Unsupervised segmentation and quantification of COVID-19 lesions on computed Tomography scans using CycleGAN.

BACKGROUND: Lesion segmentation is a critical step in medical image analysis, and methods to identify pathology without time-intensive manual labeling of data are of utmost importance during a pandemic and in resource-constrained healthcare settings. Here, we describe a method for fully automated segmentation and quantification of pathological COVID-19 lung tissue on chest Computed Tomography (CT) scans without the need for manually segmented training data. METHODS: We trained a cycle-consistent generative adversarial network (CycleGAN) to convert images of COVID-19 scans into their generated healthy equivalents. Subtraction of the generated healthy images from their corresponding original CT scans yielded maps of pathological tissue, without background lung parenchyma, fissures, airways, or vessels. We then used these maps to construct three-dimensional lesion segmentations. Using a validation dataset, Dice scores were computed for our lesion segmentations and other published segmentation networks using ground truth segmentations reviewed by radiologists. RESULTS: The COVID-to-Healthy generator eliminated high Hounsfield unit (HU) voxels within pulmonary lesions and replaced them with lower HU voxels. The generator did not distort normal anatomy such as vessels, airways, or fissures. The generated healthy images had higher gas content (2.45 ± 0.93 vs 3.01 ± 0.84 L, P < 0.001) and lower tissue density (1.27 ± 0.40 vs 0.73 ± 0.29 Kg, P < 0.001) than their corresponding original COVID-19 images, and they were not significantly different from those of the healthy images (P < 0.001). Using the validation dataset, lesion segmentations scored an average Dice score of 55.9, comparable to other weakly supervised networks that do require manual segmentations. CONCLUSION: Our CycleGAN model successfully segmented pulmonary lesions in mild and severe COVID-19 cases. Our model's performance was comparable to other published models; however, our model is unique in its ability to segment lesions without the need for manual segmentations.

Unstable Inflation Causing Injury. Insight from Prone Position and Paired Computed Tomography Scans.

RATIONALE: It remains unclear how prone positioning improves survival in acute respiratory distress syndrome. Using serial computed tomography (CT), we previously reported that "unstable" inflation (i.e., partial aeration with large tidal density swings, indicating increased local strain) is associated with injury progression. OBJECTIVES: We prospectively tested whether prone position contains the early propagation of experimental lung injury by stabilizing inflation. METHODS: Injury was induced by tracheal hydrochloric acid in rats; after randomization to supine or prone position, injurious ventilation was commenced using high tidal volume and low positive end-expiratory pressure. Paired end-inspiratory (EI) and end-expiratory (EE) CT scans were acquired at baseline and hourly up to 3 hours. Each sequential pair (EI, EE) of CT images was superimposed in parametric response maps to analyze inflation. Unstable inflation was then measured in each voxel in both dependent and nondependent lung. In addition, five pigs were imaged (EI and EE) prone versus supine, before and (1 hour) after hydrochloric acid aspiration. MEASUREMENTS AND MAIN RESULTS: In rats, prone position limited lung injury propagation and increased survival (11/12 vs. 7/12 supine; P = 0.01). EI-EE densities, respiratory mechanics, and blood gases deteriorated more in supine versus prone rats. At baseline, more voxels with unstable inflation occurred in dependent versus nondependent regions when supine (41 ± 6% vs. 18 ± 7%; P < 0.01) but not when prone. In supine pigs, unstable inflation predominated in dorsal regions and was attenuated by prone positioning. CONCLUSIONS: Prone position limits the radiologic progression of early lung injury. Minimizing unstable inflation in this setting may alleviate the burden of acute respiratory distress syndrome.

Tidal changes on CT and progression of ARDS.

BACKGROUND: Uncertain prediction of outcome in acute respiratory distress syndrome (ARDS) impedes individual patient management and clinical trial design. OBJECTIVES: To develop a radiological metric of injurious inflation derived from matched inspiratory and expiratory CT scans, calibrate it in a model of experimental lung injury, and test it in patients with ARDS. METHODS: 73 anaesthetised rats (acid aspiration model) were ventilated (protective or non-protective) for up to 4 hours to generate a spectrum of lung injury. CT was performed (inspiratory and expiratory) at baseline each hour, paired inspiratory and expiratory images were superimposed and voxels tracked in sequential scans. In nine patients with ARDS, paired inspiratory and expiratory CT scans from the first intensive care unit week were analysed. RESULTS: In experimental studies, regions of lung with unstable inflation (ie, partial or reversible airspace filling reflecting local strain) were the areas in which subsequent progression of injury was greatest in terms of progressive infiltrates (R=0.77) and impaired compliance (R=0.67, p<0.01). In patients with ARDS, a threshold fraction of tissue with unstable inflation was apparent: >28% in all patients who died and ≤28% in all who survived, whereas segregation of survivors versus non-survivors was not possible based on oxygenation or lung mechanics. CONCLUSIONS: A single set of superimposed inspiratory-expiratory CT scans may predict progression of lung injury and outcome in ARDS; if these preliminary results are validated, this could facilitate clinical trial recruitment and individualised care.

