Detecting regional overdistention during mechanical ventilation using 4-dimensional computed tomography Authors: Yong Song, David Trevascus, Melissa Preissner, Ella Smalley, Kaye Morgan, Martin Donnelley, Graeme R. Zosky, Stephen Dubsky Published: 2026 ABSTRACT Mechanical ventilation (MV) is a critical intervention for managing respiratory compromise and acute respiratory conditions. However, MV can also cause ventilator-induced lung injury (VILI) through mechanisms such as volutrauma, atelectrauma, and biotrauma. Conventional global pressure–volume (PV) loop analysis cannot capture regional mechanical heterogeneity within the lungs, which is central to amplifying VILI, particularly in the context of pre-existing lung injury. In this study, we employed four-dimensional computed tomography (4DCT) combined with propagation-based phase-contrast X-ray imaging to evaluate regional lung mechanics (overdistention and cyclic collapse) in a mouse model subjected to four MV protocols with varying peak inspiratory pressure and positive end-expiratory pressure. Regional PV loop analysis revealed significant overdistention in low tidal volume ventilation settings, which was not apparent in global PV loops, particularly in the group that was ventilated with a low tidal volume and zero PEEP (LVZP). Although global and regional collapse metrics showed no significant differences, regional overdistention indices were markedly higher than global measurements in LVZP, underscoring the limitations of global assessments. These findings highlight the utility of 4DCT for detecting subtle regional mechanical stress and support the integration of regional metrics into MV strategies to enhance lung protection and reduce the risk of VILI. Introduction Mechanical ventilation (MV) is routinely employed during general anesthesia and serves as a critical care intervention for patients with severe acute lung injuries, such as acute respiratory distress syndrome (ARDS). However, while MV is lifesaving, it can also impose excessive mechanical stress on the lungs, leading to ventilator-induced lung injury (VILI)1,2. This iatrogenic complication significantly contributes to patient morbidity and mortality2, highlighting the paradox of MV as both a therapeutic necessity and a potential source of harm. Volutrauma, atelectrauma, and biotrauma are key mechanisms underlying VILI1–3, though their individual contributions are not fully understood. A central concern in these processes is alveolar overdistention (volutrauma), which can overstretch lung tissues, leading to structural damage, increased alveolar-capillary permeability, pulmonary edema, hemorrhage, and surfactant dysfunction2. This cascade promotes alveolar collapse (atelectrauma) and initiates both local and systemic inflammatory responses (biotrauma)2,3. In ARDS, the lungs are highly heterogeneous, with this variability correlating with disease severity4. Overdistention can worsen this heterogeneity, complicating ventilatory management5. Therefore, minimizing alveolar overdistention while maintaining adequate ventilation is critical to mitigating VILI. Strategies include using low tidal volume ventilation to prevent overdistension and appropriately adjusting positive end-expiratory pressure (PEEP) to recruit collapsed alveoli and improve oxygenation—while avoiding further overdistention. Clinically, the respiratory pressure–volume (PV) curve is considered a useful tool for detecting alveolar overdistention and guiding individualized ventilator settings5. Common methods for generating PV curves include the super-syringe technique, constant flow method, and multiple-occlusion approach6. However, these techniques produce global PV curves, which reflect the overall mechanical behavior of the entire lungs and cannot capture regional variations in lung mechanics. Even in healthy lungs, structural and mechanical heterogeneity exists, and this variability is significantly amplified in ARDS1,4,7,8. The spatially heterogeneous response of the lungs to MV can lead to the simultaneous presence of opposing pathological processes—such as overdistention in some regions and collapse in others—within the same lung9,10. This complexity cannot be quantified using global PV analysis. Supporting this, a study using electrical impedance tomography demonstrated that regional PV curves exhibited substantial variability, highlighting the limitations of global assessments and underscoring the importance of regional analysis in optimizing ventilatory strategies11. Unlike electrical impedance tomography, which estimates ventilation distribution based on impedance changes, X-ray Velocimetry (XV), a four-dimensional computed tomography (4DCT) technique, directly measures local lung expansion, offering more accurate quantification of regional ventilation12. With its high spatial and temporal resolution, 4DCT enables precise visualization