We analyzed the fractal dimension (Df) of lung gas and blood distribution imaged with synchrotron radiation K-edge subtraction (KES), in six anesthetized adult New Zealand White rabbits. KES imaging was performed in upright position during stable Xe gas (64% in O₂ ) inhalation and iodine infusion (Iomeron, 350 mg/mL), respectively, at baseline and after induced bronchoconstriction by aerosolized methacholine (125 mg/mL, 90 s) and bronchodilator (salbutamol, 10 mg/mL, 90 s) inhalation, at two axial image levels. Lung Xe and iodine images were segmented, and maps of regional lung gas and blood fractions were computed. The Df of lung gas (Df_Xe ) and blood (Df_Iodine ) distribution was computed based on a log-log plot of variation coefficient as a function of region volume. Df_Xe decreased significantly during bronchoconstriction (P < 0.0001), and remained low after salbutamol. Df_Iodine depended on the axial image level (P < 0.0001), but did not change with bronchoconstriction. Df_Xe was significantly associated with arterial [Formula: see text] (R = 0.67, P = 0.002), and negatively associated with [Formula: see text] (R = -0.62, P = 0.006), respiratory resistance (R = -0.58, P = 0.011), and elastance (R = -0.55, P = 0.023). These data demonstrate the reduced Df of gas distribution during acute bronchoconstriction, and the association of this parameter with physiologically meaningful variables. This finding suggests a decreased complexity and space-filling properties of lung ventilation during bronchoconstriction, and could serve as a functional imaging biomarker in obstructive airway diseases.

NEW & NOTEWORTHY Here, we used an energy-subtractive imaging technique to assess the fractal dimension (Df) of lung gas and blood distribution and the effect of acute bronchoconstriction. We found that Df of gas significantly decreases in bronchoconstriction. Conversely, Df of blood exhibits gravity-dependent changes only, and is not affected by acute bronchoconstriction. Our data show that the fractal dimension of lung gas detects the emergence of clustered rather than scattered loss of ventilatory units during bronchoconstriction.