Journal of Computers, Mechanical and Management

PDF

Cite article

Rights and permissions

Rights and permissions

© The Author(s) 2026.

Open Access

This article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which permits the non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as appropriate credit to the original author(s) and the source is given by providing a link to the Creative Commons License and changes need to be indicated if there are any. The images or other third-party material in this article are included in the article's Creative Commons License, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons License and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this License, visit: httpshttps://creativecommons.org/licenses/by-nc/4.0/

S. Narad, K. T. V. Reddy, A hybrid CNN–Transformer model for detection and recurrence risk prediction of non-small cell lung cancer, Journal of Smart Sensors and Computing, 2026, 2(1), 26203, doi: https://doi.org/10.64189/ssc.26203.

Abstract

Critical challenges in medical diagnosis are being increasingly addressed through applications of artificial intelligence. Accurate detection and classification of non-small cell lung cancer (NSCLC) nodules smaller than 3 mm, along with reliable recurrence risk prediction, are essential for early diagnosis and improved patient outcomes. However, these tasks remain technically challenging. Existing approaches often struggle to detect and classify very small nodules because of limited image resolution and inadequate feature representation, which in turn negatively impacts recurrence risk prediction. To address these limitations, this study proposes an advanced deep learning framework that integrates a convolutional neural network (CNN)–transformer hybrid model. The CNN component extracts fine-grained local features from high-resolution computed tomography (CT) scans, while the transformer captures long-range contextual dependencies to enhance classification and prediction performance. The experimental results demonstrate a detection accuracy of 95% and a classification accuracy of 93% for nodules smaller than 3 mm. Overall, the proposed framework achieves 96% detection accuracy, 94% classification accuracy for small nodules, and 90% accuracy in recurrence risk prediction. Furthermore, the model provides enhanced interpretability, thereby supporting clinical decision-making. These findings indicate significant advancements in early NSCLC diagnosis and treatment planning.

Keywords:

• Medical imaging
• Non-small cell lung cancer
• CNN-Transformer
• Generative adversarial network-based super-resolution
• Recurrence prediction
• Machine learning

Graphical Abstract

Novelty statement

By integrating CNN–Transformer hybrids, SRGAN, self-attention, and LSTM, the proposed method enable accurate early-stage NSCLC detection.