Predicting the initiation for home oxygen therapy after lung resection using low-attenuation lung volume on computed tomography

Introduction

Anatomical lung resection has long been established as the standard treatment for lung cancer. However, with the increasing age of patients with lung cancer (1), the management of postoperative complications has become increasingly important. One such complication, particularly in patients with reduced pulmonary function, is postoperative hypoxemia, which may necessitate home oxygen therapy (HOT).

HOT is indicated for various chronic respiratory diseases. According to the American Thoracic Society guidelines, long-term continuous oxygen therapy administered for at least 15 hours per day is recommended for adults with chronic obstructive pulmonary disease (COPD) or interstitial pneumonia (IP) who exhibit severe chronic resting hypoxemia while breathing room air (2). However, while HOT provides a survival benefit for patients with severe resting hypoxemia, recent evidence suggests it may not improve survival in those with moderate hypoxemia (3).

Patients with chronic lung diseases, such as COPD or IP, are at increased risk for developing primary lung cancer in comparison to the general population (4-6). Among patients with lung cancer, those with underlying chronic lung diseases are more likely to experience postoperative pulmonary dysfunction after curative resection and subsequently require postoperative HOT. Previous studies have reported that approximately 15% of lung cancer surgery patients require HOT postoperatively, and that their quality of life (QOL) is negatively impacted (7). Identifying predictive factors for postoperative HOT in lung cancer patients undergoing surgery may be valuable in supporting clinical decision-making and obtaining informed consent, especially for high-risk patients.

Low attenuation volume (LAV), defined as the volume of lung parenchyma with attenuation values below −950 Hounsfield Units on high-resolution computed tomography (HRCT), is a well-recognized marker of emphysematous changes (8). Although LAV has been extensively studied in the context of COPD, its association with outcomes after lung resection remains poorly understood. Moreover, to the best of our knowledge, no studies have investigated the relationship between LAV and the need for postoperative HOT.

This study aimed to clarify the relationship between LAV and the need for HOT after lung resection using three-dimensional image analysis of preoperative CT scans. Furthermore, we attempted to develop an exploratory prognostic model for the need for HOT following lung resection. We present this article in accordance with the TRIPOD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1032/rc).

Methods

Patients and study design

The patient flow chart was shown in Figure 1. This retrospective study included 259 patients who underwent lobectomy or segmentectomy at the Department of General Thoracic Surgery, Chiba University Hospital between January 2019 and December 2020. Patients presenting with tumor-induced atelectasis in non-tumor-bearing lobes were excluded. Clinical data, including age, sex, smoking index, body mass index (BMI), comorbidities, preoperative pulmonary function test results, surgical details, and postoperative complications, were retrospectively collected from medical records. All patients received at least one week of postoperative respiratory rehabilitation. Following rehabilitation, patients with a percutaneous oxygen saturation <90% at rest or during exertion were considered for HOT at the time of discharge. Postoperative complications were recorded if they were classified as grade ≥3 according to the Clavien-Dindo classification.

Figure 1 Flow diagram of the patient selection process. Between January 2019 and December 2020, a total of 461 patients underwent anatomical lung resection at our institution. Of these, 202 patients were excluded based on the following criteria: undergoing bilobectomy, pneumonectomy, or anatomical resection combined with wedge resection (n=43); having a non-contrast preoperative CT scan (n=157); and an inability to measure the LAV from the CT images (n=2). After these exclusions, a final cohort of 259 patients was included in the analysis. CT, computed tomography; LAV, low attenuation volume.

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of Chiba University Hospital (No. HK202509-06) and individual consent for this retrospective analysis was waived.

