Smart Phone Based Diabetic Wound Image Processing System

weakassuredAI and Robotics

Nov 6, 2013 (3 years and 8 months ago)

90 views

Smart Phone Based
Diabetic Wound Image Processing System

Abstract

Diabetic foot ulcers represent

a s
ignificant health issue. The ulcers are

painful, susceptible to
infection and even can lead to amputation of affected limbs.
According to clinical
experience
,
good daily care, especially
through
pat
ients’ self
-
management, is

necessary for accelerating
wound healing.
In view of the prevalence of smart cell
-
phones with camera, monitoring
wounds by real
-
time taken images would be an effective and conve
nient method.

In this project, w
e proposed a novel wound image processing system
, to be integrated into an
Android platform,

to accomplish the task of wound image analysis and wound care
recommendation. This system consists o
f several sub
-
modules:
1) JEPG
-
decompression, 2)
image
-
preprocessing, 3) wound area segmentation, 4) color segmentation for single wound
area, 5) trend analysis,
and
6) wound care recommendation. We plan
to segment all wound
areas

in the image

separately by applying the distance regular
ized level set evolution (DRLSE)
method, which eliminates the need for reinitialization of level set function and thereby improve
the both the accuracy and efficiency
compared with

traditional level set evolution algorithm.
Moreover, we plan to accelerate
the segmentation process further by focus the evolution only
to a narrow
-
band
surrounding the wound area. Within

each wound area, we will
separate
different tissues

by
a
fast K
-
Mean clustering algorithm based on the color feature defined in
black
-
yellow
-
r
ed wound analysis model.

M
odules of JPEG decompression, image
-
processing, wound segmentation and color
segmentation for a single wound area have been implemented. The boundaries for wound
areas
and different wound tissues
have

be
en
accurately

defined

with
an appropriate set of
initial curve and evolution control parameters.
In our future work, we plan to improve the
segmentation algorithm further to adjust to variety of wound situations. Besides, the sample
training, trend analysis and wound care recommenda
tion modules are also in progress of
development.

After all the

software

modules are realized on PC, we will consider about
integrating this system into Android system for smart phones.