Lung Screening Research Articles

Leveraging Serial Low-Dose CT Scans in Radiomics-based Reinforcement Learning to Improve Early Diagnosis of Lung Cancer at Baseline Screening.

Purpose To evaluate the feasibility of leveraging serial low-dose CT (LDCT) scans to develop a radiomics-based reinforcement learning (RRL) model for improving early diagnosis of lung cancer at baseline screening. Materials and Methods In this retrospective study, 1951 participants (female patients, 822; median age, 61 years [range, 55-74 years]) (male patients, 1129; median age, 62 years [range, 55-74 years]) were randomly selected from the National Lung Screening Trial between August 2002 and April 2004. An RRL model using serial LDCT scans (S-RRL) was trained and validated using data from 1404 participants (372 with lung cancer) containing 2525 available serial LDCT scans up to 3 years. A baseline RRL (B-RRL) model was trained with only LDCT scans acquired at baseline screening for comparison. The 547 held-out individuals (150 with lung cancer) were used as an independent test set for performance evaluation. The area under the receiver operating characteristic curve (AUC) and the net reclassification index (NRI) were used to assess the performances of the models in the classification of screen-detected nodules. Results Deployment to the held-out baseline scans showed that the S-RRL model achieved a significantly higher test AUC (0.88 [95% CI: 0.85, 0.91]) than both the Brock model (AUC, 0.84 [95% CI: 0.81, 0.88]; P = .02) and the B-RRL model (AUC, 0.86 [95% CI: 0.83, 0.90]; P = .02). Lung cancer risk stratification was significantly improved by the S-RRL model as compared with Lung CT Screening Reporting and Data System (NRI, 0.29; P < .001) and the Brock model (NRI, 0.12; P = .008). Conclusion The S-RRL model demonstrated the potential to improve early diagnosis and risk stratification for lung cancer at baseline screening as compared with the B-RRL model and clinical models. Keywords: Radiomics-based Reinforcement Learning, Lung Cancer Screening, Low-Dose CT, Machine Learning © RSNA, 2024 Supplemental material is available for this article.

Pulmonary adenocarcinoma of low malignant potential defines indolent NSCLC associated with overdiagnosis in the national lung screening trial.

The national lung screening trial (NLST) demonstrated a reduction in lung cancer mortality with lowdose CT (LDCT) compared to chest x-ray (CXR) screening. Overdiagnosis was high (79%) among bronchoalveolar carcinoma (BAC) currently replaced by adenocarcinoma in situ (AIS), minimally invasive adenocarcinoma (MIA) and adenocarcinoma of low malignant potential (LMP) exhibiting 100% disease specific survival (DSS). Compare the outcomes and proportions of BAC, AIS, MIA, and LMP among NLST screendetected stage IA NSCLC with overdiagnosis rate. Whole slide images were reviewed by a thoracic pathologist from 174 of 409 NLST screen-detected stage IA LUAD. Overdiagnosis rates were calculated from follow-up cancer incidence rates. Most BAC were reclassified as AIS/MIA/LMP (20/35 = 57%). The 7-year DSS was 100% for AIS/MIA/LMP and 94% for BAC. Excluding AIS/MIA/LMP, BAC behaved similarly to NSCLC (7-year DSS: 86% vs. 83%, p= 0.85) The overdiagnosis rate of LDCT stage IA NSCLC was 16.6% at 11.3-years, matching the proportion of AIS/MIA/LMP (16.2%) but not AIS/MIA (3.5%) or BAC (22.8%). AIS/MIA/LMP proportionally matches the overdiagnosis rate among stage IA NSCLC in the NLST, exhibiting 100% 7-year DSS. Biomarkers designed to recognize AIS/MIA/LMP preoperatively, would be useful to prevent overtreatment of indolent screen-detected cancers.

LDCT image biomarkers that matter most for the deep learning classification of indeterminate pulmonary nodules.

