Association of Ground Glass Opacities with Systemic Inflammation and Progression of Emphysema.

RATIONAL: Ground glass opacities (GGO) in the absence of interstitial lung disease are understudied. OBJECTIVE: To assess the association of GGO with white blood cells (WBCs) and progression of quantified chest CT emphysema. METHODS: We analyzed data of participants in the Subpopulations and Intermediate Outcome Measures In COPD Study (SPIROMICS). Chest radiologists and pulmonologists labeled regions of the lung as GGO and adaptive multiple feature method (AMFM) trained the computer to assign those labels to image voxels and quantify the volume of the lung with GGO (%GGOAMFM). We used multivariable linear regression, zero-inflated negative binomial, and proportional hazards regression models to assess the association of %GGOAMFM with WBC, changes in %emphysema, and clinical outcomes. MEASUREMENTS AND MAIN RESULTS: Among 2,714 participants, 1,680 had COPD and 1,034 had normal spirometry. Among COPD participants, based on the multivariable analysis, current smoking and chronic productive cough was associated with higher %GGOAMFM. Higher %GGOAMFM was cross-sectionally associated with higher WBCs and neutrophils levels. Higher %GGOAMFM per interquartile range at visit 1 (baseline) was associated with an increase in emphysema at one-year follow visit by 11.7% (Relative increase; 95%CI 7.5-16.1%;P<0.001). We found no association between %GGOAMFM and one-year FEV1 decline but %GGOAMFM was associated with exacerbations and all-cause mortality during a median follow-up time of 1,544 days (Interquartile Interval=1,118-2,059). Among normal spirometry participants, we found similar results except that %GGOAMFM was associated with progression to COPD at one-year follow-up. CONCLUSIONS: Our findings suggest that GGOAMFM is associated with increased systemic inflammation and emphysema progression.

Quantifying the Preferential Direction of the Model Gradient in Adversarial Training With Projected Gradient Descent.

Adversarial training, especially projected gradient descent (PGD), has proven to be a successful approach for improving robustness against adversarial attacks. After adversarial training, gradients of models with respect to their inputs have a preferential direction. However, the direction of alignment is not mathematically well established, making it difficult to evaluate quantitatively. We propose a novel definition of this direction as the direction of the vector pointing toward the closest point of the support of the closest inaccurate class in decision space. To evaluate the alignment with this direction after adversarial training, we apply a metric that uses generative adversarial networks to produce the smallest residual needed to change the class present in the image. We show that PGD-trained models have a higher alignment than the baseline according to our definition, that our metric presents higher alignment values than a competing metric formulation, and that enforcing this alignment increases the robustness of models.

Mucus Plugs and Emphysema in the Pathophysiology of Airflow Obstruction and Hypoxemia in Smokers.

Rationale: The relative roles of mucus plugs and emphysema in mechanisms of airflow limitation and hypoxemia in smokers with chronic obstructive pulmonary disease (COPD) are uncertain.Objectives: To relate image-based measures of mucus plugs and emphysema to measures of airflow obstruction and oxygenation in patients with COPD.Methods: We analyzed computed tomographic (CT) lung images and lung function in participants in the Subpopulations and Intermediate Outcome Measures in COPD Study. Radiologists scored mucus plugs on CT lung images, and imaging software automatically quantified emphysema percentage. Unadjusted and adjusted relationships between mucus plug score, emphysema percentage, and lung function were determined using regression.Measurements and Main Results: Among 400 smokers, 229 (57%) had mucus plugs and 207 (52%) had emphysema, and subgroups could be identified with mucus-dominant and emphysema-dominant disease. Only 33% of smokers with high mucus plug scores had mucus symptoms. Mucus plug score and emphysema percentage were independently associated with lower values for FEV1 and peripheral oxygen saturation (P < 0.001). The relationships between mucus plug score and lung function outcomes were strongest in smokers with limited emphysema (P < 0.001). Compared with smokers with low mucus plug scores, those with high scores had worse COPD Assessment Test scores (17.4 ± 7.7 vs. 14.4 ± 13.3), more frequent annual exacerbations (0.75 ± 1.1 vs. 0.43 ± 0.85), and shorter 6-minute-walk distance (329 ± 115 vs. 392 ± 117 m) (P < 0.001).Conclusions: Symptomatically silent mucus plugs are highly prevalent in smokers and independently associate with lung function outcomes. These data provide rationale for targeting patients with mucus-high/emphysema-low COPD in clinical trials of mucoactive treatments.Clinical trial registered with www.clinicaltrials.gov (NCT01969344).