of lung structures and dynamic ventilation patterns13. Additionally, propagation-based phase-contrast X-ray imaging (PCXI)—a technique that leverages differences in X-ray phase—produces high-contrast images of air–tissue interfaces14,15, allowing for detailed tracking of lung tissue motion. The combination of 4DCT and PCXI provides the images that facilitate high-resolution, within-breath assessment of lung volume and movement using XV. Using this advanced imaging approach, we have previously published a series of studies investigating regional ventilation distribution and lung heterogeneity in animal models with and without pre-existing lung disease16–24. In the present study, we aimed to explore the value of 4DXV lung imaging technology in assessing regional overdistention and atelectasis under various MV strategies, and to understand the regional effects of different inspiratory and expiratory pressure settings. Results Global Vt and EEV The spatial distributions of tidal volume (Vt) and end-expiratory volume (EEV) were not uniform across the lung and varied depending on the ventilation settings after 1 h of mechanical ventilation (MV) (Fig. 1). Fig. 1Cross-sectional images of the lungs from one representative mouse per ventilation group, taken from scan 2 (after 1 h of mechanical ventilation). The top row (A) shows specific tidal volume (Vt), while the bottom row (B) shows specific end-expiratory volume (EEV), with color intensity indicating volume distribution. Cross-sectional images of the lungs from one representative mouse per ventilation group, taken from scan 2 (after 1 h of mechanical ventilation). The top row (A) shows specific tidal volume (Vt), while the bottom row (B) shows specific end-expiratory volume (EEV), with color intensity indicating volume distribution. After 1 h of MV, global Vt was significantly lower in the PEEP groups (HVP and LVP) compared to the zero PEEP groups (HVZP and LVZP) (p < 0.001, Fig. 2A), indicating that the addition of PEEP reduced Vt in both high and low tidal volume ventilation strategies. As expected, the high PIP groups (HVP and HVZP) exhibited higher global Vt than the low PIP groups (LVP and LVZP) (p < 0.001, Fig. 2A), confirming that increased PIP leads to greater tidal volume. Fig. 2Comparison of global tidal volume (Vt, panel A) and end-expiratory volume (EEV, panel B) across the four mechanical ventilation strategies. Values are presented as mean (SD). *p < 0.05, **p < 0.001. Comparison of global tidal volume (Vt, panel A) and end-expiratory volume (EEV, panel B) across the four mechanical ventilation strategies. Values are presented as mean (SD). *p < 0.05, **p < 0.001. For EEV, global values were significantly lower in HVZP compared to HVP (p < 0.01), while no significant difference was observed between LVZP and LVP (p > 0.05, Fig. 2B). This suggests that the absence of PEEP can reduce EEV, but primarily in high PIP ventilation strategies. Global vs. regional overdistention Analysis of the global pressure-volume (PV) loop revealed the presence of an upper inflection point (UIP), indicating potential alveolar overdistention, in both high tidal volume ventilation groups (HVP and HVZP) (Fig. 3A, B), but not in either low tidal volume group (LVZP and LVP) (Fig. 3C, D). Fig. 3Global pressure–volume (PV) loops from representative animals in each ventilation group: (A) HVP, (B) HVZP, (C) LVP, and (D) LVZP. Dashed vertical lines indicate the positions of the lower inflection point (LIP) and upper inflection point (UIP), where present, based on spline fitting. Global pressure–volume (PV) loops from representative animals in each ventilation group: (A) HVP, (B) HVZP, (C) LVP, and (D) LVZP. Dashed vertical lines indicate the positions of the lower inflection point (LIP) and upper inflection point (UIP), where present, based on spline fitting. Importantly however, regional PV loop analysis showed that certain lung regions in the low tidal volume ventilation groups also exhibited a clear UIP. For instance, in subject M71 from the LVZP group, region R1 showed no signs of regional overdistention (Fig. 4A), while region R3 displayed beaking (Fig. 4B), despite the absence of overdistention in the global PV loop (Fig. 4C). Fig. 4Example of lobar PV curves from a mouse (M71) in the LVZP group. Region R1 (A) shows no regional overdistension, while region R3 (B) exhibits a clear UIP, (arrow) indicating regional overdistension. The global PV loop (C) shows no evidence of overdistension. Example of lobar PV curves from a mouse (M71) in the LVZP group. Region R1 (A) shows no regional overdistension, while region R3 (B) exhibits a clear UIP, (arrow) indicating regional overdistension. The global PV loop (C) shows no evidence of overdistension. Consistent with these findings, the overdistention index – which varied substantially within the same ventilation setting, particularly for low