The rib plots show only moderate agreement across all models, mainly due to segmentation errors in four of the six models, which sometimes include parts of neighboring ribs. These errors are likely a result of fractured ribs in the training data. After excluding the affected models, agreement between MOOSE and CADS improves, with Dice scores above 80% and volume overlap up to 95%. Remaining differences are due to segmentation coverage: MOOSE fully includes the costovertebral joints, while CADS captures them only partially, resulting in consistently higher rib volumes for MOOSE.