![](bg1.png)
SPECKLE MODELIZATION IN OCT IMAGES FOR SKIN LAYERS
SEGMENTATION
Ali Mcheik, Clovis Tauber, Hadj Batatia
IRIT-ENSEEIHT, 2 rue Camichel BP7122, 31017 Toulouse, France
Jerome George, Jean-Michel Lagarde
CERPER, Laboratoires Pierre Fabre, Toulouse, France
Keywords:
Medical image analysis, Statistical approach, Segmentation and grouping, Optical coherence tomography.
Abstract:
In dermatology, the optical coherence tomography (OCT) is used to visualize the skin over a few millimetre
depth. These images are affected by speckle, which can alter the interpretation, but which also carry informa-
tion that characterizes locally the visualized tissue. In this paper, we present a statistical study of the speckle
distribution in OCT images. The capability of three probability density functions (pdf) (Rayleigh, Lognormal,
and Nakagami) to differentiate the speckle distribution according to the skin layer is analysed. For each pdf,
the vector of parameters, estimated over several images which are annotated by experts, are mapped onto a
parameter space. Quantitative results over 30 images are compared to the manual delineations of 5 experts.
Results confirm the potential of the method for the segmentation of the layers of the skin.
1 INTRODUCTION
The diagnosis and the treatment of pathologies of the
skin are largely based on a visual examination by
the dermatologists. This examination requires a great
experience because the skin can present ambiguous
states that are not easily interpretable. Often, as for
the monitoring of cancer, biopsies and histological
analysis are used to resolve these ambiguities. The
development of optical coherence tomography (OCT)
imaging aims at the realization of non invasive optical
biopsies.
OCT images allow the visualization of the struc-
tures of the skin, like the sweat glands, the stratum
corneum, or the change of contrast at the junction be-
tween the dermis and the epidermis. However, the de-
tailed examination of the images is strongly disturbed
by the presence of speckle. The speckle reduces con-
trast and makes difficult the interpretation of the im-
ages. It creates inter and intra variability among the
experts for the identification of the borders between
the different layers of the skin. This is particularly
true for tissues with high diffractors density like the
skin. The speckle is thus often regarded as a noise. It
is generally admitted that two types of speckle can be
found in OCT images(Schmitt, 1999; Raju and Srini-
Figure 1: Optical coherence tomography image of the skin
with manual delineations of two layers.
vasan, 2002). The first one comes from the interfer-
ence of several reflected photons. It appears as pixel-
sized dots with random value that can be filtered via
averaging techniques. The second type of speckle re-
sults from the interferences caused by the retrodiffu-
sions of the propagating waves front within the res-
olution cell of the imaging device. This speckle can
be found everywhere in the image. Several methods
can be found in the literature to reduce the speckle
in OCT images. Among these methods, the angu-
lar and spatial compoundingsignificantly increase the
347
Mcheik A., Tauber C., Batatia H., George J. and Lagarde J. (2008).
SPECKLE MODELIZATION IN OCT IMAGES FOR SKIN LAYERS SEGMENTATION.
In Proceedings of the Third International Conference on Computer Vision Theory and Applications, pages 347-350
DOI: 10.5220/0001086603470350
Copyright
c
SciTePress