tidal MV (Fig. 5A) – was significantly higher in both high PIP groups (HVP and HVZP) compared to the low PIP groups (LVP and LVZP) at both global and regional levels (p < 0.001, Fig. 5B). No significant difference was found between global and regional overdistention in high tidal volume MV (HVP and HVZP). However, in the LVZP group, the regional overdistention index was significantly higher than the global index (p < 0.01), while the difference in the LVP group was not statistically significant (p = 0.212, Fig. 5B). These results indicate that regional overdistention can occur in LVZP, even when not evident in the global PV loop. Fig. 5(A) Histogram of regional overdistension across all groups after 1 h of mechanical ventilation (scan 2). (B) Boxplot comparing overdistension measured by global and regional PV loops and the weighted average of lobar overdistension. Values are presented as median (range). *p < 0.01, **p < 0.001. (A) Histogram of regional overdistension across all groups after 1 h of mechanical ventilation (scan 2). (B) Boxplot comparing overdistension measured by global and regional PV loops and the weighted average of lobar overdistension. Values are presented as median (range). *p < 0.01, **p < 0.001. Global vs. regional collapse The cyclic collapse metric (Fig. 6A) showed variability, though to a lesser extent than the regional overdistention metric. Fig. 6(A) Histogram of regional cyclic collapse across all groups after 1 h of mechanical ventilation (scan 2). (B) Boxplot comparing cyclic collapse metrics measured by global and regional PV loops and the weighted average of lobar redistribution. Values are presented as median (range). *p < 0.05, **p < 0.01. (A) Histogram of regional cyclic collapse across all groups after 1 h of mechanical ventilation (scan 2). (B) Boxplot comparing cyclic collapse metrics measured by global and regional PV loops and the weighted average of lobar redistribution. Values are presented as median (range). *p < 0.05, **p < 0.01. Unlike overdistention, the cyclic collapse metric was similar between global and regional measurements across all MV strategies (Fig. 6B). Notably, regional cyclic collapse in HVP was significantly lower than in the low PIP MV groups, LVP (p < 0.01) and LVZP (p < 0.05) (Fig. 6B). Discussion Using X-ray Velocimetry, we evaluated lung expansion and pressure–volume (PV) loops to assess both global and regional overdistention and collapse under various mechanical ventilation (MV) strategies in a mouse model. Overdistention was consistently observed in both global and regional PV loops during high tidal volume ventilation (HVP and HVZP). Notably, regional PV loop analysis also revealed that overdistention also occurred in the low tidal volume with zero PEEP (LVZP) group, despite its absence in the corresponding global PV loop. This observation was supported by quantitative analysis of the overdistention index, which showed no significant difference between global and regional values in the HVP and HVZP groups. However, in the LVZP group, the regional overdistention index was significantly higher than the global measurement. In contrast, cyclic collapse metrics did not differ significantly between global and regional assessments across ventilation protocols. These findings underscore the value of 4DCT imaging for quantifying regional lung mechanics and highlight the importance of regional overdistention assessment particularly under low tidal volume MV to improve monitoring and guide the choice of lung-protective ventilation strategies. In MV, positive end-expiratory pressure (PEEP) and peak inspiratory pressure (PIP) are critical parameters for optimizing oxygenation and minimizing ventilator-induced lung injury (VILI). High tidal volume ventilation promotes overdistention of aerated lung regions, whereas low tidal volume ventilation reduces this risk and is thus considered a lung-protective strategy. As expected, ventilation with high PIP (20 cmH₂O) resulted in greater tidal volume (Vt) compared to low PIP (12 cmH₂O). Although high PEEP is known to cause overdistention even under low tidal volume conditions25, our study employed a low PEEP setting (2 cmH₂O), which reduced Vt while maintaining constant PIP. As we observed that low PEEP increased end-expiratory volume (EEV) under high PIP conditions, the reduced Vt in pressure-limited modes is likely due to reduced driving pressure and increased baseline lung volume26,27. Ventilation parameters are typically adjusted based on global pressure-volume (PV) loops. The PV curve generally exhibits a sigmoidal shape, characterized by two inflection points: the lower inflection point (LIP) and the upper inflection point (UIP). These inflection points mark transitions in lung compliance, reflecting changes in the number of ventilated alveolar units. The LIP corresponds to the alveolar opening pressure and is used to