Multi-Class Skin Lesions Classification System Using Probability Map Based Region Growing and DCNN
- DOI
- 10.2991/ijcis.d.200117.002How to use a DOI?
- Keywords
- Black frame removal; Gaussian filtering; Region growing; Optimal thresholding; Geometric features; SVM classification
- Abstract
Background:
Melanoma is a type of threatening pigmented skin lesion, and as of now is among the most hazardous existing diseases. Suitable automated diagnosis of skin lesions and Melanoma classification can extraordinarily enhance early identification of melanomas.
Methods:
However, classification models based on deterministic skin lesion can influence multi-dimensional nonlinear problem which leads to inaccurate and inefficient classification. This paper presents a Deep Convolutional Neural Network (DCNN) classification approach for segmented skin lesions in dermoscopy images. As an initial step, the skin lesion is preprocessed by an automatic preprocessing algorithm together with a fusion hair detection and removal strategy. Also a new probability map based region growing and optimal thresholding algorithm is integrated in our system which yields tremendous accuracy.
Results:
For obtaining more prominent results a set of features containing ABCD features as well as geometric features are calculated in the feature extraction step to describe the malignancy of the lesion.
Conclusions:
The experimental result shows that the system is efficient and works well on dermoscopy images, achieving considerable accuracy.
- Copyright
- © 2020 The Authors. Published by Atlantis Press SARL.
- Open Access
- This is an open access article distributed under the CC BY-NC 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/).
Download article (PDF)
View full text (HTML)
Cite this article
TY - JOUR AU - T. Sreekesh Namboodiri AU - A. Jayachandran PY - 2020 DA - 2020/01/22 TI - Multi-Class Skin Lesions Classification System Using Probability Map Based Region Growing and DCNN JO - International Journal of Computational Intelligence Systems SP - 77 EP - 84 VL - 13 IS - 1 SN - 1875-6883 UR - https://doi.org/10.2991/ijcis.d.200117.002 DO - 10.2991/ijcis.d.200117.002 ID - Namboodiri2020 ER -