Table 5.

A comparative analysis of skin cancer detection using KNN-based approaches.

Ref	Skin Cancer Diagnoses	Classifier and Training Algorithm	Dataset	Description	Results (%)
[59]	Melanoma/nevus/normal skin	SOM and feed-forward NN	50 skin lesion images	PCA for decreasing spectra’s dimensionality	Accuracy (96–98)
[60]	BCC, SCC, and melanoma	SOM and RBF	DermQuest and Dermnet datasets	15 features consisting of GCM morphological and color features were extracted	Accuracy (93.15)
[61]	Cancerous/noncancerous	Modified KNN	500 lesion images	Automated Otsu method of thresholding for segmentation	Accuracy (98.3)

SOM = Self organizing map; PCA = Principal component analysis; GCM = Generalized co-occurrence matrices; RBF = Radial Basis Function; KNN = Kohonen self-organizing neural network.