Dataset Information

Prediction of lung cancer risk at follow-up screening with low-dose CT: a training and validation study of a deep learning method.

ABSTRACT: Background:Current lung cancer screening guidelines use mean diameter, volume or density of the largest lung nodule in the prior computed tomography (CT) or appearance of new nodule to determine the timing of the next CT. We aimed at developing a more accurate screening protocol by estimating the 3-year lung cancer risk after two screening CTs using deep machine learning (ML) of radiologist CT reading and other universally available clinical information. Methods:A deep machine learning (ML) algorithm was developed from 25,097 participants who had received at least two CT screenings up to two years apart in the National Lung Screening Trial. Double-blinded validation was performed using 2,294 participants from the Pan-Canadian Early Detection of Lung Cancer Study (PanCan). Performance of ML score to inform lung cancer incidence was compared with Lung-RADS and volume doubling time using time-dependent ROC analysis. Exploratory analysis was performed to identify individuals with aggressive cancers and higher mortality rates. Findings:In the PanCan validation cohort, ML showed excellent discrimination with a 1-, 2- and 3-year time-dependent AUC values for cancer diagnosis of 0·968±0·013, 0·946±0·013 and 0·899±0·017. Although high ML score cohort included only 10% of the PanCan sample, it identified 94%, 85%, and 71% of incident and interval lung cancers diagnosed within 1, 2, and 3 years, respectively, after the second screening CT. Furthermore, individuals with high ML score had significantly higher mortality rates (HR=16·07, p<0·001) compared to those with lower risk. Interpretation:ML tool that recognizes patterns in both temporal and spatial changes as well as synergy among changes in nodule and non-nodule features may be used to accurately guide clinical management after the next scheduled repeat screening CT.

SUBMITTER: Huang P

PROVIDER: S-EPMC7450858 | biostudies-literature | 2019 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Prediction of lung cancer risk at follow-up screening with low-dose CT: a training and validation study of a deep learning method.

Huang Peng P Lin Cheng T CT Li Yuliang Y Tammemagi Martin C MC Brock Malcolm V MV Atkar-Khattra Sukhinder S Xu Yanxun Y Hu Ping P Mayo John R JR Schmidt Heidi H Gingras Michel M Pasian Sergio S Stewart Lori L Tsai Scott S Seely Jean M JM Manos Daria D Burrowes Paul P Bhatia Rick R Tsao Ming-Sound MS Lam Stephen S

The Lancet. Digital health 20191017 7

<h4>Background</h4>Current lung cancer screening guidelines use mean diameter, volume or density of the largest lung nodule in the prior computed tomography (CT) or appearance of new nodule to determine the timing of the next CT. We aimed at developing a more accurate screening protocol by estimating the 3-year lung cancer risk after two screening CTs using deep machine learning (ML) of radiologist CT reading and other universally available clinical information.<h4>Methods</h4>A deep machine lea ...[more]

PMID: 32864596

Similar Datasets

Project description:PurposeTo measure reader agreement in determining whether lung nodules detected at baseline screening computed tomography (CT) had changed at subsequent screening examinations and to evaluate the variability in recommendations for further follow-up.Materials and methodsAll subjects were enrolled in the National Lung Screening Trial (NLST), and each participant consented to the use of their de-identified images for research purposes. The authors randomly selected 100 cases of nodules measuring at least 4.0 mm at 1-year screening CT that were considered by the original screening CT reader to be present on baseline CT scans; nodules considered by the original reader to have changed were oversampled. Selected images from each case showing the entire nodule at both examinations were preloaded on a picture archiving and communication system workstation. Nine radiologists served as readers, and they evaluated whether the nodule was present at baseline and recorded the bidimensional measurements and nodule characteristics at each examination, presence or absence of change, results of screening CT, and follow-up recommendations (high-level follow-up, low-level follow-up, no follow-up).ResultsOn the basis of reviews during case selection, five nodules seen at follow-up were judged not to have been present at baseline; for 19 of the remaining 95 cases, at least one reader judged the nodule not to have been present at baseline. For the 76 nodules that were unanimously considered to have been present at baseline, 21%-47% (mean ± standard deviation, 30% ± 9) were judged to have grown. The κ values were similar for growth (κ = 0.55) and a positive screening result (κ = 0.51) and were lower for a change in margins and attenuation (κ = 0.27-0.31). The κ value in the recommendation of high- versus low-level follow-up was high (κ = 0.66).ConclusionReader agreement on nodule growth and screening result was moderate to substantial. Agreement on follow-up recommendations was lower.

