Quantitative Imaging Laboratory

About Us

The Quantitative Imaging Laboratory (QIL) at National Jewish Health (NJH), directed by Stephen Humphries PhD and David Lynch MB, focuses on development and application of the latest medical image analysis methods in clinical and translational research. The QIL was established in 2008 to support radiologic imaging in the COPDGene research study and provide image archival and analysis for the NJH Department of Radiology. Today, the QIL’s main goal is to develop and apply novel quantitative image analysis methods, including those based on Artificial Intelligence (AI), that have meaningful clinical impact.

3D image of sinuses.The QIL also provides quality assurance and image data management and is currently the primary image analysis core for local, national and international multicenter imaging research studies including: COPDGene, which includes 10,000 patients and nearly 20,000 CT exams, and the Pulmonary Fibrosis Foundation Registry which aims to create an imaging directory consisting of CT images of over 1,000 subjects.

 

Research Activities

The QIL develops novel techniques to recognize and characterize normal and abnormal anatomy, researches methods for efficient large-scale processing of medical images and provides quantitative image analysis for research and clinical purposes. The QIL is actively engaged in research on deep learning, a class of AI algorithms that use multi-layer computer models called neural networks to “learn” from example data rather than being explicitly programmed.

 

Lung Fibrosis Quantification on CT

Sequential quantitative CT measurement of fibrosis in a patient with idiopathic pulmonary fibrosis at baseline and at 14 months, using Data-Driven Texture Analysis. The extent of fibrosis, shown in red, increased from 39%25 to 52%25, and the patient deteriorated physiologically, with decrease in Forced Vital Capacity from 66%25 predicted to 45%25.

Researchers in the QIL have developed deep learning programs that are capable of automatic detection and quantification of pulmonary fibrosis on CT scans. The Data-Driven Texture Analysis (DTA) algorithm developed by the QIL correlates with radiologist visual assessment and pulmonary function, is sensitive to change in extent of fibrosis, and can predict survival. Quantitative analysis of lung fibrosis has been applied in multiple completed and ongoing clinical trials of treatment in fibrotic lung disease.

Quantitative Analysis of Sinus CT

Sinus Opacification Quantification Results Visual Report, Example 1

The QIL has developed deep learning methods for automatic 3D segmentation of the nasal sinuses on CT scans. This enables quantitative assessment of sinus volumes and the extent of sinus opacification as a marker of sinusitis. Recent research indicates a correlation between CT sinusitis scores and asthma diagnosis, eosinophil count and pulmonary function. Support from the State of Colorado has facilitated QIL research and participation in a multicenter trial for severe nasal polyposis.

Quantitative Analysis of COPD and Smoking Related Disease on CT

Quantitative Analysis of COPD and Smoking Related Disease on CT

The QIL is the imaging core for the COPDGene study; imaging in COPDGene has resulted in over 100 publications and several important advances, summarized in a recent review article by Bhatt et al. The QIL has been particularly interested in the role of CT in classifying emphysema, and in the potential of deep learning to assist with this task. Emphysema pattern has been shown to be an important indicator of disease severity but is time consuming and subjective to perform visually. The QIL has developed a deep learning algorithm to automatically score emphysema severity pattern on CT according a classification system developed by the Fleischner Society. Further research includes quantitative phenotyping of emphysema, air trapping, and airway analysis.

QIL Collaborators

 

Faculty

Dr. Stephen HumphriesStephen M. Humphries, PhD, Director and Imaging Scientist

Assistant Professor in the Department of Radiology at National Jewish Health, is a medical physicist and imaging scientist with a major interest in quantitative CT of fibrotic and obstructive diffuse lung diseases.

As the Director of the Quantitative Imaging Laboratory at National Jewish Health, Dr. Humphries oversees all laboratory efforts including quality assurance, storage, and image analysis for multi-center CT studies of COPD, pulmonary fibrosis and rhinosinusitis.

 

Dr. David LynchDavid Lynch, MD, Medical Director

Professor of Radiology at National Jewish Health, principal radiology investigator for the imaging core of COPDGene (which has enrolled over 10,000 individuals with and without COPD, now followed for over 10 years), and IPFNet. He has authored or coauthored over 300 papers and several books on lung imaging, He is Past President of the Society of Thoracic Radiology and of the Fleischner Society.

 

 

Staff

Alex Kluiber, Database Administrator
KluiberA@njhealth.org

Lucas Veitel, QIL Quality Project Coordinator
VeitelL@NJHealth.org

Aleena Notary, Research Software Engineer
NotaryA@NJHealth.org

Juan Pablo Centeno, Research Software Engineer
CentenoJP@NJHealth.org

Devlin Thieke, Research Analyst
ThiekeD@NJHealth.org

Alison Pegg, Administrative Assistant
PeggA@njhealth.org

Resources & Services

The QIL utilizes a variety of software to view and analyze radiologic images including proprietary software developed in the lab.

Learn about our Lab Services.

 

Contact Information

Attn: Quantitative Imaging Lab
National Jewish Health
1400 Jackson Street, K012c
Denver, CO 80206

Phone: 303.270.2529
Fax: 303.270.2538

Publications

Humphries SM, Notary AM, Centeno JP, Strand MJ, Crapo JD, Silverman EK, Lynch DA, Genetic Epidemiology of COPD (COPDGene) Investigators. Deep Learning Enables Automatic Classification of Emphysema Pattern at CT. Radiology. 2019 Dec 3:191022. https://www.ncbi.nlm.nih.gov/pubmed/31793851

View all Publications