Unsupervised Restoration of Hair-Occluded Lesion in Dermoscopic Images.
Damilola A. Okuboyejo, Oludayo Olufolorunsho Olugbara, Solomon Adeyemi Odunaike · MIUA · 2014
The presence of artefacts such as hair shaft, thin blood vessel, ruler marking and air bubble in medical images makes the diagnosis of skin-related medical images very difficult. This paper uses a two-stage artefact detection termed Fast Image Restoration (FIR) via Canny algorithm and Line Segment Detection (LSD) operation for effective detection of artefacts. The Fast Marching Method (FMM) was applied at each stage for the removal of artefacts from a dermoscopic image in an unsupervised environment while ensuring morphological features of the lesion areas of the image data are preserved. Statistical Analysis performed to determine the accuracy of artefact recognition and repair validation of our method yields a Sensitivity of 98.27%, Specificity of 93.75% and Diagnostic Accuracy of 96.10%. These results indicate that our method gives an acceptable level of accuracy.