A computational platform for quantification of infectious lung disease using PET-CT imaging

Objectives To design fast and robust automated image analysis methods that can accurately facilitate diagnostic measurements of pulmonary lesions in small animal models.

Methods We analyzed 85 PET-CTs from 18 different rabbits at various time points (0-to-38 weeks). Each rabbit was infected with the Mycobacterium tuberculosis strain H37Rv. In addition, 10 PET-CTs of Ferrets, all infected with the H1N1 influenza virus, were analyzed at four different time points (0-to-72 hours). See Figure1 for the flowchart of the overall methodology to segment lesion using PET-CTs. The resulting delineations were analyzed using dice coefficient (DSC) and metabolic volume changes over time, the standard uptake value (SUV), and correlation with expert derived volumes were also given. For longitudinal quantification, we applied group-wise image registration with CT images of the same animal, at different time points; resulting deformation fields were applied to PET counterparts to align images accordingly (Figure2). For observer agreement studies, two experts conducted their evaluations blindly.

Results DSC of PET was 86.0 +/- 3.6% and 84.0 +/- 5.8%. Lung (CT) segmentation accuracy was 86.0+/-7.1% and 83.4+/-8.6% in rabbits and ferrets, respectively. Observer agreement rates were reported (in DSC) as 84.7+/-6.0% (intra, rabbit) and 80.6+/-9.5% (inter, rabbit), 80.3+/-4.4% (intra, ferret), and 77.2+/-11.2% (inter, ferret). Mean registration errors were 2.66 mm and 3.93 mm, in rabbits and ferrets respectively. While manual segmentation can take up to 90 minutes, our segmentation method took only 16 seconds for approximately 250 slices; segmentation of PET images took only 11 seconds on average.

Conclusions Quantifying infectious lung diseases using PET-CT images can be acquired quickly and accurately with our comprehensive automated software platform.

Research Support This research is supported by CIDI, the intramural research program of the National Institute of Allergy and Infectious Diseases (NIAID) and the National Institute of Biomedical Imaging and Bioengineering (NIBIB). Dr. Jain acknowledges the NIH Director’s New Innovator Award (OD006492). The rabbit infection study is funded by HHMI, NIAD R01AI079590, and R01A1035272.