Censor-aware semi-supervised survival time prediction in lung cancer using clinical and radiomics features

Abstract

Aim: Lung cancer remains a major global health challenge, and this study presents a censor-aware semi-supervised learning framework (SSL) that integrates clinical and imaging data to improve prognostic modeling and address biases in handling censored data.

Methods: We analyzed clinical, positron emission tomography (PET), and computed tomography (CT) data from 199 lung cancer patients from public and local databases, focusing on overall survival time as the primary outcome. Handcrafted (HRF) and Deep Radiomics features were extracted after preprocessing using Visualized & Standardized Environment for Radiomics Analysis (ViSERA) software and were combined with clinical features. Features were reduced using Pearson’s correlation coefficient regression (RR) and the F-test for regression (FR), followed by supervised learning (SL) and SSL. In SSL, censored data were pseudo-labeled using the Weibull accelerated failure time (AFT) model to enrich the training data. Seven regressors and three hazard ratio survival analyses (HRSAs) were optimized using five-fold cross-validation, grid search, and holdout test bootstrapping.

Results: For PET-HRFs, the SSL approach reduced the mean absolute error by 14.81%, achieving 1.04 years with FR + AdaBoost Regression (ABR) compared to 1.20 years with SL. For clinical features, SSL with RR + ABR reached a mean absolute error of 1.04 years, outperforming SL (1.09 years) with a 4.9% improvement. In HRSA, CT_HRF combined with principal component analysis (PCA) + Component-Wise Gradient Boosting Survival Analysis yielded an external C-index of 0.65 ± 0.02, effectively distinguishing high- and low-risk groups.

Conclusions: The SSL strategy applied to HRFs from PET imaging significantly enhanced survival prediction compared to SL and uncovered complementary biological information that may remain hidden when only limited labeled data are used.

Keywords

Lung cancer / handcrafted radiomics features / deep radiomics features / machine learning / censor aware semi-supervised learning

References

[1] Siegel RL,Wagle NS.Cancer statistics, 2023.CA Cancer J Clin2023;73:17-48

[2] Sharma R.Mapping of global, regional and national incidence, mortality and mortality-to-incidence ratio of lung cancer in 2020 and 2050.Int J Clin Oncol2022;27:665-75 PMCID:PMC8753949

[3] Jouzdani AF,Rajabi M.Impact of clinical features combined with PET/CT imaging features on survival prediction of outcome in lung cancer.J Nucl Med2024;65:242130Available from: https://jnm.snmjournals.org/content/65/supplement_2/242130 [Last accessed on 21 October 2025]

[4] Salmanpour MR,Gorji A.Semi-supervised vs. supervised machine learning approaches for improved overall survival prediction: application to lung cancer PET/CT images.J Nucl Med2024;65:242097Available from: https://jnm.snmjournals.org/content/65/supplement_2/242097.abstract [Last accessed on 21 October 2025]

[5] Gorji A,Jouzdani A F.Region-of-interest and handcrafted vs. deep radiomics feature comparisons for survival outcome prediction: application to lung PET/CT imaging. In: 2023 IEEE Nuclear Science Symposium, Medical Imaging Conference and International Symposium on Room-Temperature Semiconductor Detectors (NSS MIC RTSD); 2023 Nov 4-11; Vancouver, BC, Canada, IEEE; 2023, pp. 1-2.

[6] Gorji A,Sanati N.PET-CT fusion based outcome prediction in lung cancer using deep and handcrafted radiomics features and machine learning.J Nucl Med2023;64:P1196Available from: https://jnm.snmjournals.org/content/64/supplement_1/P1196 [Last accessed on 21 October 2025]

[7] Dixon D,Moros N.Unveiling the Influence of AI predictive analytics on patient outcomes: a comprehensive narrative review.Cureus2024;16:e59954 PMCID:PMC11161909

[8] Salmanpour MR,Mousavi A.Enhanced lung cancer survival prediction using semi-supervised pseudo-labeling and learning from diverse PET/CT datasets.Cancers2025;17:285 PMCID:PMC11763441

[9] Zheng S,Langendijk JA.Survival prediction for stage I-IIIA non-small cell lung cancer using deep learning.Radiother Oncol2023;180:109483

[10] Atlam M,El-Fishawy N.Coronavirus disease 2019 (COVID-19): survival analysis using deep learning and Cox regression model.Pattern Anal Appl2021;24:993-1005 PMCID:PMC7883884

