Bio-medical Ontologies Maintenance and Change Management

Multimedia Medical Databases 103
We have chosen HSV and CIE-LUV color systems because they have good
characteristics for content-based visual query, as it was previously shown. It is
taken into consideration also the RGB color space, because of its large use, even if
it is not accomplished the above properties. There are considered different levels
of quantization to determine the way they affect the retrieval quality.
We have computed the similarity between the query and the target image with
three methods: the Euclidian distance (D1), the histogram intersection (D2) and
the histogram quadratic distance (D3).
The experiments were performed in the following conditions:
• It was created the test database with medical images extracted from DICOM
files
• Each image from the database was processed before the execution of any
query.
• For each experimental query, an image was chosen like query image and there
were established by a human factor, the relevant images for query.
• Each of the images relevant for the considered query was utilized, one by one,
for querying the database containing images. The final values of the precision
and the recall represent an average of the values resulted in the case of each
image taken one by one as query image.
• For comparing the obtained results, for each experimental query we draw the
graphic of the precision vs. recall for each of the three distances in the case of
each quantization method (figure 3.9). Also we present under a tabular form the
values that represent the number of relevant images, existing in the first 5,
respectively 10 retrieved images, and also number of images that must be
retrieved for finding among them the first 5, respectively 10 relevant images
(table 3.1).
Table 3.1. Query 1: Stomach and Duodenum Ulcers. Comparison of three distances in the
case of three methods of transformation and quantization.
M1
M2
M3
D1 D2 D3
5(9) 5(9) 5(9)
8(13) 6(11) 7(12)
5(9) 5(9) 5(9)
7(18) 6(11) 5(11)
4(7) 5(9) 4(7)
8(23) 6(13) 7(17)
We have performed four types of queries on color medical images representing
the followings diagnostics: stomach and duodenum ulcer, ulcerate cancer, hernias
and esophagus varicose. The values from table 3.1 represent an average of the
resulted values in the case of each image taken, one by one, as query image, for
the first diagnosis.