LAV calculation

This study obtained data concerning patients with contrast-enhanced computed tomography (CT) scans with a slice thickness of 1mm acquired at Chiba University Hospital, using Aquilion PRIME (Canon Medical Systems, Tochigi, Japan). The entire chest was scanned during inspiratory breath-hold with 0.5×80 mm detector collimation and a pitch factor of 0.813 using automatic exposure control at 120 kVp. REVORAS^® (Ziosoft, Tokyo, Japan), a novel volume-rendering three-dimensional (3D) image reconstruction software, was used for 3D image reconstruction from the CT images. After excluding the soft tissues surrounding the lungs, as well as the large vessels and interstitial tissues within the lungs, morphological parameters were calculated, including lung volume, LAV, and the percentage of LAV-to-lung volume (LAV%). LAV was defined as regions showing <−950 HU on the CT. The LAV% was calculated using the following formula:

LAV%=LAVLungvolume×100[1]

Statistical analyses

Continuous variables were compared using the Mann-Whitney U test and categorical variables were analyzed using Fisher’s exact test. Univariable and multivariable logistic regression analyses were performed to identify independent predictive factors for the initiation of postoperative HOT. Multicollinearity between the variables was evaluated using Pearson’s correlation test. Furthermore, receiver operating characteristic (ROC) curves were generated to determine the optimal cutoff values and to calculate the area under the curve (AUC) for assessing model performance. Statistical significance was set at P<0.05. All statistical analyses were performed using R version 4.4.2 (The R Foundation for Statistical Computing, Vienna, Austria) and RStudio (version 2025.05.1+513; Posit Software, PBC).

Results

Patient characteristics

Among the 259 enrolled patients, 14 (5.4%) were classified into the HOT group. The preoperative characteristics of the HOT and non-HOT groups are shown in Table 1. The median age of the HOT group was 74.5 years (range, 56–83 years) and that of the non-HOT group was 71.0 years (range, 27–88 years). Thirteen of 14 patients in the HOT group were male.

Significant differences were observed between the HOT and non-HOT groups in the following variables: sex, pack-years, preoperative percent predicted forced expiratory volume in 1 second (%FEV_1.0), preoperative PaO₂, presence of COPD, preoperative LAV% (Figure 2), operation time, estimated blood loss and postoperative pleurodesis.

Figure 2 Comparison of preoperative LAV% between the HOT and non-HOT groups. The box plot illustrates the distribution of LAV% in patients who required postoperative HOT and those who did not. The median LAV% was 28.8% in the HOT group and 19.9% in the non-HOT group. HOT, home oxygen therapy; LAV%, the percentage of low attenuation volume-to-lung volume.

There were no differences in age, BMI, surgical procedure, preoperative percent predicted forced vital capacity (%FVC), number of resected subsegments, postoperative pneumonia, postoperative acute exacerbation of IP, and postoperative cardiac complications between the groups.

Univariate analysis

Only preoperative variables were included in the predictive factor analysis. The results of the univariate analysis are presented in Table 2. The following variables were associated with the need for HOT: sex [odds ratio (OR) 8.38, 95% confidence interval (CI): 1.63–153, P=0.042], smoking index (OR 1.02, 95% CI: 1.01–1.03, P=0.003), LAV% (OR 1.22, 95% CI: 1.12–1.36, P<0.001), preoperative %FEV_1.0 (OR 0.94, 95% CI: 0.91–0.98, P<0.001), and presence of COPD (OR 5.45, 95% CI: 1.76–20.4, P=0.005). Age, BMI, and preoperative PaO₂ were not associated with the need for HOT in the univariate logistic regression analysis, despite PaO₂ showing a significant difference in the initial group comparison.

Multivariate analysis

Prior to the multivariate analysis, we evaluated the multicollinearity between the predictor variables by Pearson’s correlation test (Figure 3A,3B). First, we assessed the relationship between the smoking index (pack-years) and pulmonary parameters. The smoking index showed a significant positive correlation with LAV% (r=0.187, P=0.002) and a significant moderate negative correlation with %FEV_1.0 (r=−0.441, P<0.001). Because smoking history was significantly correlated with variables that more directly and objectively reflect the current structural and functional lung impairment, it was intentionally excluded from the logistic regression analyses to prevent potential multicollinearity. Next, we evaluated the multicollinearity between the LAV% and %FEV_1.0 (Figure 3C). The correlation coefficient between LAV% and %FEV_1.0 was weak (r=−0.099, P=0.11). Furthermore, the variance inflation factor for both variables was 1.11, which is well below the commonly accepted threshold of 5. This result indicated negligible multicollinearity and confirmed LAV% and %FEV_1.0 were independent predictors for requiring HOT. The results of the multivariate logistic regression analysis are presented in Table 3. Although sex, LAV%, %FEV_1.0, and presence of COPD were identified as predictive factors in the univariate analysis, the number of explanatory variables entered into the multivariate analysis was limited to two to avoid overfitting, considering the small number of the HOT group. Therefore, LAV% and %FEV_1.0, which demonstrated the most significant P values (smallest P values), were selected for the multivariate logistic regression model. The analysis identified preoperative LAV% (OR 1.17, 95% CI: 1.07–1.31, P=0.001) and preoperative %FEV_1.0 (OR 0.96, 95% CI: 0.93–1.00, P=0.04) as independent predictive factors for requiring HOT after lung resection.

