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

Intensive Care Medicine Experimental

Table 2 CT-derived lung measurement

Variables	Median [IQR]	Range
Alveolar recruitment PEEP_5–15—% lung weight	5.8 [4.2–9.2]	− 3.3–21.6
Lung weight at PEEP 5—g	1346 [1152–1681]	747–2343
Lung weight at PEEP 15—g	1347 [1155–1715]	712–2328
Non-aerated lung at PEEP 5—% lung weight	59 [44–68]	25–91
Non-aerated lung at PEEP 15—% lung weight	55 [36–60]	21–86
Tidal hyperinflation—mL.kg⁻¹ PBW	0.4 [0.2–1]	0.0–5.2
Tidal recruitment—% lung weight	4.2 [1.9–6.4]	− 6.2–14.3
End-expiratory lung volume at PEEP 5—mL	682 [473–880]	141–2748
End-expiratory lung volume at PEEP 15—mL	970 [789–1354]	251–3365
Hyperinflation at PEEP 5—mL.kg⁻¹ PBW	0.1 [0.1–0.4]	0.0–20.3
Hyperinflation at PEEP 15—mL.kg⁻¹ PBW	0.4 [0.1–0.7]	0.0–29.7

Values are pooled values from measurements of observer#1 (measurement #1 and #2), observer#2 and machine learning
CT denotes computed tomography; IQR, interquartile range; PEEP, positive end-expiratory pressure; PEEP_5–15, PEEP change from 5 to 15 cmH₂O; and PBW predicted body weight