TABLE VIII.

LOLA11 Segmentation Challenge results with a short description of the submitted methods.

Team Name	Score	Summary	Further Reading
Fraunhofer ME VIS	0.973	Region-based segmentation initiated by automatically detected seed points followed by morphological closing	[55]
yacta	0.97	Segmentation of trachea-bronchial tree. Landmark determination in left and right lung. Threshold based region growing. Filling of holes in axial slices.	[56]
NIH	0.968	Fully automated PLS method with- out considering pleural fluid as part of lung field.
SmartPaint	0.969	Interactive segmentation	[57]
UCLA Historic	0.963	Threshold-based 3D region growing followed by 6-neighborhood connected component analysis.	[2]
DIAG	0.962	Initial segmentation is performed using a conventional region growing and morphological smoothing. Next, automatic error detection (using shape analysis) is applied. The scans that are likely to contain errors are then segmented by a multi- atlas segmentation.	[58]
GE Research Niskayuna NY	0.952	Context Selective Decision Forests classification.	[59]
MCVGL	0.949	Rib cage detection followed by a robust active shape model for rough segmentation of lung field, and optimal surface finding approach for final refinement.	[60]
Barcelona	0.949	Sequential classifiers using pixel appearance and multiscale analysis of neighbors class likelihood.
CREATIS CLB Lyon	0.948	Region-growing followed by morphological operations.