Quantitative airway assessment is often performed in specific branches toenable comparison of measurements between patients and over time.
Little isknown on the accuracy in locating these branches.
We determined inter- andintra-observer agreement of manual labeling of segmental bronchi from low-dosechest CT scans.
Methods and Materials
Data We selected 40 participants of the Danish Lung Cancer Screening Trial (Pedersen et al.
and 9 severe COPD according to the GOLD standard (Rabe et al (2007)).
Each subjectcontributed 2 CT scans with an average interval of 4years.
Images were obtained using a Multi Detector CT scanner (16 rowsPhilips Mx 8000) with a low dose (120 kV and 40 mAs),
and reconstructed usinga hard kernel (D) with a resolution of approximately 0.78 mm x 0.78 mm x 1 mmat...
The majority of the labels were assigned to every scan,
however labels R4,
were found to be missing in at least one expert annotation 2,
and 1 times respectively.
6 shows the average inter- and intra-expert agreement for each segmental bronchus.
Average inter- and intra-expert agreement was 74.3% (range 39.2%-100.0%) and 75.3% (range 37.5%-100.0%) respectively.
Agreement was lowest in the lower left lobe (intra- and inter-expert agreement not...
We found considerable disagreement in expert labeling,
large anatomical heterogeneity and changes with inspiration.
Consistent airwaymeasurement cannot be guaranteed based on manual localization.
The Danish Randomized Lung CancerTrial -- Overal Design and Results of the Prevalence Round,
J Thorac Oncol,2009,
608--614 Rabe K.
J.; Global Initiative for...
Jens Petersen is with the Image Group at the Department of Computer Science at the University of Copenhagen,
Denmark; email@example.com. Aasa Feragen is with the Image Group at the Department of Computer Science at the University of Copenhagen,
She is also with theMachine Learning and Computational Biology Research Group at the Max Planck institutefor Developmental Biology and Max Planck Institute for Intelligent Systems,
Germany. Laura Hohwü Thomsen is...