Detecting the Pigment Network in Dermoscopy Images: A Directional Approach
This paper presents an algorithm for the detection of the pigment network. The proposed algorithm explores the line color and geometry as well as the network topology using a bank of directional filters. Experimental results in a set of annotated images from Hospital Pedro Hispano show that the algorithm achieves good detection scores and it is therefore a useful tool in a dermoscopy analysis system. Future work should focus on a detailed evaluation of the proposed algorithm in a larger data base and the characterization of the detected network in order to discriminate typical from atypical patterns which are important clues to detect malign lesions.
• Improves the diagnosis accuracy.
• Addresses the detection of the pigment network using a bank of directional filters.
• Difficult to reliably detect the pigment network and discriminate it from other structures.
Dermoscopic image-analysis system: estimation of a typical pigment network and a typical vascular pattern
Image processing algorithms which estimate chromatic and shape parameters involved in the detection of the occurrence of two important criteria of the “7-point checklist” method have been proposed. The realized set-up allows the detection of two specific dermoscopic criteria: “A typical Pigment Network”, which is indicated by black, brown, or gray network with irregular meshes and thick lines, and “A typical Vascular Pattern”, which is indicated by linear irregular or dotted vessels. First experiments have demonstrated the usefulness of the algorithms in the detection of the occurrence of the A typical Pigment Network. About the second criterion further studies will be carried out. It will be also useful to have a wider image test set in order to tune the thresholds employed in the algorithm. In order to realize a system for the computer-aided diagnosis of melanocytic lesions the procedure will have to implement all the seven ELM criteria, and further efforts should have to be devoted to the expression of the results in a fuzzy way, in order to give clinicians a valid support in the diagnosis activity.
• Develop efficient schemes for the clinical diagnosis and computer-aided diagnosis systems, to assist dermatologists in different analysis steps.
• Simple geometric structures or tint estimations easily described to and determined by a computer algorithm.
• Problem consist in identifying the small percentage of skin lesions that develop into melanoma soon enough to treat them effectively.