Skin Cancer Detection Using CNN with Texture and Feature Analysis

Authors

  • Shaik Imam Saheb B.E. Student Department of CSE, Lords Institute of Engineering and Technology, Hyderabad, India. Author
  • Syed Ahmeduddin Assistant Professor, Department of CSE, Lords Institute of Engineering and Technology, Hyderabad, India. Author

Keywords:

Skin cancer; Melanoma; Dermoscopy; Convolutional neural network; Texture analysis; GLCM; Local binary patterns; ABCD rule; Feature fusion; Attention mechanism

Abstract

Early and accurate diagnosis of skin cancer significantly improves patient outcomes, yet access to trained 
dermatologists remains limited in many institutional and community healthcare settings, and manual visual 
inspection of dermoscopic images is prone to inter-observer variability. Existing automated approaches often rely 
solely on deep convolutional features and under-use the complementary texture and handcrafted information that 
dermatologists themselves rely on. This paper presents TexFuse-Net, a lightweight skin cancer detection 
framework that combines convolutional neural network (CNN) feature learning with texture and handcrafted 
feature analysis to classify dermoscopic lesion images as benign or malignant. Images are first cleaned by 
morphological hair removal and Shades-of-Gray colour constancy, and the lesion is segmented with a U-Net. 
Three feature streams are then computed: deep features from a fine-tuned EfficientNet-B0, texture features from 
grey-level co-occurrence matrices (GLCM), multi-scale local binary patterns (LBP) and colour moments, and 
clinically motivated ABCD features (asymmetry, border, colour and diameter). An attention-based fusion layer 
learns how much weight to give each stream, and the model is trained with focal loss to address class imbalance. 
The system returns a calibrated malignancy probability, a Monte Carlo dropout uncertainty estimate and a Grad
CAM heat-map as a confidence report. On the HAM10000 dataset, the framework is expected to achieve 93.3% 
accuracy, 84.0% malignant-class recall, 83.0% F1-score and an AUC of 0.961, outperforming CNN-only and 
handcrafted-only baselines 

Downloads

Published

2023-08-28

How to Cite

Skin Cancer Detection Using CNN with Texture and Feature Analysis . (2023). International Journal of Engineering and Science Research, 13(3), 01-12. https://ijesr.org/index.php/ijesr/article/view/1899

Similar Articles

221-230 of 1052

You may also start an advanced similarity search for this article.