Abstract #4078

Automatic lung segmentation for hyperpolarized gas MRI using transferred generative adversarial network and three-view aggregation

Shih-Kang Chao¹, Ummul Afia Shammi², Lucia Flors-Blasco³, Talissa Altes⁴, John Mugler^5,6, Craig Meyer^5,6, Jaime Mata⁶, Wilson Miller⁶, and Robert Thomen^2,4

¹Department of Statistics, University of Missouri, Columbia, MO, United States, ²Department of Biomedical, Biological & Chemical Engineering, University of Missouri, Columbia, MO, United States, ³Keck School of Medicine, University of Southern California, Los Angeles, CA, United States, ⁴Department of Radiology, University of Missouri, Columbia, MO, United States, ⁵Department of Biomedical Engineering, University of Virginia, Charlottesville, VA, United States, ⁶Department of Radiology and Medical Imaging, University of Virginia, Charlottesville, VA, United States

Synopsis

We evaluate an automatic lung segmentation approach that aggregates the predicted mask of coronal, axial, and sagittal views generated by a deep conditional generative adversarial network (GAN) whose only input is the hyperpolarized gas (HPG) MRI. On five test subjects with ventilation defect percentages [VDP] of 25-38%, our method achieved an average Dice score of 87.72, and above 90 on a healthy control subject. The slice-wise Dice score had an average correlation of 0.72 with the human expert and a median correlation of -0.79 with VDP, and both are significant for 4 out of 5 test patients at level 1%.

How to access this content:

For one year after publication, abstracts and videos are only open to registrants of this annual meeting. Registrants should use their existing login information. Non-registrant access can be purchased via the ISMRM E-Library.

After one year, current ISMRM & ISMRT members get free access to both the abstracts and videos. Non-members and non-registrants must purchase access via the ISMRM E-Library.

After two years, the meeting proceedings (abstracts) are opened to the public and require no login information. Videos remain behind password for access by members, registrants and E-Library customers.

Click here for more information on becoming a member.