Continued improvement in deep learning methodologies has increased the rate at which deep neural networks are being evaluated for medical applications, including diagnosis of lung cancer. However, there has been limited exploration of the underlying radiological characteristics that the network relies on to identify lung cancer in computed tomography (CT) images. In this study, we used a combination of image masking and saliency activation maps to systematically explore the contributions of both parenchymal and tumor regions in a CT image to the classification of indeterminate lung nodules. We selected individuals from the National Lung Screening Trial (NLST) with solid pulmonary nodules 4-20 mm in diameter. Segmentation masks were used to generate three distinct datasets; 1) an Original Dataset containing the complete low-dose CT scans from the NLST, 2) a Parenchyma-Only Dataset in which the tumor regions were covered by a mask, and 3) a Tumor-Only Dataset in which only the tumor regions were included. The Original Dataset significantly outperformed the Parenchyma-Only Dataset and the Tumor-Only Dataset with an AUC of 80.80 ± 3.77% compared to 76.39 ± 3.16% and 78.11 ± 4.32%, respectively. Gradient-weighted class activation mapping (Grad-CAM) of the Original Dataset showed increased attention was being given to the nodule and the tumor-parenchyma boundary when nodules were classified as malignant. This pattern of attention remained unchanged in the case of the Parenchyma-Only Dataset. Nodule size and first-order statistical features of the nodules were significantly different with the average malignant and benign nodule maximum 3d diameter being 23mm and 12mm, respectively. We conclude that network performance is linked to textural features of nodules such as kurtosis, entropy and intensity, as well as morphological features such as sphericity and diameter. Furthermore, textural features are more positively associated with malignancy than morphological features.

Impact of OLD/Emphysema in LC Mortality Risk in Screening Programs: An Analysis of NLST and P-IELCAP

IntroductionThe impact of obstructive lung disease (OLD) and emphysema on lung cancer (LC) mortality in patients undergoing LC screening is controversial. MethodsPatients with spirometry and LC diagnosed within the first three rounds of screening were selected from the National Lung Screening Trial (NLST) and from the Pamplona International Early Lung Cancer Detection Program (P-IELCAP). Medical and demographic data, tumor characteristics, comorbidities and presence of emphysema were collected. The effect of OLD and emphysema on the risk of overall survival was assessed using unadjusted and adjusted Cox models, competing risk regression analysis, and propensity score matching. ResultsData from 353 patients with LC, including 291 with OLD and/or emphysema and 62 with neither, were analyzed. The median age was 67.3 years-old and 56.1% met OLD criteria, predominantly mild (1: 28.3%, 2: 65.2%). Emphysema was present in 69.4% of the patients. Patients with OLD and/or emphysema had worse survival on univariate analysis (HR: 1.40; 95% CI: 0.86–2.31; p=0.179). However, after adjusting for LC stage, age, and sex, the HR was 1.02 (95% CI: 0.61–1.70; p=0.952). Specific LC survival between both groups showed an adjusted HR of 0.90 (95% CI: 0.47–1.72; p=0.76). Propensity score matching found no statistically significant difference in overall survival (HR: 1.03; 95% CI: 0.59–1.9; p=0.929). ConclusionThe survival of LC patients diagnosed in the context of screening is not negatively impacted by the coexistence of mild OLD and/or emphysema.

Artificial Intelligence: Can It Save Lives, Hospitals, and Lung Screening?

BackgroundEarly detection is essential in lung cancer survival. Lung screening or incidental detection on unrelated imaging holds the most promise for early detection. With the large volume of imaging performed today, management of incidental pulmonary nodules can be difficult. We hypothesized an artificial intelligence (AI) tool could reliably read all imaging reports, detect, and effectively triage indeterminate pulmonary nodules without adding additional personnel, helping save lives. MethodsAn incidental lung nodule clinic (ILNC) was created using AI and an existing nurse practitioner. Over 26 months, the software read all radiology reports, visualizing any lung tissue. Patients with nodules >3 mm and considered indeterminate by the nurse practitioner were referred to the ILNC. High-risk patients with benign nodules were offered entry into the lung screening program. ResultsOf 502,632 imaging reports analyzed, 22,136 (4.4%) had positive findings. Follow-up data were lacking in 11,797 (2.3%), 911 (7.7%) were verified lost, and 518 (4.4%) were referred to the ILNC. There were 393 patients with benign nodules and accepted enrollment in the lung screening program. Mean age of enrolled patients was 61 years, and 53% were men. Workup included 499 diagnostic computed tomographic scans, 39 positron emission tomographic scans, and 27 biopsy samples that identified 15 malignancies (2.9%), with 14 lung cancers (8 stage I, 4 stage III, and 2 stage IV). Treatment included 5 lobectomies, and 4 underwent stereotactic body radiation therapy. Financials were favorable. ConclusionsAI software can supplement practitioners, help diagnose lung cancer earlier, save lives, and generate value-based revenue for the hospital.

