Search Results for author:
Found 3 papers, 0 papers with code
Can self-training identify suspicious ugly duckling lesions?
We first automatically detect and extract all the lesions from a wide-field skin image, and calculate an embedding for each detected lesion in a patient image, based on automatically identified features.
Out-of-Distribution Detection for Dermoscopic Image Classification
This problem is especially important for medical image diagnosis, when an image of a hitherto unknown disease is presented for diagnosis, especially when the images come from the same image domain, such as dermoscopic skin images.
Diagnostic Accuracy of Content Based Dermatoscopic Image Retrieval with Deep Classification Features
Methods: A neural network was trained to predict disease classes on dermatoscopic images from three retrospectively collected image datasets containing 888, 2750 and 16691 images respectively.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
