Please consider downloading the latest version of Internet Explorer
to experience this site as intended.
Tools Search Main Menu


Interpretable Prediction of Obstructive Lung Disease from Chest Radiographs with Deep Learning

Please join the Goergen Institute for Data Science for Interpretable Prediction of Obstructive Lung Disease from Chest Radiographs with Deep Learning, a research seminar with Tolga Tasdizen, Professor of Electrical and Computer Engineering at University of Utah.

Abstract: Chronic obstructive pulmonary disease (COPD), the third leading cause of death worldwide, is often underdiagnosed. Our hypothesis is that the diagnosis rate for COPD can be improved using the routine chest x-ray (CXR) as a screening tool. CXRs can show indirect evidence of COPD such as emphysema; however, radiologist do not routinely comment on findings related to COPD while reading CXRs. Pulmonary function testing (PFT) which measures the amounts of air a patient can inhale and exhale is the main diagnostic exam for COPD. We train a convolutional neural network to predict COPD from CXRs using near-concurrent PFT data as ground truth. We demonstrate that this image model has better prediction accuracy compared to a state-of-the-art natural language processing model that operates on radiologist text reports. Furthermore, we propose a method for making our image model human interpretable. We use a conditional generative adversarial network where the generator attempts to learn to shift the output of a regressor through creating disease effect maps that are added to the original images. Meanwhile, the regressor is trained to predict the original regression value for the modified images. A model trained with this technique learns to provide visualization for how the image would appear at different stages of the disease which is used as an interpretation for the regressor’s output. 

Bio: Dr. Tasdizen is a Professor of Electrical and Computer Engineering at the University of Utah and a faculty member of the Scientific Computing an Imaging (SCI) Institute.  Dr. Tasdizen earned the BS degree in Electrical Engineering from Bogazici University in 1995, and the PhD degree in Engineering from Brown University in 2001. He is a senior member of the IEEE and is Senior Area Editor for the IEEE Transactions on Image Processing. His research interests are in the areas of machine learning, image processing and analysis. He has worked on applications of these research areas to biomedicine, radiology, neuroscience, climate science, material science and security. He has over 120 peer-reviewed journal and conference publications. Dr. Tasdizen is the recipient of a NSF Early Career Award and his research group has received funding from the NIH, NSF, DHS, DOE and ONR.

Dial-In Information

Please click the link below to join the webinar:

Or iPhone one-tap : 
    US: +16468769923,,97885004386#  or +13017158592,,97885004386# 
Or Telephone:
    Dial(for higher quality, dial a number based on your current location):
        US: +1 646 876 9923  or +1 301 715 8592  or +1 312 626 6799  or +1 669 900 6833  or +1 253 215 8782  or +1 346 248 7799 
Webinar ID: 978 8500 4386
    International numbers available:

Or an H.323/SIP room system:
    H.323: (US West) (US East) (India Mumbai) (India Hyderabad) (Amsterdam Netherlands) (Germany) (Australia Sydney) (Australia Melbourne) (Singapore) (Brazil) (Canada Toronto) (Canada Vancouver) (Japan Tokyo) (Japan Osaka)
    Meeting ID: 978 8500 4386

Friday, April 2 at 2:00pm to 3:00pm

Virtual Event

Recent Activity