Information Technology Reference
In-Depth Information
image-type data. This is why their uses have included cognitive analyses of images of
the central nervous system - the spinal cord - and the understanding of long bone
fracture images [16]-[24], [27]-[33]. Here we will present an attempt at using them for
a new class of medical images: to analyse foot bone images shown further down in
this publication.
The cognitive analysis of images showing foot bones has been conducted using
formalisms for the linguistic description, analysis and interpretation of data. These
include such formalisms as graph grammar. The purpose is to identify and intelli-
gently understand the analysed X-ray images of bones of the foot. In order to perform
a cognitive analysis aimed at understanding the analysed data showing foot bone
lesions, a linguistic formalism was proposed. It takes the form of an image grammar
whose purpose is to define a language describing the possible layouts of foot bones
which are within physiological norms and the possible lesions of foot bones.
The aim of this research project was to determine the utility of cognitive analysis
methods for this specific class of medical information systems and the analysed
images of foot bones. This utility will be measured by the effectiveness of executing a
task during which the system detects lesions indicating the presence of selected
disease units. The following have been distinguished among these units: fractures,
deformations, bone displacements and the appearance of an additional bone. The
lesions described can be divided further into various types of foot bone fractures,
degenerations leading to the skeleton being deformed, bone displacements,the appear-
ance of an additional bone among foot bones, the appearance of hematomas, calcifica-
tions and various irregularities in the structure of foot bones.
Cognitive reasoning and analysis methods were used in this project to detect all the
above groups of pathological phenomena related to foot bones. We will prove that the
results achieved confirm the suitability of the cognitive approach, although the
unanimous identification of all disease units appeared to be extremely difficult. This
was due to subtle differences in the input data (images) which were used to take the
decision to classify the case under consideration to a specific disease unit.
However, before foot visualisations are analysed, it is necessary to complete a
sequence of pre-processing operations. These operations help to extract all bones
making up the foot skeleton from the X-ray image. For this purpose it is necessary to
segment the image, label the detected bones, and determine their centres of gravity.
These centres will then be represented by the apexes of graph descriptions introduced.
A special algorithm previously developed for separating wrist bones [19]-[20] was
used to segment such images. the characteristic features of wrist and foot bones are
similar. Consequently, this algorithm is also suitable for segmenting the individual
small bones found in the foot. After the proper segmentation, the image showing
bones was subjected to median filtration to smooth out minor irregularities of the
contour.
After the necessary pre-processing operations are completed, we get binary images
showing the contours of all bones. These pre-processed images will be subjected to
further analysis aimed at creating graph representations in the projection considered
below.
One example of using the cognitive interpretation of image-type data to analyse
data depicting foot bone pathologies is an analysis of images acquired in the lateral
Search WWH ::
Custom Search