Share Email Print

Proceedings Paper

Human wound photogrammetry with low-cost hardware based on automatic calibration of geometry and color
Author(s): Abin Jose; Daniel Haak; Stephan Jonas; Vincent Brandenburg; Thomas M. Deserno
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Photographic documentation and image-based wound assessment is frequently performed in medical diagnostics, patient care, and clinical research. To support quantitative assessment, photographic imaging is based on expensive and high-quality hardware and still needs appropriate registration and calibration. Using inexpensive consumer hardware such as smartphone-integrated cameras, calibration of geometry, color, and contrast is challenging. Some methods involve color calibration using a reference pattern such as a standard color card, which is located manually in the photographs. In this paper, we adopt the lattice detection algorithm by Park et al. from real world to medicine. At first, the algorithm extracts and clusters feature points according to their local intensity patterns. Groups of similar points are fed into a selection process, which tests for suitability as a lattice grid. The group which describes the largest probability of the meshes of a lattice is selected and from it a template for an initial lattice cell is extracted. Then, a Markov random field is modeled. Using the mean-shift belief propagation, the detection of the 2D lattice is solved iteratively as a spatial tracking problem. Least-squares geometric calibration of projective distortions and non-linear color calibration in RGB space is supported by 35 corner points of 24 color patches, respectively. The method is tested on 37 photographs taken from the German Calciphylaxis registry, where non-standardized photographic documentation is collected nationwide from all contributing trial sites. In all images, the reference card location is correctly identified. At least, 28 out of 35 lattice points were detected, outperforming the SIFT-based approach previously applied. Based on these coordinates, robust geometry and color registration is performed making the photographs comparable for quantitative analysis.

Paper Details

Date Published: 20 March 2015
PDF: 8 pages
Proc. SPIE 9414, Medical Imaging 2015: Computer-Aided Diagnosis, 94143J (20 March 2015); doi: 10.1117/12.2081809
Show Author Affiliations
Abin Jose, RWTH Aachen Univ. (Germany)
Daniel Haak, RWTH Aachen Univ. (Germany)
Stephan Jonas, RWTH Aachen Univ. (Germany)
Vincent Brandenburg, RWTH Aachen Univ. (Germany)
Thomas M. Deserno, RWTH Aachen Univ. (Germany)

Published in SPIE Proceedings Vol. 9414:
Medical Imaging 2015: Computer-Aided Diagnosis
Lubomir M. Hadjiiski; Georgia D. Tourassi, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?