Skip to main content
Intensive Care Medicine Experimental
Fig. 2 | Intensive Care Medicine Experimental

Fig. 2

From: Precision of CT-derived alveolar recruitment assessed by human observers and a machine learning algorithm in moderate and severe ARDS

Fig. 2

Accuracy of lung segmentation in various conditions (intra-observer, inter-observer human–human and inter-observer human–ML). Each point refers to one computed tomography volume acquisition (four computed tomography volumes were acquired for each patient, i.e., at both end-expiration and end-inspiration at the PEEP level chosen by attending physician, and at end-expiration at PEEP 5 and 15 cmH2O). †p < 0.05 vs intra-observer, ‡p < 0.05 vs inter-observer (human–human). ML denotes machine learning and PEEP positive end-expiratory pressure

Back to article page