1. MMC
  2. All Volumes and Issues
  3. Volume 12, Number 4, 2025
  4. Radiomics and Transformer-Based Deep Learning for Non-Invasive Prediction of PD-L1 Expression in Non-Small Cell Lung Cancer: A Paradigm Shift in Precision Oncology

Radiomics and Transformer-Based Deep Learning for Non-Invasive Prediction of PD-L1 Expression in Non-Small Cell Lung Cancer: A Paradigm Shift in Precision Oncology

MMC.
2025;
Volume 12, Number 4
: pp. 1145–1156
Received: March 28, 2025
Revised: October 24, 2025
Accepted: October 30, 2025

Karim A., Berquedich A., El Habib B., Zellou A.  Radiomics and Transformer-Based Deep Learning for Non-Invasive Prediction of PD-L1 Expression in Non-Small Cell Lung Cancer: A Paradigm Shift in Precision Oncology.  Mathematical Modeling and Computing. Vol. 12, No. 4, pp. 1145–1156 (2025) 

Authors:
  1. A. Karim
  2. A. Berquedich
  3. B. El Habib
  4. A. Zellou
1
Information Technology and Modeling Laboratory, Faculty of Sciences Ben M'sick, Hassan II University, Casablanca
2
SPM Team, ENSIAS, Mohammed V University, Rabat
3
Information Technology and Modeling Laboratory, Faculty of Sciences Ben M'sick, Hassan II University, Casablanca
4
SPM Team, ENSIAS, Mohammed V University, Rabat

Traditional biomarker testing for Programmed Death-Ligand 1 (PD-L1) in Non-Small Cell Lung Cancer (NSCLC) remains invasive and costly.  This study proposes a non-invasive alternative by integrating radiomic features extracted from CT scans with advanced deep learning architectures.  We evaluated Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), and Recurrent Neural Networks (RNN), including Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU).  Our results demonstrate that Transformer-based models significantly outperform conventional approaches, achieving a test Mean Squared Error (MSE) of 18.25 compared to 294.59 for ANN and 127.12 for CNN.  The optimized Complex Transformer architecture reduced MSE to 17.41 after 1000 epochs, with early stopping at epoch 261.  These findings highlight the potential of radiomics combined with Transformer models to enable accurate, cost-effective PD-L1 prediction, advancing personalized oncology while reducing reliance on invasive procedures.

radiomics
deep learning
PD-L1 biomarker
non-invasive diagnostics
non-small cell lung cancer
transformer models
  1. Xu T., Liu X., Chen Y., Wang S., Jiang C., Gong J.  CT-based deep learning radiomics biomarker for programmed cell death ligand 1 expression in non-small cell lung cancer.  BMC Med Imaging.  24 (1), 196 (2024).
  2. Zhang X., Zhang G., Qiu X., Yin J., Tan W., Yin X., Yang H., Wang H., Zhang Y.  Exploring non-invasive precision treatment in non-small cell lung cancer patients through deep learning radiomics across imaging features and molecular phenotypes.  Biomarker Research.  12 (1), 12 (2024).
  3. Li B., Su J., Liu K., Hu C.  Deep learning radiomics model based on PET/CT predicts PD-L1 expression in non-small cell lung cancer.  European Journal of Radiology Open.  12, 100549 (2024).
  4. Ferro A., Bottosso M., Dieci M. Vi., Scagliori E., Miglietta F., Aldegheri V., Bonanno L., Caumo F., Guarneri V., Griguolo G., Pasello G.  Clinical applications of radiomics and deep learning in breast and lung cancer: A narrative literature review on current evidence and future perspectives.  Critical Reviews in Oncology/Hematology.  203, 104479 (2024).
  5. Atmakuru A., Chakraborty S., Faust O., Salvi M., Barua P. D., Molinari F., Acharya U. R., Homaira N.  Deep learning in radiology for lung cancer diagnostics: A systematic review of classification, segmentation, and predictive modeling techniques.  Expert Systems with Applications.  255 (B), 124665 (2024).
  6. Wang C., Ma J., Shao J., Zhang S., Liu Z., Yu Y., Li W.  Predicting EGFR and PD-L1 Status in NSCLC Patients Using Multitask AI System Based on CT Images.  Frontiers in Immunology.  13, 813072 (2022).
  7. Wu J., Meng H., Zhou L., Wang M., Jin S., Ji H., Liu B., Jin P., Du C.  Habitat radiomics and deep learning fusion nomogram to predict EGFR mutation status in stage I non-small cell lung cancer: a multicenter study.  Scientific Reports.  14, 15877 (2024).
  8. Shi S., Wang Y., Wu J., Zha B., Li P., Liu Y., Yang Y., Kong J., Gao S., Cui H., Huangfu L., Sun X., Li Z., Liang T., Zheng Y., Yang D.  Predictive value of PD-L1 and TMB for short-term efficacy prognosis in non-small cell lung cancer and construction of prediction models.  Frontiers in Oncology.  14, 1342262 (2024).
  9. Sui D., Guo M., Ma X., Baptiste J., Zhang L.  Imaging Biomarkers and Gene Expression Data Correlation Framework for Lung Cancer Radiogenomics Analysis Based on Deep Learning.  IEEE Access.  9, 125247–125257 (2021).
  10. Bakr S., Gevaert O., Echegaray S., Ayers K., Zhou M., Shafiq M., Zheng H., Zhang W., Leung A., Kadoch M., Shrager J., Quon A., Rubin D., Plevritis S., Napel S.  Data for NSCLC Radiogenomics (Version 4) [Data set].  The Cancer Imaging Archive (2017).
  11. Mlynar J., Depeursinge A., Prior J. O., Schaer R., Martroye De Joly A., Evequoz F.  Making sense of radiomics: insights on human AI collaboration in medical interaction from an observational user study.  Frontiers in Communication.  8, 1234987 (2024).
  12. Van Griethuysen J. J. M., Fedorov A., Parmar C., Hosny A., Aucoin N., Narayan V., Beets-Tan R. G. H., Fillon-Robin J. C., Pieper S., Aerts H. J. W. L.  Computational Radiomics System to Decode the Radiographic Phenotype.  Cancer Res.  77 (21), e104–e107 (2017).