Idiopathic Pulmonary Fibrosis: Data-driven Textural Analysis of Extent of Fibrosis at Baseline and 15-Month Follow-up.

Purpose To evaluate associations between pulmonary function and both quantitative analysis and visual assessment of thin-section computed tomography (CT) images at baseline and at 15-month follow-up in subjects with idiopathic pulmonary fibrosis (IPF). Materials and Methods This retrospective analysis of preexisting anonymized data, collected prospectively between 2007 and 2013 in a HIPAA-compliant study, was exempt from additional institutional review board approval. The extent of lung fibrosis at baseline inspiratory chest CT in 280 subjects enrolled in the IPF Network was evaluated. Visual analysis was performed by using a semiquantitative scoring system. Computer-based quantitative analysis included CT histogram-based measurements and a data-driven textural analysis (DTA). Follow-up CT images in 72 of these subjects were also analyzed. Univariate comparisons were performed by using Spearman rank correlation. Multivariate and longitudinal analyses were performed by using a linear mixed model approach, in which models were compared by using asymptotic χ2 tests. Results At baseline, all CT-derived measures showed moderate significant correlation (P < .001) with pulmonary function. At follow-up CT, changes in DTA scores showed significant correlation with changes in both forced vital capacity percentage predicted (ρ = -0.41, P < .001) and diffusing capacity for carbon monoxide percentage predicted (ρ = -0.40, P < .001). Asymptotic χ2 tests showed that inclusion of DTA score significantly improved fit of both baseline and longitudinal linear mixed models in the prediction of pulmonary function (P < .001 for both). Conclusion When compared with semiquantitative visual assessment and CT histogram-based measurements, DTA score provides additional information that can be used to predict diminished function. Automatic quantification of lung fibrosis at CT yields an index of severity that correlates with visual assessment and functional change in subjects with IPF. © RSNA, 2017.

4D magnetic resonance flow imaging for estimating pulmonary vascular resistance in pulmonary hypertension.

PURPOSE: To develop an estimate of pulmonary vascular resistance (PVR) using blood flow measurements from 3D velocity-encoded phase contract magnetic resonance imaging (here termed 4D MRI). MATERIALS AND METHODS: In all, 17 patients with pulmonary hypertension (PH) and five controls underwent right heart catheterization (RHC), 4D and 2D Cine MRI (1.5T) within 24 hours. MRI was used to compute maximum spatial peak systolic vorticity in the main pulmonary artery (MPA) and right pulmonary artery (RPA), cardiac output, and relative area change in the MPA. These parameters were combined in a four-parameter multivariate linear regression model to arrive at an estimate of PVR. Agreement between model predicted and measured PVR was also evaluated using Bland-Altman plots. Finally, model accuracy was tested by randomly withholding a patient from regression analysis and using them to validate the multivariate equation. RESULTS: A decrease in vorticity in the MPA and RPA were correlated with an increase in PVR (MPA: R(2) = 0.54, P < 0.05; RPA: R(2) = 0.75, P < 0.05). Expanding on this finding, we identified a multivariate regression equation that accurately estimates PVR (R(2) = 0.94, P < 0.05) across severe PH and normotensive populations. Bland-Altman plots showed 95% of the differences between predicted and measured PVR to lie within 1.49 Wood units. Model accuracy testing revealed a prediction error of ∼20%. CONCLUSION: A multivariate model that includes MPA relative area change and flow characteristics, measured using 4D and 2D Cine MRI, offers a promising technique for noninvasively estimating PVR in PH patients. J. MAGN. RESON. IMAGING 2016;44:914-922.

Galectin-3 levels are associated with right ventricular functional and morphologic changes in pulmonary arterial hypertension.

