Pattern Analysis in Dermoscopic Images

Melanoma Detection From Lesion Images Using Optimized Features Selected by Metaheuristic Algorithms

This paper deals with a simple but efficient method for detection of deadly malignant melanoma with optimized hand-crafted feature sets selected by three alternative metaheuristic algorithms, namely Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO) and Simulated Annealing (SA). Total 1898 number of features relating to lesion shapes, colors and textures are extracted from each of the 170 non-dermoscopy camera images of the popular MED-NODE dataset. This large feature set is then optimized and the number of features is reduced to up-to the range of single digit using metaheuristic algorithms as feature selector. Two well-known supervised classifiers, i.e. Support Vector Machine (SVM) and Artificial Neural Network (ANN) are used to classify malignant and benign lesions. The best classification accuracy result found by this method is 87.69% with only 7 features selected by PSO using ANN classifier which is far better than the results found in the literature so far.

Development and clinical validation of a novel photography-based skin pigmentation evaluation system: a comparison with the calculated consensus of dermatologists

OBJECTIVE

Various cosmetics, medicines, and light and laser treatments have been increasingly developed to improve pigmentary skin alterations such as melasma, actinic lentigo and dyschromia. To determine the efficacy of these modalities in view of the changes in pigmentation, an objective and reliable device that has a comparable performance to that of physicians is required. We developed a novel photography-based skin pigmentation evaluation system and validated its accuracy and reliability with a newly proposed method.

METHODS

A novel photography-based system was developed that integrates a consistent photography setting and image processing diagnostic algorithms. To automatically detect areas of pigmentation, the diagnostic algorithms were applied to photographs, which were obtained from 31 female patients. To validate its performance in comparison with the physicians' evaluation, five dermatologists independently evaluated the area of pigmentation. The clinical consensus area of pigmentation (CCAP) was calculated based on the consensus of five dermatologists' to exclude subjectivity or bias, and it was compared with the pigmentation area determined by the system.

RESULTS

Forty-four photographs with pigmented areas were evaluated by the system and the physicians. In contrast to the individual physician assessments, CCAP reduced the error that occurred due to subjectivity and bias, particularly for areas with indistinct pigmentation, and it was set as the gold standard. The results from the system showed a mean accuracy of 92.1% and a standard deviation of 4.6% in comparison with CCAP.

CONCLUSION

This pigmentation evaluation system can reproduce the physicians' consensus, suggesting that this system can support the dermatologists' objective evaluation of pigmentation.