determine the minimum PEEP required to initiate alveolar recruitment28. In contrast, the UIP indicates the onset of overdistention in already ventilated lung regions. Therefore, setting MV within the linear portion of the PV curve—between the LIP and UIP—can help prevent both intra-tidal alveolar opening and closing (atelectrauma) and overinflation (barotrauma or volutrauma), promoting lung-protective ventilation. Consistent with lung volume data we observed above, we identified the UIP, indicative of overstretch, in the global PV loops of both HVP and HVZP groups. In contrast, no UIP was observed under low tidal volume ventilation. However, in other studies the regional inflection points derived from electrical impedance tomography (EIT) exhibit high variability, reflecting underlying lung heterogeneity11. Importantly, EIT-derived global and regional PV curves have been shown to provide distinct information: the inflection points of the global PV curve differ from those of the regional PV curve which showed a higher LIP and a lower UIP11. Our results confirmed overdistention in the global PV loops for high tidal volume ventilation groups. Notably, significant regional overdistention was also detected in the LVZP group, despite its absence in the global PV loops. This finding was further validated by the regional overdistention index, which was substantially higher than the corresponding global measurement in the LVZP protocol. These data suggest that regional overdistention can occur even under lung-protective ventilation strategies, and that global PV analysis may fail to detect it. Therefore, we conclude that relying solely on global PV loops to titrate ventilator settings may not adequately protect against VILI. Regional PV analysis should be incorporated, particularly in low tidal volume ventilation strategies, to ensure optimal MV settings. Mechanistically, these patterns likely reflect regional redistribution of EEV at low PEEP: small-airway closure and long time-constants in dependent regions reduce local EEV, while non-dependent units retain gas. Therefore, the global EEV remains unchanged even as EEV heterogeneity increases. During the subsequent inspiration, flow preferentially enters already-patent, fast compartments adjacent to closed/poorly aerated areas, creating local stress raisers and regional overdistention despite low global Vt and airway pressures. Such a pattern is consistent with variability in regional PV inflection points and stress amplification in heterogeneous lungs7,11. Similarly, global EEV does not preclude regional airway closure at zero PEEP/low PIP, which can channel flow into already-patent units and manifest as regional overdistention on 4DCT/XV-derived PV loops. Cyclic alveolar collapse refers to the repetitive opening and closing of alveoli during each breath cycle, particularly under MV29. This phenomenon is especially relevant in conditions such as ARDS, where alveolar instability and collapse are common29,30. In our analysis of cyclic collapse metrics derived from global and regional PV loops, we found that PEEP implementation under high PIP settings effectively prevented cyclic collapse. However, unlike the overdistention index, comparisons between global and regional collapse metrics did not reveal significant differences across ventilation strategies in our healthy mouse model. Previous studies have shown that biomechanical stress in lung tissue can vary regionally, and MV may amplify these stresses, leading to non-uniform collapse in injured lungs7. These findings underscore the importance of considering regional metrics in the context of pre-existing lung injury. Thus, incorporating regional assessments of cyclic collapse into ventilation strategies may enhance personalization and help mitigate VILI, particularly in heterogeneous lung conditions such as ARDS. This study has several limitations. Firstly, it was conducted in healthy mouse lungs, which may not reflect the heterogeneity of injured lungs. However, regional overdistention is likely to be more pronounced in ARDS, where a reduced ventilatable “baby lung” and local stress raisers concentrate strain in aerated units adjacent to nonaerated tissue7,8. Imaging work in ARDS also demonstrates marked regional variability in inflection points/compliance, emphasizing a narrow balance between recruitment and overinflation that global metrics may obscure and supporting the use of regional PV analysis (e.g., 4DCT/XV) to titrate PEEP and pressures in heterogeneous lungs. Secondly, airway pressure and volume were measured dynamically during tidal breathing rather than with conventional quasi-static PV curves, meaning resistive and viscoelastic components may have influenced results. Finally, collapse and overdistention were inferred indirectly from spline-derived slope ratios, which provide useful indicators but do not directly quantify