The response of the right ventricle (RV) to pulmonary arterial hypertension (PAH) involves changes in contractile function, chamber size, hypertrophy, and extracellular matrix (ECM). Galectin-3 (Gal-3) is a mediator of myocardial ECM metabolism and biomarker for left heart remodeling, yet its ability to reflect RV remodeling is unknown. We hypothesized that serum Gal-3 levels correlate with RV morphology and function in PAH, and that Gal-3 is associated with circulating markers of ECM. Fifteen subjects with PAH and 10 age-matched controls underwent same-day echocardiography, cardiac magnetic resonance (CMR) imaging, and phlebotomy for Gal-3 and ECM biomarkers including N-terminal propeptide of type III collagen type (PIIINP), tissue inhibitor of metalloproteinase-1 (TIMP-1), and hyaluronic acid (HA). RV ejection fraction, end diastolic volume index, end systolic volume index, and mass index were calculated using CMR. Echocardiography was used to estimate RV systolic pressure and measure RV strain. Serum Gal-3, TIMP-1, and HA levels were all significantly increased in PAH subjects when compared to controls. Gal-3 correlated with RV ejection fraction (ρ -0.44, p 0.03), end diastolic volume index (ρ 0.42, p 0.03), end systolic volume index (ρ 0.44, p 0.027), mass index (ρ 0.47, p 0.016), systolic pressure (ρ 0.55, p < 0.001), and strain (ρ 0.43, p 0.03). Gal-3 levels positively correlated with the ECM markers TIMP-1 and HA but not with PIIINP. In conclusion, Gal-3 levels are associated with multiple indices of RV function and morphology. Gal-3 may represent a novel biomarker for RV remodeling and associated ECM turnover in PAH.

Vorticity is a marker of right ventricular diastolic dysfunction.

Right ventricular diastolic dysfunction (RVDD) is an important prognostic indicator in pulmonary arterial hypertension (PAH). RV vortex rings have been observed in healthy subjects, but their significance in RVDD is unknown. Vorticity, the local spinning motion of an element of fluid, may be a sensitive measure of RV vortex dynamics. Using four-dimensional (4D) flow cardiac magnetic resonance imaging (CMR), we investigated the relationship between right heart vorticity with echocardiographic indexes of RVDD. Thirteen (13) PAH subjects and 10 controls underwent same-day 4D flow CMR and echocardiography. RV diastolic function was assessed using trans-tricuspid valve (TV) early (E) and late (A) velocities, E/A ratio, and e' and a' tissue Doppler velocities. RV and right atrial (RA) integrated mean vorticity was calculated for E and A-wave filling periods using 4D datasets. Compared with controls, A-wave vorticity was significantly increased in RVDD subjects in both the RV [2343 (1,559-3,295) vs. 492 (267-2,649) 1/s, P = 0.028] and RA [30 (27-44) vs. 9 (5-27) 1/s, P = 0.005]. RA E vorticity was significantly decreased [13 (7-22) vs. 28 (15-31) 1/s, P = 0.038] in RVDD. E-wave vorticity correlated TV e', E-,and TV E/A (P < 0.05), and A-wave vorticity associated with both TV A and E/A (P < 0.02). RVDD is associated with alterations in E- and A-wave vorticity, and vorticity correlates with multiple echocardiographic markers of RVDD. Vorticity may be a robust noninvasive research tool for the investigation of RV fluid and tissue mechanical interactions in PAH.

Cystatin C: a potential biomarker for pulmonary arterial hypertension.

BACKGROUND AND OBJECTIVE: Cystatin C (CysC), a novel marker of renal function, predicts left heart failure and cardiovascular mortality. The hypothesis that serum CysC levels correlate with right ventricular (RV) morphology, function and pressure in pulmonary arterial hypertension (PAH) was tested. METHODS: As part of a prospective study, 14 PAH subjects and 10 matched controls underwent same-day echocardiography, cardiac magnetic resonance imaging (CMR), and phlebotomy for CysC, brain natriuretic peptide (BNP), and N-terminal BNP (NT-ProBNP). RV ejection fraction (RVEF), end-diastolic volume, end-systolic volume and mass were calculated using CMR. RV systolic pressure (RVSP), strain and diastolic function (including tricuspid valve (TV) E velocity, A velocity, e' velocity, E/A ratio and E/e' ratio) were assessed using echocardiography. RESULTS: RVSP was significantly elevated in PAH subjects versus controls (57 ± 17 vs. 28 ± 8 mm Hg, P < 0.0001). CysC was abnormally elevated in the PAH cohort when compared with controls (1.00 ± 0.23 vs 0.78 ± 0.05 mg/L, P = 0.001). CysC positively correlated with RVSP (rho 0.61, P = 0.002), RV end-diastolic volume (rho 0.50, P = 0.01), RV end-systolic volume (rho 0.58, P = 0.003), mass index (rho 0.66, P = 0.0004), strain (rho 0.51, P = 0.01) and strain rate (rho 0.51, P = 0.01) and negatively correlated with RVEF (rho -0.58, P = 0.003) and TV e' (rho -0.75, P < 0.0001). The same correlations with BNP and NT-ProBNP were comparable with CysC. CONCLUSIONS: In a small cohort, CysC accurately correlates with RV pressure, function and morphology. CysC may represent a novel PAH biomarker.