Figure 3 Scatter plot illustrating the correlation among smoking, preoperative LAV% and %FEV_1.0. (A) The smoking index (pack-years) showed a significant positive correlation with LAV% (r=0.187, P=0.002). (B) The smoking index showed a significant negative correlation with %FEV_1.0 (r=−0.441, P<0.001). Because of these significant correlations, the smoking index was excluded from the multivariate analysis. (C) The correlation between LAV% and %FEV_1.0 was weak (r=−0.099, P=0.11), indicating negligible multicollinearity prior to the multivariate analysis. Red circles represent the non-HOT group, and blue circles represent the HOT group. %FEV_1.0, percent predicted forced expiratory volume in 1 second; HOT, home oxygen therapy; LAV%, the percentage of low attenuation volume-to-lung volume.

ROC curve analysis

ROC curve analyses were performed to evaluate the predictive capability of preoperative variables for HOT. The analysis using LAV% identified an optimal cutoff value of 24.6%, yielding a sensitivity of 0.857, a specificity of 0.767, and an AUC of 0.808 (Figure 4A). For %FEV_1.0, the optimal cutoff value was 80.7%, with a sensitivity of 0.643, a specificity of 0.763, and an AUC of 0.741 (Figure 4B). Furthermore, the combined model incorporating both LAV% and %FEV_1.0 demonstrated the highest predictive performance, with a sensitivity of 0.786, a specificity of 0.829, and an AUC of 0.822 (Figure 4C). The probability of requiring HOT can be calculated using the following formula:

Linearpredictor(LP)=−3.5107+0.1581×LAV%−0.0362×%FEV1.0[2]

Probability=1(1+e−LP)[3]

Figure 4 ROC curves for predicting the initiation of HOT. (A) ROC curve for preoperative LAV%, demonstrating an AUC of 0.808. (B) ROC curve for preoperative %FEV_1.0, with an AUC of 0.741. (C) ROC curve for the combined model incorporating both LAV% and %FEV_1.0, achieving the acceptable predictive performance with an AUC of 0.822. %FEV_1.0, percent predicted forced expiratory volume in 1 second; AUC, area under the curve; HOT, home oxygen therapy; LAV%, the percentage of low attenuation volume-to-lung volume; ROC, receiver operating characteristic.

According to the Youden index, the optimal cutoff value of the predicted probability for requiring postoperative HOT was calculated to be 6.5%.

Development of a predictive nomogram

To facilitate the clinical application of our findings, a predictive nomogram was developed based on the multivariate logistic regression model (Figure 5). The nomogram visually incorporates the two independent predictive factors: preoperative LAV% and %FEV_1.0. Clinicians can estimate a patient’s individualized probability of requiring postoperative HOT by assigning points to the value of each variable on the top Points scale, summing these points, and projecting the Total Points vertically down onto the bottom probability of HOT scale.

Figure 5 Predictive nomogram for estimating the probability of requiring HOT following lung resection. The model incorporates two independents preoperative predictive factors: LAV% and %FEV_1.0. To use the nomogram, locate the patient’s value on each variable axis, draw a vertical line upward to the Points scale to determine the score for each variable, sum the scores, locate this sum on the Total points scale, and draw a vertical line downward to determine the patient’s individualized the probability of HOT. %FEV_1.0, percent predicted forced expiratory volume in 1 second; HOT, home oxygen therapy; LAV%, the percentage of low attenuation volume-to-lung volume.