Hyperpolarized gas diffusion MRI of biphasic lung inflation in short- and long-term emphysema models.

During lung inflation, airspace dimensions are affected nonlinearly by both alveolar expansion and recruitment, potentially confounding the identification of emphysematous lung by hyperpolarized helium-3 diffusion magnetic resonance imaging (HP MRI). This study aimed to characterize lung inflation over a broad range of inflation volume and pressure values in two different models of emphysema, as well as in normal lungs. Elastase-treated rats (n = 7) and healthy controls (n = 7) were imaged with HP MRI. Gradual inflation was achieved by incremental changes to both inflation volume and airway pressure. The apparent diffusion coefficient (ADC) was measured at each level of inflation and fitted to the corresponding airway pressures as the second-order response equation, with minimizing residue (χ2 < 0.001). A biphasic ADC response was detected, with an initial ADC increase followed by a decrease at airway pressures >18 cmH2O. Discrimination between treated and control rats was optimal when airway pressure was intermediate (between 10 and 11 cmH2O). Similar findings were confirmed in mice following long-term exposure to cigarette smoke, where optimal discrimination between treated and healthy mice occurred at a similar airway pressure as in the rats. We subsequently explored the evolution of ADC measured at the intermediate inflation level in mice after prolonged smoke exposure and found a significant increase (P < 0.01) in ADC over time. Our results demonstrate that measuring ADC at intermediate inflation enhances the distinction between healthy and diseased lungs, thereby establishing a model that may improve the diagnostic accuracy of future HP gas diffusion studies.

In vivo imaging of the progression of acute lung injury using hyperpolarized [1-13 C] pyruvate.

PURPOSE: To investigate pulmonary metabolic alterations during progression of acute lung injury. METHODS: Using hyperpolarized [1-13 C] pyruvate imaging, we measured pulmonary lactate and pyruvate in 15 ventilated rats 1, 2, and 4 h after initiation of mechanical ventilation. Lung compliance was used as a marker for injury progression. 5 untreated rats were used as controls; 5 rats (injured-1) received 1 ml/kg and another 5 rats (injured-2) received 2 ml/kg hydrochloric acid (pH 1.25) in the trachea at 70 min. RESULTS: The mean lactate-to-pyruvate ratio of the injured-1 cohort was 0.15 ± 0.02 and 0.15 ± 0.03 at baseline and 1 h after the injury, and significantly increased from the baseline value 3 h after the injury to 0.23 ± 0.02 (P = 0.002). The mean lactate-to-pyruvate ratio of the injured-2 cohort decreased from 0.14 ± 0.03 at baseline to 0.08 ± 0.02 1 h after the injury and further decreased to 0.07 ± 0.02 (P = 0.08) 3 h after injury. No significant change was observed in the control group. Compliance in both injured groups decreased significantly after the injury (P < 0.01). CONCLUSIONS: Our findings suggest that in severe cases of lung injury, edema and hyperperfusion in the injured lung tissue may complicate interpretation of the pulmonary lactate-to-pyruvate ratio as a marker of inflammation. However, combining the lactate-to-pyruvate ratio with pulmonary compliance provides more insight into the progression of the injury and its severity. Magn Reson Med 78:2106-2115, 2017. © 2017 International Society for Magnetic Resonance in Medicine.

Semiautomatic segmentation of longitudinal computed tomography images in a rat model of lung injury by surfactant depletion.

Quantitative analysis of computed tomography (CT) is essential to the study of acute lung injury. However, quantitative CT is made difficult by poor lung aeration, which complicates the critical step of image segmentation. To overcome this obstacle, this study sought to develop and validate a semiautomated, multilandmark, registration-based scheme for lung segmentation that is effective in conditions of poor aeration. Expiratory and inspiratory CT images were obtained in rats (n = 8) with surfactant depletion of incremental severity to mimic worsening aeration. Trained operators manually delineated the images to provide a comparative landmark. Semiautomatic segmentation originated from a single, previously segmented reference image obtained at healthy baseline. Deformable registration of the target images (after surfactant depletion) was performed using the symmetric diffeomorphic transformation model with B-spline regularization. Registration used multiple landmarks (i.e., rib cage, spine, and lung parenchyma) to minimize the effect of poor aeration. Then target images were automatically segmented by applying the calculated transformation function to the reference image contour. Semiautomatically and manually segmented contours proved to be highly similar in all aeration conditions, including those characterized by more severe surfactant depletion and expiration. The Dice similarity coefficient was over 0.9 in most conditions, confirming high agreement, irrespective of poor aeration. Furthermore, CT density-based measurements of gas volume, tissue mass, and lung aeration distribution were minimally affected by the method of segmentation. Moving forward, multilandmark registration has the potential to streamline quantitative CT analysis by enabling semiautomatic image segmentation of lungs with a broad range of injury severity.

