I need a dermoscopic image database to test an algorithm for automatic diagnose. In particular I am interested in images with blue-black colors within the lesion.
I recommend to visit derm101.com, there you can find an "Clinical Atlas of 101 Common Skin Diseases" with over 3,200 color photographs. There are also 900 "Quizzes", especially containing pigmented lesions. The Quizzes can be sorted by "pigmented lesions"
The issue here is the diagnosis. An interesting idea is to see if the registration through the smartphone (with the lenses) has identification power comparable to the dermoscope.
To test this question, it is necessary to obtain a set of images through the smartphone and to evaluate them also by dermoscopy. Then use the same automatic learning algorithm to see if the results are similar or not in terms of sensitivity and specificity.
Sending images to a database, in which much information is known, but without diagnosis, has little relevance in my opinion.
Hi, for those interested in skin imaging, it was recently published a light field dataset of skin lesions (SKINL2). The data set is captured with a focused plenoptic camera with extended depth of field.