Discussion

This study assessed the relationship between LAV and the need for HOT after lung resection. The multivariate analysis identified preoperative LAV% and %FEV_1.0 as independent predictive factors for the need of HOT. In the individual ROC curve analysis, the optimal cutoff values were 24.6% for LAV% (sensitivity of 0.857, specificity of 0.767) and 80.7% for %FEV_1.0 (sensitivity of 0.643, specificity of 0.763). Furthermore, the combined model incorporating both LAV% and %FEV_1.0 demonstrated a potential predictive performance with an AUC of 0.822, a sensitivity of 0.786, and a specificity of 0.829. These findings suggest that the quantitative assessment of emphysematous changes via LAV%, when used in conjunction with conventional pulmonary function tests, provides a robust clinical indicator for predicting need for HOT. Our developed predictive model could provide a clear clinical threshold: patients with a calculated probability of 6.5% or greater could be managed as a high-risk group for postoperative HOT.

A systematic review of 17 studies by Oslock et al. reported that female sex, Caucasian race, obesity, and non-adenocarcinoma histology were most strongly associated with an increased risk of HOT, while smoking history and pulmonary function parameters, such as FEV_1.0 and diffusing capacity for carbon monoxide (DLCO), were not identified as significant factors in their multivariate analysis (9). However, that review did not account for surgery-related variables or incorporate imaging findings, and only 4 of the 17 studies performed multivariate analysis. Previous work by Yamanaka et al. identified pulmonary comorbidities and postoperative pulmonary complications as risk factors for HOT (10). Because our study focused exclusively on identifying predictive factors using preoperative parameters, our findings can be integrated with these previous reports to better identify patients at a higher risk of requiring postoperative HOT. Sekihara et al. further reported that, in lung cancer patients with IP, those requiring postoperative HOT had significantly lower 3-year overall survival than those who did not (11). Given that postoperative HOT may not only reduce QOL, but also impact the prognosis, careful preoperative risk assessment for HOT is critically important.

The incidence of postoperative HOT initiation in the present study was 5.4%, which is lower than the ~15% rate reported in the previous study (7). We attribute this discrepancy primarily to differences in patient demographics and perioperative management. First, our cohort had a lower baseline respiratory risk; the proportion of patients with pulmonary comorbidities was approximately 33.6% in our study, compared to 52.2% in the previous report. Consequently, our rate of postoperative pulmonary complications was also lower (approximately 17% vs. 31.6%) (7). This relatively healthier patient background likely reduced the overall need for HOT. Second, while the prior study defined strict SpO₂ criteria (≤88%, or ≤89% with pulmonary neoplasm) for HOT eligibility without detailing postoperative rehabilitation protocols (7), our institution emphasizes proactive perioperative interventions. Although the specific effect of this strategy was not assessed in the present study, it may have facilitated postoperative respiratory recovery and contributed to the lower HOT incidence observed in our cohort.

Yasuura et al. found that in lobectomy patients with COPD, age ≥70 years and preoperative LAV% ≥10% predicted postoperative cardiopulmonary complications (12). Although COPD itself was not identified as an independent predictive factor in our multivariate analysis, the identification of LAV% as a predictor of the initiation of HOT was consistent with their observations. Similarly, Ozeki et al. reported that patients with postoperative LAV% ≥0.1% and resected lung volume ≥20% of total lung volume experienced reduced exercise tolerance at 6 months postoperatively (13). Together with our results, these findings support the use of LAV as a valuable imaging biomarker for postoperative risk assessment in lung resection.