Evaluation of the Efficacy of OSU-2S in the Treatment of Non-Small-Cell Lung Cancer and Screening of Potential Targets of Action.

(1) Background: OSU-2S is a derivative of FTY720 and exhibits significant inhibitory effects on various cancer cells. There is currently no research on the mechanism of the impact of OSU-2S on NSCLC development. We analysed and validated the hub genes and pharmacodynamic effects of OSU-2S to treat NSCLC. (2) Methods: The hub genes of OSU-2S for the treatment of NSCLC were screened in PharmMapper, genecard, and KM Plotter database by survival and expression analysis. The effect of OSU-2S on hub gene expression was verified by Western blot analysis. The ex vivo and in vivo efficacy of OSU-2S on tumour growth was verified using A549 cells and a xenografted animal model. (3) Results: A total of 7 marker genes for OSU-2S treatment of NSCLC were obtained. AURKA and S1PR1 were screened as hub genes. Significant differences in the expression of AURKA and S1PR1 between normal and lung adenocarcinoma (LUAD) tissues were found in the GEPIA2 database; Western blot showed that OSU-2S could affect p-AURKA and S1PR1 protein expression. OSU-2S significantly inhibited tumour growth in A549 cells and xenografted animal models. (4) Conclusions: Our study confirms the inhibitory effect of OSU-2S on NSCLC, screens and demonstrates its potential targets AURKA(p-AURKA) and S1PR1, and provides a research basis for treating NSCLC with OSU-2S.

Screening for Lung Cancer in India: Expert Opinion Statement.

Lung cancer (LC) has the highest rate of disability-adjusted life years (DALY) of all cancers in India. A large majority of patients with LC present with advanced disease, resulting in poor survival rates. Early diagnosis can improve survival outcomes as the patients can be treated with curative intent. The National Lung Screening Trial (NLST), in 53,454 persons at high risk for LC in the US, showed a 20% (95% confidence interval of 6.8-26.7; p = 0.004) relative reduction in LC-specific mortality in the patients screened with low-dose computed tomography (LDCT) compared with chest X-ray. To date, India does not have a formal LC screening (LCS) program. As a panel of experts, we reviewed a synthesis of a targeted literature search on the burden of LC, the current status of diagnosis of LC, barriers to early diagnosis, current referral pathways, LC risk patterns, use of artificial intelligence (AI) and risk calculators for risk assessment, and a multidisciplinary team (MDT) approach to diagnosis LC. We used the existing international LCS guidelines, data from published literature, and clinical experience to depict the characteristics of the population at risk of LC in India-young age (<40 years), smoking, especially the predominance of bidi smoking (an indigenous form of tobacco smoking), exposure to biomass fuel smoke, especially in rural women, and air pollution being the prominent features. LC in India is characterized by a higher rate of driver mutations and adenocarcinomatous histology. Here, we present the expert opinion on risk-based LCS in India and discuss the challenges, facilitators, and research priorities for the effective rollout of LCS in India.

MIB-Net: Balance the mutual information flow in deep learning network for multi-dimensional segmentation of COVID-19 CT images

Computed tomography (CT) based lung screening followed by computer-vision-based automated segmentation has great potential in early diagnosis of COVID-19, but low sample size and high patient-to-patient variability of infection characteristics are two of the major challenges. We propose a novel Mutual Information Balanced Net (MIB-Net) with three distinctive features. First, it captures mutual information across feature maps, CT scans, and predicted segmentation labels to optimize the contribution of different feature maps for better segmentation accuracy. Second, it incorporates a three-dimensional structural prior of CT scans based on predicted segmentation of neighboring CT slices to improve accuracy. Third, a novel Weight Optimization Filter (WOF) is used to adjust the proportion of 3D structural prior to be incorporated, which further improves the segmentation accuracy. Experimental results show that the proposed approach outperforms state-of-the-art biomedical image segmentation models by 7.87%, 3.85%, and 5.01% in Dice Score, respectively for three popular COVID-19 CT datasets. The better performance of the proposed approach suggests that it may serve as an attractive alternative to Lung CT-based screening for COVID-19.

