Nabin K. Mishra;Ravneet Kaur;William Van Dover Stoecker;Reda Kasmi
发明人:
Ravneet Kaur,William Van Dover Stoecker,Nabin K. Mishra,Reda Kasmi
申请号：
US15785232
公开号：
US20180103892A1
申请日:
2017.10.16
申请国别(地区):
US
年份:
2018
代理人:
摘要：
Systems and methods facilitate segmenting a dermoscopy image of a lesion to facilitate classification of the lesion. A dermoscopy image is received from an image source； pre-processed； and segmented. Segmenting the pre-processed dermoscopy image includes applying a thresholding algorithm to the dermoscopy image. The thresholding algorithm includes at least one of a Huang auto-thresholding algorithm, a Li auto-thresholding algorithm, a Shanbhag auto-thresholding algorithm, an Otsu auto-thresholding algorithm, or an Isodata thresholding algorithm. A weighted border error metric is provided to allow prediction of whether segmentation is likely to avoid excluding portions of the lesion. A test is provided for determination of whether image inversion is needed for the methods to provide optimal borders. An iterative expansion algorithm is provided to for determine whether expansion is needed for the methods to provide optimal borders and to determine the degree of expansion needed for more inclusive segmentation.
