Sign in Subscribe
Cancer research

Improved Performance in Melanoma Skin Cancer Classification using Deep Learning Based Ensemble Technique

Strategic Insights

— 1 min read

Research conducted by Mohammad Farukh Hashmi and colleagues at the National Institute of Technology in Warangal, India, has introduced a new deep learning based ensemble method to enhance the accuracy of melanoma skin cancer detection. The study presents a thorough performance evaluation of five ensemble techniques and achieves an overall accuracy of 81.99% and melanoma classification accuracy of 89.85% on a combined dataset of HAM10000 and ISIC2019 images. This research demonstrates the potential utility of ensemble techniques with adjusted model architectures in the classification of skin cancer images.

Key Takeaways:

  • The research introduced a deep learning based ensemble method to improve the accuracy of melanoma skin cancer detection, achieving an overall accuracy of 81.99% and melanoma classification accuracy of 89.85%.
  • The ensemble technique used a weighted average approach, combining the modified architectures of fine-tuned pre-trained models such as VGG16, MobileNetV2, and DenseNet169.
  • The dataset consisted of 15,606 images from the HAM10000 and ISIC2019 datasets, with images of seven skin lesion classes.
  • The research demonstrated the effectiveness of ensemble techniques in improving melanoma detection accuracy, affirming their potential utility in the classification of skin cancer images.
  • Mohammad Farukh Hashmi and colleagues presented their findings in the paper "Improved Performance On Melanoma Skin Cancer Classification Using Deep Learning Based Ensemble Technique" in the Intelligent Data Analysis: An International Journal.
  • The research was conducted at the National Institute of Technology in Warangal, India, and was peer-reviewed.

Statistics:

  • Overall accuracy of 81.99% on a combined dataset of HAM10000 and ISIC2019 images.
  • Melanoma classification accuracy of 89.85% using the weighted average ensemble model.
  • Consisted of 15,606 images from the HAM10000 and ISIC2019 datasets.
  • Seven skin lesion classes were included in the dataset.
  • Employed five ensemble techniques: weighted average, voting, bagging, boosting, and stacking.
  • Utilized a deep learning based ensemble method to enhance the accuracy of melanoma skin cancer detection.

Sources:

  • Hashmi, M. F., et al. "Improved Performance On Melanoma Skin Cancer Classification Using Deep Learning Based Ensemble Technique." Intelligent Data Analysis: An International Journal, vol. 2025, 2025, pp. 3973-3983.
  • National Institute of Technology, Warangal, India.
  • Ios Press, Nieuwe Hemweg 6B, 1013 BG Amsterdam, Netherlands.
  • NewsRx. "Research Conducted at National Institute of Technology Has Provided New Information about Skin Cancer (Improved Performance On Melanoma Skin Cancer Classification Using Deep Learning Based Ensemble Technique)." Health & Medicine Week, vol. 2025, no. 20, May 30, 2025, p. 3973.

Read more