alveolar mechanics; absence of a slope change does not necessarily exclude overdistention. These factors should be considered when interpreting our findings and their clinical relevance. In conclusion, our study demonstrates the utility of XV imaging technology in evaluating regional lung mechanics during MV. While global PV loops effectively identified overdistention in high tidal volume settings, regional PV analysis revealed significant overdistention even under low tidal volume ventilation—highlighting the limitations of relying solely on global metrics. The discrepancy between global and regional overdistention indices, particularly in the LVZP group, underscores the importance of incorporating regional assessments to detect subtle mechanical stress that may contribute to VILI. Although our data showed no significant differences between global and regional collapse metrics across ventilation protocols, further studies are required to understand how these measurements vary in models with pre-existing lung disease. Overall, our findings suggest a shift toward regional PV analysis in MV management. This approach enables more accurate detection of overdistention and collapse, thereby enhancing lung-protective strategies and minimizing the risk of VILI. Methods Mouse preparation and mechanical ventilation Eight-week-old female BALB/c mice (approximately 20 g) were obtained from the Monash Animal Research Platform (Melbourne, Australia) and housed at the Imaging and Medical Beamline, Australian Synchrotron (Clayton, Victoria). Mice were anesthetized via intraperitoneal injection of ketamine (150 mg/kg) and xylazine (15 mg/kg). Once a surgical plane of anesthesia was achieved, animals were surgically intubated, mounted upright in a custom-built holder on a rotating stage, and mechanically ventilated for 2 h. Ventilation was performed using one of four combinations of positive end-expiratory pressure (PEEP) and positive inspiratory pressure (PIP): High tidal volume with PEEP (HVP, 2 cmH2O PEEP, 20 cmH2O PIP) (n = 13);High tidal volume with zero PEEP (HVZP, 0 cmH2O PEEP, 20 cmH2O PIP) (n = 13);Low tidal volume with PEEP (LVP, 2 cmH2O PEEP, 12 cmH2O PIP) (n = 13);Low tidal volume with zero PEEP (LVZP, 0 cmH2O PEEP, 12 cmH2O PIP) (n = 14). High tidal volume with PEEP (HVP, 2 cmH2O PEEP, 20 cmH2O PIP) (n = 13); High tidal volume with PEEP (HVP, 2 cmH2O PEEP, 20 cmH2O PIP) (n = 13); High tidal volume with zero PEEP (HVZP, 0 cmH2O PEEP, 20 cmH2O PIP) (n = 13); High tidal volume with zero PEEP (HVZP, 0 cmH2O PEEP, 20 cmH2O PIP) (n = 13); Low tidal volume with PEEP (LVP, 2 cmH2O PEEP, 12 cmH2O PIP) (n = 13); Low tidal volume with PEEP (LVP, 2 cmH2O PEEP, 12 cmH2O PIP) (n = 13); Low tidal volume with zero PEEP (LVZP, 0 cmH2O PEEP, 12 cmH2O PIP) (n = 14). Low tidal volume with zero PEEP (LVZP, 0 cmH2O PEEP, 12 cmH2O PIP) (n = 14). Animal welfare was continuously monitored throughout the procedure. Electrocardiogram (ECG) recordings were used to assess anesthetic depth and detect signs of distress or inadequate sedation. Supplemental anesthesia was administered at the 1-h timepoint if required, and animals received an intraperitoneal saline bolus (200 µL) to maintain hydration during ventilation. Humane endpoints were predefined: euthanasia was performed if animals failed to maintain adequate anesthesia, exhibited physiological distress (e.g., abnormal ECG patterns, irregular respiration despite ventilation), or were unable to recover from the procedure. At the predetermined endpoint (following 2 h of imaging and ventilation), all mice were humanely euthanized via intraperitoneal injection of sodium pentobarbitone (200 mg/kg), in accordance with institutional animal ethics guidelines. The detailed experimental protocol has been described previously16,17. All procedures were approved by the Australian Nuclear Science and Technology Organization (AS2019_02) and the University of Tasmania (A0017845) Animal Ethics Committees, and were conducted in accordance with the National Health and Medical Research Council (Australia) guidelines. The study was reported in compliance with the ARRIVE guidelines (https://arriveguidelines.org). Lung imaging Mice were imaged at the Imaging and Medical Beamline (IMBL) at the Australian Synchrotron using a combination of 4DCT and PCXI. During image capture each mouse was rotated 180 degrees to capture sufficient x-ray projections to perform CT reconstruction. Image capture was coordinated using a custom-designed timing hub, with acquisitions timed to occur 15 times per mouse breath. Images captured at equivalent points during the breath were combined during reconstruction to create 3D lung images at 15 timepoints across the mouse breath31. In this study, only 1-h mechanical ventilation (MV) imaging data were analyzed, as portions of the 2-h dataset have been reported previously17. Using the reconstructed