Vertical gradients in regional alveolar oxygen tension in supine human lung imaged by hyperpolarized 3He MRI.

The purpose of this study was to evaluate whether regional alveolar oxygen tension (P(A)O2) vertical gradients imaged with hyperpolarized (3)He can identify smoking-induced pulmonary alterations. These gradients are compared with common clinical measurements including pulmonary function tests (PFTs), the six minute walk test, and the St. George's Respiratory Questionnaire. 8 healthy non-smokers, 12 asymptomatic smokers, and 7 symptomatic subjects with chronic obstructive pulmonary disease (COPD) underwent two sets of back-to-back P(A)O2 imaging acquisitions in the supine position in two opposite directions (top to bottom and bottom to top), followed by clinically standard pulmonary tests. The whole-lung mean, standard deviation (DP(A)O2) and vertical gradients of P(A)O2 along the slices were extracted, and the results were compared with clinically derived metrics. Statistical tests were performed to analyze the differences between cohorts. The anterior-posterior vertical gradients and DP(A)O2 effectively differentiated all three cohorts (p < 0.05). The average vertical gradient P(A)O2 in healthy subjects was -1.03 ± 0.51 Torr/cm toward lower values in the posterior/dependent regions. The directional gradient was absent in smokers (0.36 ± 1.22 Torr/cm) and was in the opposite direction in COPD subjects (2.18 ± 1.54 Torr/cm). The vertical gradients correlated with smoking history (p = 0.004); body mass index (p = 0.037), PFT metrics (forced expiratory volume in 1 s, p = 0.025; residual volume/total lung capacity percent predicted, p = 0.033) and with distance walked in 6 min (p = 0.009). Regional P(A)O2 data indicate that cigarette smoke induces physiological alterations that are not being detected by the most widely used physiological tests.

Hyperpolarized gas diffusion MRI for the study of atelectasis and acute respiratory distress syndrome.

Considerable uncertainty remains about the best ventilator strategies for the mitigation of atelectasis and associated airspace stretch in patients with acute respiratory distress syndrome (ARDS). In addition to several immediate physiological effects, atelectasis increases the risk of ventilator-associated lung injury, which has been shown to significantly worsen ARDS outcomes. A number of lung imaging techniques have made substantial headway in clarifying the mechanisms of atelectasis. This paper reviews the contributions of computed tomography, positron emission tomography, and conventional MRI to understanding this phenomenon. In doing so, it also reveals several important shortcomings inherent to each of these approaches. Once these shortcomings have been made apparent, we describe how hyperpolarized (HP) gas MRI--a technique that is uniquely able to assess responses to mechanical ventilation and lung injury in peripheral airspaces--is poised to fill several of these knowledge gaps. The HP-MRI-derived apparent diffusion coefficient (ADC) quantifies the restriction of (3) He diffusion by peripheral airspaces, thereby obtaining pulmonary structural information at an extremely small scale. Lastly, this paper reports the results of a series of experiments that measured ADC in mechanically ventilated rats in order to investigate (i) the effect of atelectasis on ventilated airspaces, (ii) the relationship between positive end-expiratory pressure (PEEP), hysteresis, and the dimensions of peripheral airspaces, and (iii) the ability of PEEP and surfactant to reduce airspace dimensions after lung injury. An increase in ADC was found to be a marker of atelectasis-induced overdistension. With recruitment, higher airway pressures were shown to reduce stretch rather than worsen it. Moving forward, HP MRI has significant potential to shed further light on the atelectatic processes that occur during mechanical ventilation.

Positive end-expiratory pressure increments during anesthesia in normal lung result in hysteresis and greater numbers of smaller aerated airspaces.