Interestingly, we observed only a weak negative correlation between LAV% and %FEV_1.0 (r=−0.099) in our study, despite their well-known biological association. We considered three main reasons for this discrepancy. First, due to surgical selection bias, patients with severely low %FEV_1.0 were generally deemed ineligible for anatomical resection, leading to a restricted range of %FEV_1.0 values that statistically attenuates the correlation. Second, COPD is a heterogeneous disease comprising distinct phenotypes, such as emphysema-dominant and non-emphysematous types, which can cause a dissociation between structural destruction and functional impairment (14). Third, the preoperative use of inhaled bronchodilators can significantly improve and maintain %FEV_1.0 without altering the irreversible structural changes quantified by LAV%. Thus, our cohort likely included a subset of patients whose airflow obstruction was medically well-controlled despite a high emphysema burden.

The LAV can be easily and quantitatively measured from preoperative CT scans, making it a practical addition to risk assessment models. Regarding the intended clinical decision pathway, this exploratory model is designed to be used before surgery. By calculating the risk of postoperative HOT initiation using the proposed nomogram, clinicians can identify high-risk patients early. For these patients, proactive interventions and tailored strategies can be planned. These include opting for less invasive surgical approaches, intensive preoperative respiratory rehabilitation, careful intraoperative lung-protective management, and proactively preparing for the possible initiation of HOT at discharge to reduce complications. However, further external validation is required before this model can be widely implemented in routine clinical practice.

This study had several limitations. First, owing to its retrospective single center design and the sample size was limited, it is difficult to eliminate the influence of selection bias and confounding factors, and caution is required when generalizing the results. For example, detailed data regarding the preoperative use of inhaled bronchodilators were unavailable, limiting our ability to adjust for their potential modifying effects on baseline pulmonary function. Additionally, because the number of events per variable was approximately 7 and neither internal validation nor external validation was performed in the present study, there is a potential risk of overfitting, which further underscores the need to interpret our model’s performance cautiously as an exploratory finding. Second, we derived the LAV% from contrast-enhanced CT scans using the standard −950 HU threshold, whereas LAV% was originally developed and validated using non-contrast inspiratory HRCT. Intravenous contrast media increase the attenuation of the pulmonary vasculature and parenchyma, which may result in a slight underestimation of the actual LAV% when applying the strict −950 HU threshold (15,16). Furthermore, since all patients in our cohort were evaluated under similar contrast-enhanced protocols, and because our objective was risk stratification rather than absolute quantification of emphysema burden, the relative ranking of patients according to LAV% was expected to remain largely preserved. However, the use of contrast-enhanced CT may have influenced the measured LAV% values, and future studies should evaluate the robustness of our findings using non-contrast HRCT or alternative attenuation thresholds. Third, DLCO was not included in the present analysis despite being a well-recognized predictor of postoperative pulmonary complications and hypoxemia after lung resection. In our institutional clinical practice, diffusion capacity is not measured routinely for all surgical candidates, but rather selectively performed for patients with strongly suspected IP or severe COPD. Therefore, a large amount of DLCO data was unavailable in this retrospective cohort. Furthermore, because DLCO testing was performed based on clinical indications, the missing data were unlikely to be random, limiting the feasibility of incorporating DLCO into the multivariable analysis. Future prospective studies with systematic DLCO assessment are needed to determine the incremental value of DLCO beyond CT-based emphysema quantification. Finally, in patients who started HOT postoperatively, the timing, duration, and follow-up protocols were not standardized, leading to incomplete data on long-term HOT management. Furthermore, the clinical definition of exertion during the postoperative desaturation assessment was not strictly standardized. In our routine real-world practice, exertion was not evaluated using a formal protocol, such as the 6-minute walk test. Instead, it was based on routine ward ambulation and rehabilitation activities, the intensity and duration of which were tailored to individual patients and left to the discretion of the attending physical therapists. This lack of standardization in exertion assessment might affect the reproducibility of our results, and future prospective studies using standardized exercise protocols are warranted.

Given these limitations, it is necessary to validate the findings through multicenter prospective studies. Integrating LAV% with other imaging metrics, clinical variables, and laboratory data may enable the development of more accurate predictive models for the initiation of HOT after lung resection.

Conclusions

This study demonstrated that preoperative LAV% is an independent predictor of the initiation of HOT following lung resection. Incorporating preoperative CT image analysis into surgical risk assessments may help identify patients at high risk of requiring postoperative HOT, thereby guiding optimal perioperative management strategies.

