Information Technology Reference
In-Depth Information
and results of the classification). At this stage a 3D representation with the information
retrieved is displayed. The dimensionality is reduced by using MDS [18] [19] [20].
5 Case Study: Computational Intelligence Techniques for
Classification of CLL Leukemia
Microarray analysis has made it possible to characterize the molecular mechanisms
that cause several cancers. Regarding leukemia, microarray analysis has facilitated the
identification of certain characteristic genes in the different variants of leukemia [24]
[41] [45]. Cancer experts remark on the importance of the identification of the genes
associated to each type of cancer in order to establish the most efficient treatments for
the patients [46] [47]. The Cancer Institute in the city of Salamanca was interested in
novel tools for decision support in the process of CLL (Chronic Lymphocytic Leuke-
mia) patient classification. In this way, we focus on a concrete leukemia subtype, while
our previous works were aimed at classifying patients into leukemia subtypes [12].
The Institute provided us with 91 samples of patient data and asked for a tool to
provide decision support in the expression array analysis process and to incorporate
innovative techniques to reduce the dimensionality of the data and identify the vari-
ables with a higher influence in the patient's classification. The samples corresponded
to patients affected by chronic lymphocytic leukemia. CLL is a disease of lympho-
cytes that appear to be mature but are biologically immature. These B lymphocytes
arise from a subset of CD5-B cells that appear to have a role in autoimmunity. The
pathogenesis of chronic lymphocytic leukemia is likely a multistep process, initially
involving a polyclonal expansion of CD5-B cells followed by the transformation of a
single cell [48]. CLL is one of four main types of leukemia. About 15.110 new cases
of CLL will be diagnosed in 2008. Approximately 90.179 people are currently living
with CLL, more than the number of people living with any other type of leukemia.
Most people with CLL are at least 50 years old [49]. CLL starts with a change to a
single cell called a lymphocyte. Over time, the CLL cells multiply and replace normal
lymphocytes in the marrow and lymph nodes. The high number of CLL cells in the
marrow may crowd out normal blood-forming cells, and CLL cells are not able
to fight off infection like normal lymphocytes do [49]. The aim of the tests performed
in this study is to determine whether our system is able to classify new patients based
on previously analyzed and stored cases.
6 Results and Conclusions
This chapter has presented a case-based reasoning system, that evolved from a previ-
ous work in leukemia patients classification [12], specifically designed to analyze
data from microarrays, facilitating the grouping and classification of individuals.
Moreover, the system provides an innovative method for exploring the classification
process and extracting knowledge in the form of rules which help the human experts
to understand the classification process and obtain conclusions about the relevance of
the probes. The human experts in the laboratory have remarked on the advantages of
using the system as a decision support system for CLL classification, and have espe-
cially noted the facility in acquiring knowledge and explanations.
Search WWH ::
Custom Search