image from the end of expiration, a mask that differentiated between the lungs and external material was created for each scan using the 3D imaging software Avizo (Thermo Fisher Scientific, version: 2022.1, https://www.thermofisher.com/au/en/home/electron-microscopy/products/software-em-3d-vis/avizo-software.html). Cross-correlation-based X-ray Velocimetry (XV) was performed using 32 × 32 × 32 pixel cubes within this mask to create a vector field of lung movement between each pair of consecutive timepoints. The change in air volume between each pair of timepoints was determined by calculating the divergence of these vector fields. This workflow provides time-resolved, regional quantification of lung expansion, with phase-contrast–enhanced air–tissue interface definition, and offers an advantage over impedance-based inference by directly measuring regional expansion. Global and regional analyses The lobes of the mouse right lung cannot be readily segmented in the CT images. Thus, a map of the airway tree for each mouse lung was created by combining a vessel enhancement filter with the lung mask and the CT reconstruction at the end of expiration. The airways leading to right lobes R1-4, and therefore the location of these lobes within the CT reconstruction, were determined manually for each scan using this map. Air volume data was allocated to each lobe using this lobe location data, as shown in Fig. 7. The volume data for the left lobe was split into six sub-lobes (L1-6) using three cuts (Fig. 7). Fig. 7Anatomical positions of lobes L1–L6 within the left lung. The three black dotted lines indicate the cuts used to separate the lobes. Anatomical positions of lobes L1–L6 within the left lung. The three black dotted lines indicate the cuts used to separate the lobes. Average lobar (regional) quantities were calculated by weighing each measure to the mask volume of each lobe and dividing the sum of these weightings by the volume of the whole lung (global):\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$x_{{average}} = \frac{{\sum\nolimits_{{lobe}} {x_{{lobe}} V_{{lobe}} } }}{{V_{{global}} }}$$\end{document} where xaverage is the average of a given quantity (e.g. the overdistention measure), xlobe is the value of the quantity for a given lobe, Vlobe is the volume of that same lobe and Vglobal is the volume of the whole lung. A ventilation pressure value was obtained for each timepoint across the mouse breath by averaging the pressure values at the corresponding times during each scan. These values were paired with whole lung and lobar air volume values throughout the breath to create a set of pressure-volume (PV) loops. The inspiration curve for each PV loop was fitted with a 3-part linear spline32. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V = \left\{ {\begin{array}{*{20}l} {a_{1} } & {P + b_{1} } & {0 \le P < LIP} \\ {a_{2} } & {P + b_{2} } & {LIP \le P < UIP} \\ {a_{3} } & {P + b_{3} } & {UIP \le P} \\ \end{array} } \right\},$$\end{document}where P is the pressure from MV, V is the volume within the given lobe (or for the whole lung), and LIP and UIP are the lower and upper inflection points of the spline respectively. For any inspiration curve, overdistention is marked by a decrease in volume intake after UIP. To determine the amount of overdistention present in each lobe, an overdistention index was created using the parameters of the linear spline fit:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$od = \left\{ {\begin{array}{*{20}l} 1 & {r \le 0,a_{2} < 0} \\ {1 - r} & {0 < r < 1} \\ 0 & {r \ge 1} \\ \end{array} } \right\},$$\end{document} where r = a3/a2. This index decreases as the volume intake after UIP increases and gets closer to the intake between LIP and UIP. In cases where either a2 < 0 or a3 < 0 the index defaults to one. In cases where a3 > a2 the index defaults to zero. Similarly, collapse is marked by lower volume intake before LIP. A collapse metric was calculated using parameters from the linear spline fit: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$rd = \left\{ {\begin{array}{*{20}l} {1 - a_{1} /a_{2} } & {a_{1} < a_{2} } \\ 0 & {a_{1} \ge a_{2} } \\ \end{array} } \right\}$$\end{document} This index decreases as a1 increases and gets closer to a2. In the case where a1 > a2 the index defaults to zero. Statistical analysis Statistical analysis was conducted using SPSS (version 20.0; SPSS Inc., Chicago, IL). Paired t-tests were employed to compare differences between global and regional parameters (overdistention and cyclic collapse indices). Comparisons among different ventilatory strategies were performed using one-way ANOVA, followed by Tukey’s Honestly Significant Difference (HSD) test for post hoc analysis. Data were assessed for normality, and variables not normally distributed were log₁₀-transformed. A p-value of less than 0.05 was considered statistically significant.