[11] Guo S.An overview of semiparametric models in survival analysis.J Stat Plan Inference2014;151-152:1-16

[12] Schober P.Survival analysis and interpretation of time-to-event data: the tortoise and the hare.Anesth Analg2018;127:792-8 PMCID:PMC6110618

[13] George B,Aban I.Survival analysis and regression models.J Nucl Cardiol2014;21:686-94 PMCID:PMC4111957

[14] Kapil A,Zuraw A.Deep semi supervised generative learning for automated tumor proportion scoring on NSCLC tissue needle biopsies.Sci Rep2018;8:17343 PMCID:PMC6255873

[15] Xie Y,Xia Y.Semi-supervised adversarial model for benign-malignant lung nodule classification on chest CT.Med Image Anal2019;57:237-48

[16] Shi F,Cao Q.Semi-supervised deep transfer learning for benign-malignant diagnosis of pulmonary nodules in chest CT images.IEEE Trans Med Imaging2022;41:771-81

[17] Chai H,Meng DY,Liang Y.A new semi-supervised learning model combined with Cox and SP-AFT models in cancer survival analysis.Sci Rep2017;7:13053 PMCID:PMC5638936

[18] Hermoza R,Nascimento JC.Censor-aware semi-supervised learning for survival time prediction from medical images. arXiv 2022;arXiv:2205.13226.

[19] Haredasht F.Predicting survival outcomes in the presence of unlabeled data.Mach Learn2022;111:4139-57

[20] Bakr S,Echegaray S.A radiogenomic dataset of non-small cell lung cancer.Sci Data2018;5:180202 PMCID:PMC6190740

[21] Lucignani G,Bombardieri E.The use of standardized uptake values for assessing FDG uptake with PET in oncology: a clinical perspective.Nucl Med Commun2004;25:651-6

[22] Brahim A,Ramírez J.Intensity normalization of DaTSCAN SPECT imaging using a model-based clustering approach.Appl Soft Comput2015;37:234-44

[23] Salmanpour MR,Hosseinzadeh M.ViSERA: visualized & standardized environment for radiomics analysis-a shareable, executable, and reproducible workflow generator. In: 2023 IEEE Nuclear Science Symposium, Medical Imaging Conference and International Symposium on Room-Temperature Semiconductor Detectors (NSS MIC RTSD); 2023 Nov 4-11; Vancouver, BC, Canada, IEEE; 2023: pp. 1-2.

[24] Freund Y.A decision-theoretic generalization of on-line learning and an application to boosting.J Comput Syst Sci1997;55:119-39

[25] Ho TK.Random decision forests. In: Proceedings of 3rd international conference on document analysis and recognition; 1995 Aug 14-16; Montreal, QC, Canada, IEEE; 1995, pp. 278-82.

[26] Srisuradetchai P.Random kernel k-nearest neighbors regression.Front Big Data2024;7:1402384

[27] Ahmed AM,Ulusoy AH.A decision tree algorithm combined with linear regression for data classification. In: 2018 International Conference on Computer, Control, Electrical, and Electronics Engineering (ICCCEEE); 2018 Aug 12-14; Khartoum, Sudan, IEEE; 2018, pp. 1-5.

[28] Stanton JM.Galton, Pearson, and the peas: a brief history of linear regression for statistics instructors.J Stat Educ2001;9:1-13

[29] Cybenko G.Approximation by superpositions of a sigmoidal function.Math Control Signal Systems1989;2:303-14

[30] Cortes C.Support-vector networks.Mach Learn1995;20:273-97

[31] Pölsterl S. scikit-survival: A library for time-to-event analysis built on top of scikit-learn. J Mach Learn Res 2020;21:1-6. Available from: https://www.jmlr.org/papers/volume21/20-729/20-729.pdf?utm_source [Last accessed on 21 October 2025]

[32] Pölsterl S. Gradient boosted models. 2020. Available from: https://scikit-survival.readthedocs.io/en/stable/user_guide/boosting.html [Last accessed on 21 October 2025]

[33] Pölsterl S. Using random survival forests. 2020. Available from: https://scikit-survival.readthedocs.io/en/stable/user_guide/random-survival-forest.html [Last accessed on 21 October 2025]

[34] Goel MK,Kishore J.Understanding survival analysis: Kaplan-Meier estimate.Int J Ayurveda Res2010;1:274-8 PMCID:PMC3059453