Computed tomographic measures of pulmonary vascular morphology in smokers and their clinical implications.

RATIONALE: Angiographic investigation suggests that pulmonary vascular remodeling in smokers is characterized by distal pruning of the blood vessels. OBJECTIVES: Using volumetric computed tomography scans of the chest we sought to quantitatively evaluate this process and assess its clinical associations. METHODS: Pulmonary vessels were automatically identified, segmented, and measured. Total blood vessel volume (TBV) and the aggregate vessel volume for vessels less than 5 mm(2) (BV5) were calculated for all lobes. The lobe-specific BV5 measures were normalized to the TBV of that lobe and the nonvascular tissue volume (BV5/T(issue)V) to calculate lobe-specific BV5/TBV and BV5/T(issue)V ratios. Densitometric measures of emphysema were obtained using a Hounsfield unit threshold of -950 (%LAA-950). Measures of chronic obstructive pulmonary disease severity included single breath measures of diffusing capacity of carbon monoxide, oxygen saturation, the 6-minute-walk distance, St George's Respiratory Questionnaire total score (SGRQ), and the body mass index, airflow obstruction, dyspnea, and exercise capacity (BODE) index. MEASUREMENTS AND MAIN RESULTS: The %LAA-950 was inversely related to all calculated vascular ratios. In multivariate models including age, sex, and %LAA-950, lobe-specific measurements of BV5/TBV were directly related to resting oxygen saturation and inversely associated with both the SGRQ and BODE scores. In similar multivariate adjustment lobe-specific BV5/T(issue)V ratios were inversely related to resting oxygen saturation, diffusing capacity of carbon monoxide, 6-minute-walk distance, and directly related to the SGRQ and BODE. CONCLUSIONS: Smoking-related chronic obstructive pulmonary disease is characterized by distal pruning of the small blood vessels (<5 mm(2)) and loss of tissue in excess of the vasculature. The magnitude of these changes predicts the clinical severity of disease.

Non-emphysematous chronic obstructive pulmonary disease is associated with diabetes mellitus.

BACKGROUND: Chronic obstructive pulmonary disease (COPD) has been classically divided into blue bloaters and pink puffers. The utility of these clinical subtypes is unclear. However, the broader distinction between airway-predominant and emphysema-predominant COPD may be clinically relevant. The objective was to define clinical features of emphysema-predominant and non-emphysematous COPD patients. METHODS: Current and former smokers from the Genetic Epidemiology of COPD Study (COPDGene) had chest computed tomography (CT) scans with quantitative image analysis. Emphysema-predominant COPD was defined by low attenuation area at -950 Hounsfield Units (LAA-950) ≥10%. Non-emphysematous COPD was defined by airflow obstruction with minimal to no emphysema (LAA-950 < 5%). RESULTS: Out of 4197 COPD subjects, 1687 were classified as emphysema-predominant and 1817 as non-emphysematous; 693 had LAA-950 between 5-10% and were not categorized. Subjects with emphysema-predominant COPD were older (65.6 vs 60.6 years, p < 0.0001) with more severe COPD based on airflow obstruction (FEV1 44.5 vs 68.4%, p < 0.0001), greater exercise limitation (6-minute walk distance 1138 vs 1331 ft, p < 0.0001) and reduced quality of life (St. George's Respiratory Questionnaire score 43 vs 31, p < 0.0001). Self-reported diabetes was more frequent in non-emphysematous COPD (OR 2.13, p < 0.001), which was also confirmed using a strict definition of diabetes based on medication use. The association between diabetes and non-emphysematous COPD was replicated in the ECLIPSE study. CONCLUSIONS: Non-emphysematous COPD, defined by airflow obstruction with a paucity of emphysema on chest CT scan, is associated with an increased risk of diabetes. COPD patients without emphysema may warrant closer monitoring for diabetes, hypertension, and hyperlipidemia and vice versa. TRIAL REGISTRATION: Clinicaltrials.gov identifiers: COPDGene NCT00608764, ECLIPSE NCT00292552.