Project description:BackgroundOver 420,000 people have initiated life-saving antiretroviral therapy (ART) in Ethiopia; however, lost-to-follow-up (LTFU) rates continues to be high. A clinical decision tool is needed to identify patients at higher risk for LTFU to provide individualized risk prediction to intervention. Therefore, this study aimed to develop and validate a statistical risk prediction tool that predicts the probability of LTFU among adult clients on ART.MethodsA retrospective follow-up study was conducted among 432 clients on ART in Gondar Town, northwest, Ethiopia. Prognostic determinates included in the analysis were determined by multivariable logistic regression. The area under the receiver operating characteristic (AUROC) and calibration plot were used to assess the model discriminative ability and predictive accuracy, respectively. Individual risk prediction for LTFU was determined using both regression formula and score chart rule. Youden index value was used to determine the cut-point for risk classification. The clinical utility of the model was evaluated using decision curve analysis (DCA).ResultsThe incidence of LTFU was 11.19 (95% CI 8.95-13.99) per 100-persons years of observation. Potential prognostic determinants for LTFU were rural residence, not using prophylaxis (either cotrimoxazole or Isoniazid or both), patient on appointment spacing model (ASM), poor drug adherence level, normal Body mass index (BMI), and high viral load (viral copies > 1000 copies/ml). The AUROC was 85.9% (95% CI 82.0-89.6) for the prediction model and the risk score was 81.0% (95% CI 76.7-85.3) which was a good discrimination probability. The maximum sensitivity and specificity of the probability of LTFU using the prediction model were 72.07% and 83.49%, respectively. The calibration plot of the model was good (p-value = 0.350). The DCA indicated that the model provides a higher net benefit following patients based on the risk prediction tool.ConclusionThe incidence of LTFU among clients on ART in Gondar town was high (> 3%). The risk prediction model presents an accurate and easily applicable prognostic prediction tool for clients on ART. A prospective follow-up study and external validation of the model is warranted before using the model.

Project description:PurposeLow-dose CT screening allows early lung cancer detection, but is affected by frequent false positive results, inter/intra observer variation and uncertain diagnoses of lung nodules. Radiomics-based models have recently been introduced to overcome these issues, but limitations in demonstrating their generalizability on independent datasets are slowing their introduction to clinic. The aim of this study is to evaluate two radiomics-based models to classify malignant pulmonary nodules in low-dose CT screening, and to externally validate them on an independent cohort. The effect of a radiomics features harmonization technique is also investigated to evaluate its impact on the classification of lung nodules from a multicenter data.MethodsPulmonary nodules from two independent cohorts were considered in this study; the first cohort (110 subjects, 113 nodules) was used to train prediction models, and the second cohort (72 nodules) to externally validate them. Literature-based radiomics features were extracted and, after feature selection, used as predictive variables in models for malignancy identification. An in-house prediction model based on artificial neural network (ANN) was implemented and evaluated, along with an alternative model from the literature, based on a support vector machine (SVM) classifier coupled with a least absolute shrinkage and selection operator (LASSO). External validation was performed on the second cohort to evaluate models' generalization ability. Additionally, the impact of the Combat harmonization method was investigated to compensate for multicenter datasets variabilities. A new training of the models based on harmonized features was performed on the first cohort, then tested separately on the harmonized and non-harmonized features of the second cohort.ResultsPreliminary results showed a good accuracy of the investigated models in distinguishing benign from malignant pulmonary nodules with both sets of radiomics features (i.e., non-harmonized and harmonized). The performance of the models, quantified in terms of Area Under the Curve (AUC), was > 0.89 in the training set and > 0.82 in the external validation set for all the investigated scenarios, outperforming the clinical standard (AUC of 0.76). Slightly higher performance was observed for the SVM-LASSO model than the ANN in the external dataset, although they did not result significantly different. For both harmonized and non-harmonized features, no statistical difference was found between Receiver operating characteristic (ROC) curves related to training and test set for both models.ConclusionsAlthough no significant improvements were observed when applying the Combat harmonization method, both in-house and literature-based models were able to classify lung nodules with good generalization to an independent dataset, thus showing their potential as tools for clinical decision-making in lung cancer screening.

Dataset Information

Prediction of lung cancer risk at follow-up screening with low-dose CT: a training and validation study of a deep learning method.

Publications

Prediction of lung cancer risk at follow-up screening with low-dose CT: a training and validation study of a deep learning method.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets