An Integrated Framework of Skin Lesion Detection and Recognition through Saliency Method and Optimal Deep Neural Network Feature Selection

Published in Neural Computing and Applications (Springer), 2019

Authors:
M. Attique Khan, Tallha Akram, Muhammad Sharif, Kashif Javed, Muhammad Rashid, Syed Ahmad Chan Bukhari


This work presents an integrated framework for skin lesion detection and recognition by combining saliency-based segmentation with optimal deep feature selection. The proposed method improves classification performance by focusing on the most discriminative regions and features extracted from dermoscopic images.


Contributions 📃

  1. A framework for automatic skin lesion detection and classification.
  2. Use of saliency-based segmentation to focus on relevant lesion regions.
  3. Extraction of deep features from pretrained CNN models.
  4. Optimal feature selection to improve classification accuracy.
  5. Evaluation on dermoscopic image datasets with improved performance.

📖 Citation (BibTeX)

@article{khan2020integrated,
  title     = {An Integrated Framework of Skin Lesion Detection and Recognition through Saliency Method and Optimal Deep Neural Network Feature Selection},
  author    = {Khan, M. Attique and Akram, Tallha and Sharif, Muhammad and Javed, Kashif and Rashid, Muhammad and Bukhari, Syed Ahmad Chan},
  journal   = {Neural Computing and Applications},
  volume    = {32},
  number    = {20},
  pages     = {15929--15948},
  year      = {2020},
  publisher = {Springer},
  doi       = {10.1007/s00521-019-04514-0}
}

Paper PDF