LungViT: Ensembling Cascade of Texture Sensitive Hierarchical Vision Transformers for Cross-Volume Chest CT Image-to-Image Translation.

Chest computed tomography (CT) at inspiration is often complemented by an expiratory CT to identify peripheral airways disease. Additionally, co-registered inspiratory-expiratory volumes can be used to derive various markers of lung function. Expiratory CT scans, however, may not be acquired due to dose or scan time considerations or may be inadequate due to motion or insufficient exhale; leading to a missed opportunity to evaluate underlying small airways disease. Here, we propose LungViT- a generative adversarial learning approach using hierarchical vision transformers for translating inspiratory CT intensities to corresponding expiratory CT intensities. LungViT addresses several limitations of the traditional generative models including slicewise discontinuities, limited size of generated volumes, and their inability to model texture transfer at volumetric level. We propose a shifted-window hierarchical vision transformer architecture with squeeze-and-excitation decoder blocks for modeling dependencies between features. We also propose a multiview texture similarity distance metric for texture and style transfer in 3D. To incorporate global information into the training process and refine the output of our model, we use ensemble cascading. LungViT is able to generate large 3D volumes of size 320×320×320 . We train and validate our model using a diverse cohort of 1500 subjects with varying disease severity. To assess model generalizability beyond the development set biases, we evaluate our model on an out-of-distribution external validation set of 200 subjects. Clinical validation on internal and external testing sets shows that synthetic volumes could be reliably adopted for deriving clinical endpoints of chronic obstructive pulmonary disease.

Vessel and Airway Characteristics in One-Year Computed Tomography-defined Rapid Emphysema Progression: SPIROMICS.

Rationale: Rates of emphysema progression vary in chronic obstructive pulmonary disease (COPD), and the relationships with vascular and airway pathophysiology remain unclear. Objectives: We sought to determine if indices of peripheral (segmental and beyond) pulmonary arterial dilation measured on computed tomography (CT) are associated with a 1-year index of emphysema (EI; percentage of voxels <-950 Hounsfield units) progression. Methods: Five hundred ninety-nine former and never-smokers (Global Initiative for Chronic Obstructive Lung Disease stages 0-3) were evaluated from the SPIROMICS (Subpopulations and Intermediate Outcome Measures in COPD Study) cohort: rapid emphysema progressors (RPs; n = 188, 1-year ΔEI > 1%), nonprogressors (n = 301, 1-year ΔEI ± 0.5%), and never-smokers (n = 110). Segmental pulmonary arterial cross-sectional areas were standardized to associated airway luminal areas (segmental pulmonary artery-to-airway ratio [PAARseg]). Full-inspiratory CT scan-derived total (arteries and veins) pulmonary vascular volume (TPVV) was compared with small vessel volume (radius smaller than 0.75 mm). Ratios of airway to lung volume (an index of dysanapsis and COPD risk) were compared with ratios of TPVV to lung volume. Results: Compared with nonprogressors, RPs exhibited significantly larger PAARseg (0.73 ± 0.29 vs. 0.67 ± 0.23; P = 0.001), lower ratios of TPVV to lung volume (3.21 ± 0.42% vs. 3.48 ± 0.38%; P = 5.0 × 10-12), lower ratios of airway to lung volume (0.031 ± 0.003 vs. 0.034 ± 0.004; P = 6.1 × 10-13), and larger ratios of small vessel volume to TPVV (37.91 ± 4.26% vs. 35.53 ± 4.89%; P = 1.9 × 10-7). In adjusted analyses, an increment of 1 standard deviation in PAARseg was associated with a 98.4% higher rate of severe exacerbations (95% confidence interval, 29-206%; P = 0.002) and 79.3% higher odds of being in the RP group (95% confidence interval, 24-157%; P = 0.001). At 2-year follow-up, the CT-defined RP group demonstrated a significant decline in postbronchodilator percentage predicted forced expiratory volume in 1 second. Conclusions: Rapid one-year progression of emphysema was associated with indices indicative of higher peripheral pulmonary vascular resistance and a possible role played by pulmonary vascular-airway dysanapsis.

