Content-based image retrieval techniques for the analysis of dermatological lesions using particle swarm optimization technique

• To retrieve images of similar diseases that lead to similar diagnosis.
• To provide improved diagnosis of diseases in medical applications.
• To guide the physician by predicting the disease of large number of dermatological cases.

This method presents extraction of effective color and shape features for the analysis of dermatology images. We employ three phases of operation in order to perform efficient retrieval of images of skin lesions. Our proposed algorithm used color and shape feature vectors and the features are normalized using Min–Max normalization. Particle swarm optimization (PSO) technique for multi-class classification is used to converge the search space more efficiently. The results using receiver operating characteristic (ROC) curve proved that the proposed architecture is highly contributed to computer-aided diagnosis of skin lesions. Experiments on a set of 1450 images yielded a specificity of 98.22% and a sensitivity of 94%. Our empirical evaluation has a superior retrieval and diagnosis performance when compared to the performance of other works. We present explicit combinations of feature vectors corresponding to healthy and lesion skin.

Figure optionsDownload as PowerPoint slide