Thyroid abnormalities identified on CT screening for lung cancer

PurposeBecause incidental thyroid nodules (ITNs) are common extrapulmonary findings in low-dose computed tomography (LDCT) scans for lung cancer screening, we aimed to investigate the frequency of ITNs on LDCT scans separately on baseline and annual repeat scans, the frequency of malignancy among the ITNs, and any association with demographic, clinical, CT characteristics. MethodsRetrospective case series of all 2309 participants having baseline and annual repeat screening in an Early Lung and Cardiac Action Program (MS-ELCAP) LDCT lung screening program from January 2010 to December 2016 was performed. Frequency of ITNs in baseline and annual repeat rounds were determined. Multivariable regression analysis was performed to identify significant predictors. ResultsDominant ITNs were seen in 2.5 % of 2309 participants on baseline and in 0.15 % of participants among 4792 annual repeat LDCTs. The low incidence of new ITNs suggests slow growth as it would take approximately an average of 16.8 years for a new ITN to be detected on annual rounds of screening. Newly detected ITNs on annual repeat LDCT were all smaller than 15 mm. Regression analysis showed that the increasing of age, coronary artery calcifications score and breast density grade were significant predictors for females having an ITN. No significant predictors were found for ITNs in males. ConclusionITNs are detected at LDCT however, no malignancy was found. Certain predictors for ITNs in females have been identified including breast density, which may point towards a common causal pathway.

Perspectives of potentially eligible Indigenous Māori on a lung cancer screening programme: a qualitative study

aims: Lung cancer causes more deaths than any other cancer, globally and in Aotearoa New Zealand, where it disproportionately affects Māori. We aimed to understand Māori perspectives on lung cancer screening in Aotearoa New Zealand to guide its equity focussed implementation, including identifying enablers and barriers. methods: We took a Kaupapa Māori based co-design approach to inform future screening, recruiting Māori current/ex-smokers and members of their whānau (family) for three focus group phases held in Auckland, Aotearoa New Zealand in August 2019. Participants responded to a proposed lung cancer screening pathway and shared their attitudes and beliefs about lung cancer and screening. Results were thematically analysed. results: The 21 Māori participants supported future lung cancer screening in Aotearoa New Zealand. Perceived benefits included being more informed about lung cancer and screening and enabling healthier future generations. Barriers to screening were previous negative health service experiences; fear; stigma; and access, including time, cost and transport. Enablers included providers’ cultural competence; clear communication; a one-stop shop; and support with transport. A range of factors could potentially influence a decision to participate in screening. conclusions: Participants favoured future lung cancer screening and identified key barriers and facilitators of screening.

Increasing power in screening trials by testing control-arm specimens: application to multicancer detection screening.

Cancer screening trials have required large sample sizes and long time-horizons to demonstrate cancer mortality reductions, the primary goal of cancer screening. We examine assumptions and potential power gains from exploiting information from testing control-arm specimens, which we call the "intended effect" (IE) analysis that we explain in detail herein. The IE analysis is particularly suited to tests that can be conducted on stored specimens in the control arm, such as stored blood for multicancer detection (MCD) tests. We simulated hypothetical MCD screening trials to compare power and sample size for the standard vs IE analysis. Under two assumptions that we detail herein, we projected the IE analysis for 3 existing screening trials (National Lung Screening Trial [NLST], Minnesota Colon Cancer Control Study [MINN-FOBT-A], and Prostate, Lung, Colorectal, Ovarian Cancer Screening Trial-colorectal component [PLCO-CRC]). Compared with the standard analysis for the 3 existing trials, the IE design could have reduced cancer-specific mortality P values 5-fold (NLST), 33-fold (MINN-FOBT-A), or 14 160-fold (PLCO-CRC) or, alternately, reduced sample size (90% power) by 26% (NLST), 48% (MINN-FOBT-A), or 59% (PLCO-CRC). For potential MCD trial designs requiring 100 000 subjects per arm to achieve 90% power for multicancer mortality for the standard analysis, the IE analysis achieves 90% power for only 37 500-50 000 per arm, depending on assumptions concerning control-arm test-positives. Testing stored specimens in the control arm of screening trials to conduct the IE analysis could substantially increase power to reduce sample size or accelerate trials and could provide particularly strong power gains for MCD tests.