Chronic obstructive pulmonary disease: lobe-based visual assessment of volumetric CT by Using standard images--comparison with quantitative CT and pulmonary function test in the COPDGene study.

PURPOSE: To provide a new detailed visual assessment scheme of computed tomography (CT) for chronic obstructive pulmonary disease (COPD) by using standard reference images and to compare this visual assessment method with quantitative CT and several physiologic parameters. MATERIALS AND METHODS: This research was approved by the institutional review board of each institution. CT images of 200 participants in the COPDGene study were evaluated. Four thoracic radiologists performed independent, lobar analysis of volumetric CT images for type (centrilobular, panlobular, and mixed) and extent (on a six-point scale) of emphysema, the presence of bronchiectasis, airway wall thickening, and tracheal abnormalities. Standard images for each finding, generated by two radiologists, were used for reference. The extent of emphysema, airway wall thickening, and luminal area were quantified at the lobar level by using commercial software. Spearman rank test and simple and multiple regression analyses were performed to compare the results of visual assessment with physiologic and quantitative parameters. RESULTS: The type of emphysema, determined by four readers, showed good agreement (κ = 0.63). The extent of the emphysema in each lobe showed good agreement (mean weighted κ = 0.70) and correlated with findings at quantitative CT (r = 0.75), forced expiratory volume in 1 second (FEV(1)) (r = -0.68), FEV(1)/forced vital capacity (FVC) ratio (r = -0.74) (P < .001). Agreement for airway wall thickening was fair (mean κ = 0.41), and the number of lobes with thickened bronchial walls correlated with FEV(1) (r = -0.60) and FEV(1)/FVC ratio (r = -0.60) (P < .001). CONCLUSION: Visual assessment of emphysema and airways disease in individuals with COPD can provide reproducible, physiologically substantial information that may complement that provided by quantitative CT assessment.

Relationships between airflow obstruction and quantitative CT measurements of emphysema, air trapping, and airways in subjects with and without chronic obstructive pulmonary disease.

OBJECTIVE: This study evaluates the relationships between quantitative CT (QCT) and spirometric measurements of disease severity in cigarette smokers with and without chronic obstructive pulmonary disease (COPD). MATERIALS AND METHODS: Inspiratory and expiratory CT scans of 4062 subjects in the Genetic Epidemiology of COPD (COPDGene) Study were evaluated. Measures examined included emphysema, defined as the percentage of low-attenuation areas≤-950 HU on inspiratory CT, which we refer to as "LAA-950I"; air trapping, defined as the percentage of low-attenuation areas≤-856 HU on expiratory CT, which we refer to as "LAA-856E"; and the inner diameter, inner and outer areas, wall area, airway wall thickness, and square root of the wall area of a hypothetical airway of 10-mm internal perimeter of segmental and subsegmental airways. Correlations were determined between spirometry and several QCT measures using statistics software (SAS, version 9.2). RESULTS: QCT measurements of low-attenuation areas correlate strongly and significantly (p<0.0001) with spirometry. The correlation between LAA-856E and forced expiratory volume in 1 second (FEV1) and the ratio of FEV1 to forced vital capacity (FVC) (r=-0.77 and -0.84, respectively) is stronger than the correlation between LAA-950I and FEV1 and FEV1/FVC (r=-0.67 and r=-0.76). Inspiratory and expiratory volume changes decreased with increasing disease severity, as measured by the Global Initiative for Chronic Obstructive Pulmonary Disease (GOLD) staging system (p<0.0001). When airway variables were included with low-attenuation area measures in a multiple regression model, the model accounted for a statistically greater proportion of variation in FEV1 and FEV1/FVC (R2=0.72 and 0.77, respectively). Airway measurements alone are less correlated with spirometric measures of FEV1 (r=0.15 to -0.44) and FEV1/FVC (r=0.19 to -0.34). CONCLUSION: QCT measurements are strongly associated with spirometric results showing impairment in smokers. LAA-856E strongly correlates with physiologic measurements of airway obstruction. Airway measurements can be used concurrently with QCT measures of low-attenuation areas to accurately predict lung function.

