Automatic Assessment of Lesion Development in Hemangioma Follow-Up Images (bibtex)
by Sebastian Zambanini
Abstract:
This thesis presents an automatic method for the assessment of the development of cutaneous hemangiomas in digital images. The overall method provides two measurements on photographs taken during follow-up examinations: (1) the current skin area affected by the lesion and (2) the percentage/area of the hemangioma showing a regression, a so-called graying. For both analyses a pixel-wise classification scheme is applied to the images. The actual area measurement is accomplished through an image scale computation by means of a ruler attached to skin and visible in the images. Image registration is included in the assessment procedure to align follow-up images providing a direct comparison of color values necessary for a reliable detection of regressions. For image registration a robust feature-based method is presented that is able to deal with changing hemangioma appearances during follow-up. Experimental results are reported for the individual algorithms presented as well as for the whole procedure applied to follow-up images. In general, the proposed procedure working on follow-up images is expected to provide a more accurate and objective evaluation of the course of disease than the current clinical practice of manual measurement during an examination.
Reference:
Automatic Assessment of Lesion Development in Hemangioma Follow-Up Images (Sebastian Zambanini), Technical report, PRIP, TU Wien, 2007.
Bibtex Entry:
@TechReport{TR115,
  author =	 "Sebastian Zambanini",
  title =	 "Automatic Assessment of Lesion Development in Hemangioma Follow-Up Images",
  institution =	 "PRIP, TU Wien",
  number =	 "PRIP-TR-115",
  year =	 "2007",
  url =		 "ftp://ftp.prip.tuwien.ac.at/pub/publications/trs/tr115.pdf",
  abstract =	 "This thesis presents an automatic method for the assessment of the development of cutaneous hemangiomas in digital images. The overall method provides two measurements on photographs taken during follow-up examinations: (1) the current skin area affected by the lesion and (2) the percentage/area of the hemangioma showing a regression, a so-called graying. For both analyses a pixel-wise classification scheme is applied to the images. The actual area measurement is accomplished through an image scale computation by means of a ruler attached to skin and visible in the images. Image registration is included in the assessment procedure to align follow-up images providing a direct comparison of color values necessary for a reliable detection of regressions. For image registration a robust feature-based method is presented that is able to deal with changing hemangioma appearances during follow-up. Experimental results are reported for the individual algorithms presented as well as for the whole procedure applied to follow-up images. In general, the proposed procedure working on follow-up images is expected to provide a more accurate and objective evaluation of the course of disease than the current clinical practice of manual measurement during an examination.",
}
Powered by bibtexbrowser