BACKGROUND: Although it is recognized that pulmonary hysteresis can influence the effects of positive end-expiratory pressure (PEEP), the extent to which expansion of previously opened (vs. newly opening) peripheral airspaces contribute to increased lung volume is unknown. METHODS: Following a recruitment maneuver, rats were ventilated with constant tidal volumes and imaged during ascending and descending ramps of PEEP. RESULTS: The authors estimated peripheral airspace dimensions by measuring the apparent diffusion coefficient of He in 10 rats. In a separate group (n = 5) undergoing a similar protocol, the authors used computerized tomography to quantify lung volume. Hysteresis was confirmed by larger end-inspiratory lung volume (mean ± SD; all PEEP levels included): 8.4 ± 2.8 versus 6.8 ± 2.0 ml (P < 0.001) and dynamic compliance: 0.52 ± 0.12 versus 0.42 ± 0.09 ml/cm H2O (P < 0.001) during descending versus ascending PEEP ramps. Apparent diffusion coefficient increased with PEEP, but it was smaller during the descending versus ascending ramps for corresponding levels of PEEP: 0.168 ± 0.019 versus 0.183 ± 0.019 cm/s (P < 0.001). Apparent diffusion coefficient was smaller in the posterior versus anterior lung regions, but the effect of PEEP and hysteresis on apparent diffusion coefficient was greater in the posterior regions. CONCLUSIONS: The authors' study results suggest that in healthy lungs, larger lung volumes due to hysteresis are associated with smaller individual airspaces. This may be explained by opening of previously nonaerated peripheral airspaces rather than expansion of those already aerated. Setting PEEP on a descending ramp may minimize distension of individual airspaces.

Multislice fractional ventilation imaging in large animals with hyperpolarized gas MRI.

The noninvasive assessment of regional lung ventilation is of critical importance in the quantification of the severity of disease and evaluation of response to therapy in many pulmonary diseases. This work presents, for the first time, the implementation of a hyperpolarized (HP) gas MRI technique to measure whole-lung regional fractional ventilation (r) in Yorkshire pigs (n = 5) through the use of a gas mixing and delivery device in the supine position. The proposed technique utilizes a series of back-to-back HP gas breaths with images acquired during short end-inspiratory breath-holds. In order to decouple the radiofrequency pulse decay effect from the ventilatory signal build-up in the airways, the regional distribution of the flip angle (α) was estimated in the imaged slices by acquiring a series of back-to-back images with no interscan time delay during a breath-hold at the tail end of the ventilation sequence. Analysis was performed to assess the sensitivity of the multislice ventilation model to noise, oxygen and the number of flip angle images. The optimal α value was determined on the basis of the minimization of the error in r estimation: α(opt) = 5-6º for the set of acquisition parameters in pigs. The mean r values for the group of pigs were 0.27 ± 0.09, 0.35 ± 0.06 and 0.40 ± 0.04 for the ventral, middle and dorsal slices, respectively (excluding conductive airways r 0.9). A positive gravitational (ventral-dorsal) ventilation gradient effect was present in all animals. The trachea and major conductive airways showed a uniform near-unity r value, with progressively smaller values corresponding to smaller diameter airways, and ultimately leading to lung parenchyma. The results demonstrate the feasibility of the measurement of the fractional ventilation in large species, and provide a platform to address the technical challenges associated with long breathing time scales through the optimization of acquisition parameters in species with a pulmonary physiology very similar to that of humans.

A multislice single breath-hold scheme for imaging alveolar oxygen tension in humans.

Reliable, noninvasive, and high-resolution imaging of alveolar partial pressure of oxygen (p(A)O(2)) is a potentially valuable tool in the early diagnosis of pulmonary diseases. Several techniques have been proposed for regional measurement of p(A)O(2) based on the increased depolarization rate of hyperpolarized (3) He. In this study, we explore one such technique by applying a multislice p(A)O(2) -imaging scheme that uses interleaved-slice ordering to utilize interslice time-delays more efficiently. This approach addresses the low spatial resolution and long breath-hold requirements of earlier techniques, allowing p(A)O(2) measurements to be made over the entire human lung in 10-15 s with a typical resolution of 8.3 × 8.3 × 15.6 mm(3). PO(2) measurements in a glass syringe phantom were in agreement with independent gas analysis within 4.7 ± 4.1% (R = 0.9993). The technique is demonstrated in four human subjects (healthy nonsmoker, healthy former smoker, healthy smoker, and patient with COPD), each imaged six times on 3 different days during a 2-week span. Two independent measurements were performed in each session, consisting of 12 coronal slices. The overall p(A)O(2) mean across all subjects was 95.9 ± 12.2 Torr and correlated well with end-tidal O(2) (R = 0.805, P < 0.0001). The alveolar O(2) uptake rate was consistent with the expected range of 1-2 Torr/s. Repeatable visual features were observed in p(A)O(2) maps over different days, as were characteristic differences among the subjects and gravity-dependent effects.