Localization supervision of chest x-ray classifiers using label-specific eye-tracking annotation.

Convolutional neural networks (CNNs) have been successfully applied to chest x-ray (CXR) images. Moreover, annotated bounding boxes have been shown to improve the interpretability of a CNN in terms of localizing abnormalities. However, only a few relatively small CXR datasets containing bounding boxes are available, and collecting them is very costly. Opportunely, eye-tracking (ET) data can be collected during the clinical workflow of a radiologist. We use ET data recorded from radiologists while dictating CXR reports to train CNNs. We extract snippets from the ET data by associating them with the dictation of keywords and use them to supervise the localization of specific abnormalities. We show that this method can improve a model's interpretability without impacting its image-level classification.

Response to high dose ipilimumab plus temozolomide after progression on standard or low dose ipilimumab in advanced melanoma: a retrospective analysis.

Background: Despite advancements in checkpoint inhibitor-based immunotherapy, patients with advanced melanoma who have progressed on standard dose ipilimumab (Ipi) + nivolumab continue to have poor prognosis. Several studies support a dose-response activity of Ipi, and one promising combination is Ipi 10mg/kg (Ipi10) + temozolomide (TMZ). Methods: We performed a retrospective cohort analysis of patients with advanced melanoma treated with Ipi10+TMZ in the immunotherapy refractory/resistant setting (n = 6), using similar patients treated with Ipi3+TMZ (n = 6) as comparison. Molecular profiling by whole exome sequencing (WES) and RNA-seq of tumors harvested through one responder's treatment was performed. Results: With a median follow up of 119 days, patients treated with Ipi10+TMZ had statistically significant longer median progression free survival of 144.5 days (range 27-219) vs 44 (26-75) in Ipi3+TMZ, p=0.04, and a trend for longer median overall survival of 154.5 days (27-537) vs 89.5 (26-548). All patients in the Ipi10 cohort had progressed on prior Ipi+Nivo. WES revealed only 12 shared somatic mutations including BRAF V600E. RNA-seq showed enrichment of inflammatory signatures, including interferon responses in metastatic lesions after standard dose Ipi + nivo and Ipi10 + TMZ compared to the primary tumor, and downregulated negative immune regulators including Wnt and TGFb signaling. Conclusion: Ipi10+TMZ demonstrated efficacy including dramatic responses in patients with advanced melanoma refractory to prior Ipi + anti-PD1, even with CNS metastases. Molecular data suggest a potential threshold of Ipi dose for activation of sufficient anti-tumor immune response, and higher dose Ipi is required for some patients.

Predicting severe chronic obstructive pulmonary disease exacerbations using quantitative CT: a retrospective model development and external validation study.

BACKGROUND: Quantitative CT is becoming increasingly common for the characterisation of lung disease; however, its added potential as a clinical tool for predicting severe exacerbations remains understudied. We aimed to develop and validate quantitative CT-based models for predicting severe chronic obstructive pulmonary disease (COPD) exacerbations. METHODS: We analysed the Subpopulations and Intermediate Outcome Measures In COPD Study (SPIROMICS) cohort, a multicentre study done at 12 clinical sites across the USA, of individuals aged 40-80 years from four strata: individuals who never smoked, individuals who smoked but had normal spirometry, individuals who smoked and had mild to moderate COPD, and individuals who smoked and had severe COPD. We used 3-year follow-up data to develop logistic regression classifiers for predicting severe exacerbations. Predictors included age, sex, race, BMI, pulmonary function, exacerbation history, smoking status, respiratory quality of life, and CT-based measures of density gradient texture and airway structure. We externally validated our models in a subset from the Genetic Epidemiology of COPD (COPDGene) cohort. Discriminative model performance was assessed using the area under the receiver operating characteristic curve (AUC), which was also compared with other predictors, including exacerbation history and the BMI, airflow obstruction, dyspnoea, and exercise capacity (BODE) index. We evaluated model calibration using calibration plots and Brier scores. FINDINGS: Participants in SPIROMICS were enrolled between Nov 12, 2010, and July 31, 2015. Participants in COPDGene were enrolled between Jan 10, 2008, and April 15, 2011. We included 1956 participants from the SPIROMICS cohort who had complete 3-year follow-up data: the mean age of the cohort was 63·1 years (SD 9·2) and 1017 (52%) were men and 939 (48%) were women. Among the 1956 participants, 434 (22%) had a history of at least one severe exacerbation. For the CT-based models, the AUC was 0·854 (95% CI 0·852-0·855) for at least one severe exacerbation within 3 years and 0·931 (0·930-0·933) for consistent exacerbations (defined as ≥1 acute episode in each of the 3 years). Models were well calibrated with low Brier scores (0·121 for at least one severe exacerbation; 0·039 for consistent exacerbations). For the prediction of at least one severe event during 3-year follow-up, AUCs were significantly higher with CT biomarkers (0·854 [0·852-0·855]) than exacerbation history (0·823 [0·822-0·825]) and BODE index 0·812 [0·811-0·814]). 6965 participants were included in the external validation cohort, with a mean age of 60·5 years (SD 8·9). In this cohort, AUC for at least one severe exacerbation was 0·768 (0·767-0·769; Brier score 0·088). INTERPRETATION: CT-based prediction models can be used for identification of patients with COPD who are at high risk of severe exacerbations. The newly identified CT biomarkers could potentially enable investigation into underlying disease mechanisms responsible for exacerbations. FUNDING: National Institutes of Health and the National Heart, Lung, and Blood Institute.