Acknowledgments

The abstract of this paper was presented at the 42nd Annual Meeting of the Japanese Association for Chest Surgery in Tokyo, Japan, May 2025.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1032/rc

Data Sharing Statement: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1032/dss

Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1032/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1032/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of Chiba University Hospital (No. HK202509-06) and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Committee for Scientific Affairs, The Japanese Association for Thoracic Surgery. Thoracic and cardiovascular surgeries in Japan during 2023: Annual report by the Japanese Association for Thoracic Surgery. Gen Thorac Cardiovasc Surg 2025;73:526-65.
2. Jacobs SS, Krishnan JA, Lederer DJ, et al. Home Oxygen Therapy for Adults with Chronic Lung Disease. An Official American Thoracic Society Clinical Practice Guideline. Am J Respir Crit Care Med 2020;202:e121-41.
3. Long term domiciliary oxygen therapy in chronic hypoxic cor pulmonale complicating chronic bronchitis and emphysema. Report of the Medical Research Council Working Party. Lancet 1981;1:681-6.
4. Mizuno S, Takiguchi Y, Fujikawa A, et al. Chronic obstructive pulmonary disease and interstitial lung disease in patients with lung cancer. Respirology 2009;14:377-83. [Crossref] [PubMed]
5. Choi WI, Park SH, Park BJ, et al. Interstitial Lung Disease and Lung Cancer Development: A 5-Year Nationwide Population-Based Study. Cancer Res Treat 2018;50:374-81. [Crossref] [PubMed]
6. Park HY, Kang D, Shin SH, et al. Chronic obstructive pulmonary disease and lung cancer incidence in never smokers: a cohort study. Thorax 2020;75:506-9. [Crossref] [PubMed]
7. Nicastri DG, Alpert N, Liu B, et al. Oxygen Use After Lung Cancer Surgery. Ann Thorac Surg 2018;106:1548-55. [Crossref] [PubMed]
8. Kitazawa S, Wijesinghe AI, Maki N, et al. Predicting Respiratory Complications Following Lobectomy Using Quantitative CT Measures of Emphysema. Int J Chron Obstruct Pulmon Dis 2021;16:2523-31. [Crossref] [PubMed]
9. Oslock WM, Xie R, Donahue J, et al. Lung resection and new postoperative home oxygen requirement: a systematic review. J Thorac Dis 2024;16:8796-806. [Crossref] [PubMed]
10. Yamanaka T, Sakairi Y, Sata Y, et al. What are the risk factors for postoperative home oxygen therapy in patients with lung cancer? J Thorac Dis 2024;16:989-96. [Crossref] [PubMed]
11. Sekihara K, Aokage K, Hiyama T, et al. Prognostic impact of home oxygen therapy on patients with resected non-small-cell lung cancer with interstitial lung disease. Surg Today 2021;51:1036-43. [Crossref] [PubMed]
12. Yasuura Y, Maniwa T, Mori K, et al. Quantitative computed tomography for predicting cardiopulmonary complications after lobectomy for lung cancer in patients with chronic obstructive pulmonary disease. Gen Thorac Cardiovasc Surg 2019;67:697-703. [Crossref] [PubMed]
13. Ozeki N, Iwano S, Nakamura S, et al. Chest three-dimensional-computed tomography imaging data analysis for the variation of exercise capacity after lung lobectomy. Clin Physiol Funct Imaging 2022;42:362-71. [Crossref] [PubMed]
14. Shima H, Tanabe N, Oguma A, et al. Subtyping emphysematous COPD by respiratory volume change distributions on CT. Thorax 2023;78:344-53. [Crossref] [PubMed]
15. Heussel CP, Kappes J, Hantusch R, et al. Contrast enhanced CT-scans are not comparable to non-enhanced scans in emphysema quantification. Eur J Radiol 2010;74:473-8. [Crossref] [PubMed]
16. Schachner C, Strotzer QD, Scheuermeyer L, et al. Regional differences of lung parenchyma attenuation after contrast administration in quantitative dual energy chest CT. BMC Pulm Med 2026;26:180. [Crossref] [PubMed]