Analysis of Algorithms for Effective Skin Cancer Detection Model

EasyChair Preprint 6820

Abstract

Melanoma is the deadliest of all skin cancers, yet early detection   can increase your chances of survival. Due to the lack of knowledge  of  general practitioners, early diagnosis  is  one  of  the most difficult challenges. A clinical decision support system for general practitioners is described in this study, with the goal of saving time and money throughout the diagnosis process. The key steps in our approach are segmentation, pattern recognition, and change detection. The performance of Artificial Neural Network (ANN) learning algorithms for skin cancer diagnosis is also investigated in this paper. The capabilities of three learning algorithms, namely Levenberg-Marquardt (LM), Resilient Back propagation (RP), and Scaled Conjugate Gradient (SCG), in discriminating melanoma and benign lesions are investigated and compared. The results suggest that the Levenberg-Marquardt algorithm was quick and efficient in determining benign lesions, with specificity 95.1 percent, while the SCG algorithm produced superior results in diagnosing melanoma with sensitivity 92.6 percent at the cost of a larger number of epochs.

Keyphrases: Artificial Neural Network, Levenberg-Marquardt, Resilient Back propagation, Scaled Conjugate Gradient

@booklet{EasyChair:6820,
  author    = {Sam Praveen and C.V. Suresh Babu},
  title     = {Analysis of Algorithms for Effective Skin Cancer Detection Model},
  howpublished = {EasyChair Preprint 6820},
  year      = {EasyChair, 2021}}