A combined pulmonary-radiology workshop for visual evaluation of COPD: study design, chest CT findings and concordance with quantitative evaluation.

UNLABELLED: The purposes of this study were: to describe chest CT findings in normal non-smoking controls and cigarette smokers with and without COPD; to compare the prevalence of CT abnormalities with severity of COPD; and to evaluate concordance between visual and quantitative chest CT (QCT) scoring. METHODS: Volumetric inspiratory and expiratory CT scans of 294 subjects, including normal non-smokers, smokers without COPD, and smokers with GOLD Stage I-IV COPD, were scored at a multi-reader workshop using a standardized worksheet. There were 58 observers (33 pulmonologists, 25 radiologists); each scan was scored by 9-11 observers. Interobserver agreement was calculated using kappa statistic. Median score of visual observations was compared with QCT measurements. RESULTS: Interobserver agreement was moderate for the presence or absence of emphysema and for the presence of panlobular emphysema; fair for the presence of centrilobular, paraseptal, and bullous emphysema subtypes and for the presence of bronchial wall thickening; and poor for gas trapping, centrilobular nodularity, mosaic attenuation, and bronchial dilation. Agreement was similar for radiologists and pulmonologists. The prevalence on CT readings of most abnormalities (e.g. emphysema, bronchial wall thickening, mosaic attenuation, expiratory gas trapping) increased significantly with greater COPD severity, while the prevalence of centrilobular nodularity decreased. Concordances between visual scoring and quantitative scoring of emphysema, gas trapping and airway wall thickening were 75%, 87% and 65%, respectively. CONCLUSIONS: Despite substantial inter-observer variation, visual assessment of chest CT scans in cigarette smokers provides information regarding lung disease severity; visual scoring may be complementary to quantitative evaluation.

PLOSL: Population learning followed by one shot learning pulmonary image registration using tissue volume preserving and vesselness constraints.

This paper presents the Population Learning followed by One Shot Learning (PLOSL) pulmonary image registration method. PLOSL is a fast unsupervised learning-based framework for 3D-CT pulmonary image registration algorithm based on combining population learning (PL) and one-shot learning (OSL). The PLOSL image registration has the advantages of the PL and OSL approaches while reducing their respective drawbacks. The advantages of PLOSL include improved performance over PL, substantially reducing OSL training time and reducing the likelihood of OSL getting stuck in local minima. PLOSL pulmonary image registration uses tissue volume preserving and vesselness constraints for registration of inspiration-to-expiration and expiration-to-inspiration pulmonary CT images. A coarse-to-fine convolution encoder-decoder CNN architecture is used to register large and small shape features. During training, the sum of squared tissue volume difference (SSTVD) compensates for intensity differences between inspiration and expiration computed tomography (CT) images and the sum of squared vesselness measure difference (SSVMD) helps match the lung vessel tree. Results show that the PLOSL (SSTVD+SSVMD) algorithm achieved subvoxel landmark error while preserving pulmonary topology on the SPIROMICS data set, the public DIR-LAB COPDGene and 4DCT data sets.