| Headache is the most common illness for human. There are several
types of headache commonly occurs among headache patients,
each has their own symptoms which are visually and physically
noticeable, hence, the headache diagnosis is an open problem. The
preliminary observations on headache patients have shown irisdiscoloration
on the symptomatic side. This study proposes a
quantitative classification by taking basic primary color space,
chrominance and luminance, and human perspective color space
as representation of iris color space, as well as trying to investigate
which color components are the best solution to address the
problem.
ELM, a robust modification of single layer feedforward neural
network, is implemented to do the binary and multiclass
classification on 189 subject data, distributed into 162 headache
patients and 27 control subjects. Corresponding to the
conveniences of using ELM as classifier, the classification
performance was compared by considering the number of hidden
nodes inside ELM. The result obtained relatively good result to
distinguish control subject and headache patient, along with their
type of headache.
|