Inter-vendor harmonization of CT reconstruction kernels using unpaired image translation.

The reconstruction kernel in computed tomography (CT) generation determines the texture of the image. Consistency in reconstruction kernels is important as the underlying CT texture can impact measurements during quantitative image analysis. Harmonization (i.e., kernel conversion) minimizes differences in measurements due to inconsistent reconstruction kernels. Existing methods investigate harmonization of CT scans in single or multiple manufacturers. However, these methods require paired scans of hard and soft reconstruction kernels that are spatially and anatomically aligned. Additionally, a large number of models need to be trained across different kernel pairs within manufacturers. In this study, we adopt an unpaired image translation approach to investigate harmonization between and across reconstruction kernels from different manufacturers by constructing a multipath cycle generative adversarial network (GAN). We use hard and soft reconstruction kernels from the Siemens and GE vendors from the National Lung Screening Trial dataset. We use 50 scans from each reconstruction kernel and train a multipath cycle GAN. To evaluate the effect of harmonization on the reconstruction kernels, we harmonize 50 scans each from Siemens hard kernel, GE soft kernel and GE hard kernel to a reference Siemens soft kernel (B30f) and evaluate percent emphysema. We fit a linear model by considering the age, smoking status, sex and vendor and perform an analysis of variance (ANOVA) on the emphysema scores. Our approach minimizes differences in emphysema measurement and highlights the impact of age, sex, smoking status and vendor on emphysema quantification.

Risk-based lung cancer screening performance in a universal healthcare setting.

Globally, lung cancer is the leading cause of cancer death. Previous trials demonstrated that low-dose computed tomography lung cancer screening of high-risk individuals can reduce lung cancer mortality by 20% or more. Lung cancer screening has been approved by major guidelines in the United States, and over 4,000 sites offer screening. Adoption of lung screening outside the United States has, until recently, been slow. Between June 2017 and May 2019, the Ontario Lung Cancer Screening Pilot successfully recruited 7,768 individuals at high risk identified by using the PLCOm2012noRace lung cancer risk prediction model. In total, 4,451 participants were successfully screened, retained and provided with high-quality follow-up, including appropriate treatment. In the Ontario Lung Cancer Screening Pilot, the lung cancer detection rate and the proportion of early-stage cancers were 2.4% and 79.2%, respectively; serious harms were infrequent; and sensitivity to detect lung cancers was 95.3% or more. With abnormal scans defined as ones leading to diagnostic investigation, specificity was 95.5% (positive predictive value, 35.1%), and adherence to annual recall and early surveillance scans and clinical investigations were high (>85%). The Ontario Lung Cancer Screening Pilot provides insights into how a risk-based organized lung screening program can be implemented in a large, diverse, populous geographic area within a universal healthcare system.

Uptake and 4-week quit rates from an opt-out co-located smoking cessation service delivered alongside community-based low-dose computed tomography screening within the Yorkshire Lung Screening Trial.

