods, and programming; to maintain and deepen her understanding of pulmonary fibrosis and smoking-
related lung disease; and to hone her skills in scientific manuscript preparation, grant-writing, and effective
communication. Dr. Choi’s long-term goal is to become a physician-scientist that combines her clinical expertise
in pulmonary medicine with advanced technical and research expertise in data science, in order to leverage big
data for the improved detection and treatment of lung diseases.