ResNet-fused external attention network with Taylor based mean absolute cross entropy for lung cancer subtype
Lakshmana Rao Padala1, Balajee Maram2, Naresh Tangudu3
1Department of Computer Science Engineering, School of Computer Science and Artificial Intelligence, SR University, Warangal, Telangana - 506371, India; Department of Computer Science Engineering (Data Science), Aditya Institute of Technology Management (AITAM), K Kotturu, Tekkali, Srikakulam, Andhra Pradesh - 532201, India.
Computational Biology and Chemistry
|June 15, 2026
Summary
A new ResNet-fused External Attention Network with Taylor based Mean Absolute Cross Entropy (Resf-TMACE) model accurately classifies lung cancer subtypes from histopathological images, improving diagnostic reliability.
Area of Science:
- Oncology
- Medical Imaging
- Computational Pathology
Background:
- Lung cancer is a leading cause of cancer mortality globally.
- Conventional diagnostic methods face challenges like subjective interpretation and class imbalance.
- Accurate lung cancer subtype classification is crucial for effective treatment and patient outcomes.
Purpose of the Study:
- To develop a robust lung cancer subtype classification model using histopathological images.
- To address limitations of existing methods including noise sensitivity and limited feature representation.
- To enhance early and reliable diagnosis of lung cancer.
Main Methods:
- Histopathological lung images were preprocessed using Non-Local Means (NLM) filtering and MaskmeanshiftCNN for cell segmentation.
- Image augmentation techniques including resizing, rotation, and flipping were applied.
- A novel ResNet-fused External Attention Network with Taylor based Mean Absolute Cross Entropy (Resf-TMACE) model was developed for classification.
Main Results:
- The Resf-TMACE model achieved high performance metrics.
- Accuracy: 97.738%
- True Positive Rate (TPR): 98.379%
- True Negative Rate (TNR): 96.368% at image size 512.
Conclusions:
- The proposed Resf-TMACE model demonstrates significant potential for accurate and reliable lung cancer subtype classification.
- This approach offers a promising tool to aid clinicians in early diagnosis and treatment planning.
- The integration of ResNet-fused External Attention Network and Taylor based Mean Absolute Cross Entropy effectively overcomes existing diagnostic challenges.