Up to 50% of those attending for low-dose computed tomography screening for lung cancer continue to smoke and co-delivery of smoking cessation services alongside screening may maximise clinical benefit. Here we present data from an opt-out co-located smoking cessation service delivered alongside the Yorkshire Lung Screening Trial (YLST). Eligible YLST participants were offered an immediate consultation with a smoking cessation practitioner (SCP) at their screening visit with ongoing smoking cessation support over subsequent weeks. Of 2150 eligible participants, 1905 (89%) accepted the offer of an SCP consultation during their initial visit, with 1609 (75%) receiving ongoing smoking cessation support over subsequent weeks. Uptake of ongoing support was not associated with age, ethnicity, deprivation or educational level in multivariable analyses, although men were less likely to engage (adjusted OR (ORadj) 0.71, 95% CI 0.56-0.89). Uptake was higher in those with higher nicotine dependency, motivation to stop smoking and self-efficacy for quitting. Overall, 323 participants self-reported quitting at 4 weeks (15.0% of the eligible population); 266 were validated by exhaled carbon monoxide (12.4%). Multivariable analyses of eligible smokers suggested 4-week quitting was more likely in men (ORadj 1.43, 95% CI 1.11-1.84), those with higher motivation to quit and previous quit attempts, while those with a stronger smoking habit in terms of cigarettes per day were less likely to quit. There was high uptake for co-located opt-out smoking cessation support across a wide range of participant demographics. Protected funding for integrated smoking cessation services should be considered to maximise programme equity and benefit.

Association of Lung Cancer Risk With the Presence of Both Lung Nodules and Emphysema in a Lung Cancer Screening Trial.

The coexistence of emphysema and lung nodules could interact with each other and then lead to potential higher lung cancer risk. The study aimed to explore the association between emphysema combined with lung nodules and lung cancer risk. A total of 21,949 participants from the National Lung Screening Trial (NLST) who underwent low-dose computed tomography (LDCT) examination were included. Participants were categorized into four groups (NENN group (non-emphysema and non-nodules), E group (emphysema without nodules), N group (nodules without emphysema), and E + N group (nodules with emphysema)) according to whether there were lung nodules and emphysema. Multivariable Cox regression and stratified analyses were performed to estimate the association between the four groups and lung cancer risk. Among the 21,949 participants, there were 9,040 (41.2%), 5,819 (26.5%), 4,737 (21.6%), and 2,353 (10.7%) participants in the NENN group, E group, N group, and E + N group. The risk of lung cancer incidence increased in turn in NENN group, E group, N group and E + N group. Compared with NENN group, the age-adjusted hazard ratios (HRs) (95% confidence intervals (CIs)) of lung cancer incidence were 2.07 (1.69 - 2.54) for E group, 4.13 (3.47 - 5.05) for N group, and 6.26 (5.14 - 7.62) for E + N group. The association was robust to adjustment for potential confounders (1.83 (1.47 - 2.27) for E group, 3.97 (3.24 - 4.86) for N group, and 5.23 (4.28 - 6.48) for E + N group). Comparable results as the lung cancer incidence were observed for lung cancer mortality, whether in age-adjusted model (E group: 1.85 (1.39 - 2.46), N group: 2.49 (1.89 - 3.29), E + N group: 4.27 (3.21 - 5.68)) or fully adjusted model (E group: 1.56 (1.15 - 2.11), N group: 2.43 (1.81 - 3.26), E + N group: 3.39 (2.50 - 4.61)). However, the trend of all-cause mortality risk among the four groups was somewhat different from that of lung cancer risk, whether in age-adjusted model (1.37 (1.21 - 1.54) for E group, 1.06 (0.92 - 1.21) for N group, and 1.75 (1.51 - 2.02) for E + N group) or fully adjusted model (1.26 (1.10 - 1.44) for E group, 1.09 (0.94 - 1.27) for N group, and 1.52 (1.30 - 1.79) for E + N group). Based on a large-scale lung cancer screening trial in the United States, this study demonstrated that either emphysema or lung nodules can increase lung cancer risk, and lung nodules combined with emphysema can further increase the lung cancer risk and all-cause mortality. The significance of these findings for lung cancer screening should be evaluated.

184 Prevalence and clinical outcomes of thoracic aortic disease detected within a UK lung screening programme: A 5 year follow-up study

184 Prevalence and clinical outcomes of thoracic aortic disease detected within a UK lung screening programme: A 5 year follow-up study