[35] Cui L,Hui W.A deep learning-based framework for lung cancer survival analysis with biomarker interpretation.BMC Bioinformatics2020;21:112 PMCID:PMC7079513

[36] Kurtipek E,Düzgün N.¹⁸F-FDG PET/CT mean SUV and metabolic tumor volume for mean survival time in non-small cell lung cancer.Clin Nucl Med2015;40:459-63

[37] Fried DV,Zhou S.Prognostic value and reproducibility of pretreatment CT texture features in stage III non-small cell lung cancer.Int J Radiat Oncol Biol Phys2014;90:834-42

[38] Huang B,Luo YH.Prediction of lung malignancy progression and survival with machine learning based on pre-treatment FDG-PET/CT.EBioMedicine2022;82:104127 PMCID:PMC9278031

[39] Yuan L,Zhu Y.Machine learning in diagnosis and prognosis of lung cancer by PET-CT.Cancer Manag Res2024;16:361-75

[40] Nakajo M,Ito S,Hirahara M.Clinical application of ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography radiomics-based machine learning analyses in the field of oncology.Jpn J Radiol2024;42:28-55 PMCID:PMC10764437

[41] Ahn HK,Kim SG.Pre-treatment ¹⁸F-FDG PET-based radiomics predict survival in resected non-small cell lung cancer.Clin Radiol2019;74:467-73

[42] Kirienko M,Antunovic L.Prediction of disease-free survival by the PET/CT radiomic signature in non-small cell lung cancer patients undergoing surgery.Eur J Nucl Med Mol Imaging2018;45:207-17

[43] Lee HY,Lee KS.Role of CT and PET imaging in predicting tumor recurrence and survival in patients with lung adenocarcinoma: a comparison with the international association for the study of Lung Cancer/American Thoracic Society/European Respiratory Society Classification of lung adenocarcinoma.J Thorac Oncol2015;10:1785-94

[44] Liu H,Zhang Y,Yin Y.Radiomics-based machine learning models for differentiating pathological subtypes in cervical cancer: a multicenter study.Front Oncol2024;14:1346336 PMCID:PMC11442173

[45] Hosseinzadeh M,Jouzdani AF,Rahmim A.Prediction of cognitive decline in Parkinson’s disease using clinical and DAT SPECT imaging features, and hybrid machine learning systems.Diagnostics2023;13:1691 PMCID:PMC10217464

[46] Salmanpour MR,Hosseinzadeh M.Deep versus handcrafted tensor radiomics features: prediction of survival in head and neck cancer using machine learning and fusion techniques.Diagnostics2023;13:1696 PMCID:PMC10217462

[47] Salmanpour MR,Sanati N.Tensor deep versus radiomics features: lung cancer outcome prediction using hybrid machine learning systems.J Nucl Med2023;64 (supplement 1):1174Available from: https://jnm.snmjournals.org/content/64/supplement_1/P1174.abstract [Last accessed on 27 October 2025]

[48] Le VH,Kha QH.A transfer learning approach on MRI-based radiomics signature for overall survival prediction of low-grade and high-grade gliomas.Med Biol Eng Comput2023;61:2699-712

[49] Hosny A,Mak RH.Handcrafted versus deep learning radiomics for prediction of cancer therapy response.Lancet Digit Health2019;1:e106-7

[50] Wu G,Refaee T.Structural and functional radiomics for lung cancer.Eur J Nucl Med Mol Imaging2021;48:3961-74 PMCID:PMC8484174

[51] Le NQK.Hematoma expansion prediction: still navigating the intersection of deep learning and radiomics.Eur Radiol2024;34:2905-7

[52] Lee B,Lee J.Breath analysis system with convolutional neural network (CNN) for early detection of lung cancer.Sens Actuators B Chem2024;409:135578

[53] Mazzone PJ,Dweik R.Diagnosis of lung cancer by the analysis of exhaled breath with a colorimetric sensor array.Thorax2007;62:565-8 PMCID:PMC2117257

[54] Wu HJ,Maliga Z.Spatial intra-tumor heterogeneity is associated with survival of lung adenocarcinoma patients.Cell Genom2022;2:100165 PMCID:PMC9681138

[55] Fridland S,Chae YK.Impact of intratumor heterogeneity (ITH) on survival outcomes in patients with non-small cell lung cancer (NSCLC) with high tumor mutational burden (TMB-H).J Clin Oncol2024;42:e20519-e20519