162 The LungScot study: overview of findings from a feasibility and acceptability of lung screening in Scotland

162 The LungScot study: overview of findings from a feasibility and acceptability of lung screening in Scotland

Interest in Lung Screening among a Sample of Adult Muslims in the United States

Objectives. Muslims in the United States exhibit high rates of tobacco use, which make them highly vulnerable to lung cancer. Lung cancer could be avoided by regularly undergoing preventive lung screening. This study investigated the associations between Social Cognitive Theory factors and being interested in preventive lung screening in a sample of adult US Muslims. Methods. We examined a convenience sample of adult US Muslims. A cross-sectional online survey was used to collect data. Binomial logistic regression analyses were conducted to identify the factors related to interest in preventive lung screening. Key Findings. Two hundred seventy-one eligible participants (≥18 years old Muslim individuals who live in the US) from 30 states completed the questionnaire. Individuals were more likely to express an interest in lung cancer screening if they had 1) more positive attitudes about lung screening, 2) a higher perceived value of screening, and 3) greater self-efficacy regarding the ability to undergo lung screening. Conclusion. Addressing attitudes, perceived value, and self-efficacy in future interventions could increase interest in preventive lung screening amongst adult US Muslims.

Lung CT harmonization of paired reconstruction kernel images using generative adversarial networks.

The kernel used in CT image reconstruction is an important factor that determines the texture of the CT image. Consistency of reconstruction kernel choice is important for quantitative CT-based assessment as kernel differences can lead to substantial shifts in measurements unrelated to underlying anatomical structures. In this study, we investigate kernel harmonization in a multi-vendor low-dose CT lung cancer screening cohort and evaluate our approach's validity in quantitative CT-based assessments. Using the National Lung Screening Trial, we identified CT scan pairs of the same sessions with one reconstructed from a soft tissue kernel and one from a hard kernel. In total, 1000 pairs of five different paired kernel types (200 each) were identified. We adopt the pix2pix architecture to train models for kernel conversion. Each model was trained on 100 pairs and evaluated on 100 withheld pairs. A total of 10 models were implemented. We evaluated the efficacy of kernel conversion based on image similarity metrics including root mean squared error (RMSE), peak signal-to-noise ratio (PSNR), and structural similarity index measure (SSIM) as well as the capability of the models to reduce measurement shifts in quantitative emphysema and body composition measurements. Additionally, we study the reproducibility of standard radiomic features for all kernel pairs before and after harmonization. Our approach effectively converts CT images from one kernel to another in all paired kernel types, as indicated by the reduction in RMSE (p<0.05) and an increase in the PSNR (p<0.05) and SSIM (p<0.05) for both directions of conversion for all pair types. In addition, there is an increase in the agreement for percent emphysema, skeletal muscle area, and subcutaneous adipose tissue (SAT) area for both directions of conversion. Furthermore, radiomic features were reproducible when compared with the ground truth features. Kernel conversion using deep learning reduces measurement variation in percent emphysema, muscle area, and SAT area.

Smoking, Lung Cancer Stage, and Prognostic Factors-Findings from the National Lung Screening Trial.

Low-dose computed tomography (LDCT) increases the early detection of lung cancer. Identifying modifiable behaviors that may affect tumor progression in LDCT-detected patients increases the likelihood of long-term survival and a good quality of life. We examined cigarette smoking behaviors on lung cancer stage, progression, and survival in 299 ever-smoking patients with low-dose CT-detected tumors from the National Lung Screening Trial. Univariate and multivariate Cox models were used to estimate the hazard ratio (HR) for smoking variables on survival time. Current vs. former smokers and early morning smokers (≤5 min after waking, i.e., time to first cigarette (TTFC) ≤ 5 min) had more advanced-stage lung cancer. The adjusted HR for current vs. former smokers was 1.3 (95% confidence interval [CI] 0.911-1.98, p = 0.136) for overall survival (OS) and 1.3 (0.893-1.87, p = 0.1736) for progression-free survival (PFS). The univariate hazard ratios for TTFC ≤ 5 min vs. >5 min were 1.56 (1.1-2.2, p = 0.013) for OS and 1.53 (1.1-2.12, p = 0.01) for PFS. Among current smokers, the corresponding HRs for early TTFC were 1.78 (1.16-2.74, p = 0.0088) and 1.95 (1.29-2.95, p = 0.0016) for OS and PFS, respectively. In causal mediation analysis, the TTFC effect on survival time was mediated entirely through lung cancer stage. The current findings indicate smoking behaviors at diagnosis may affect lung cancer stage and prognosis.