124

Bioinformatics of the Brain

intersection of the lower points of these two valleys [37]. All pixels below the

threshold value can be deemed as part of the background and all pixels above

the threshold value can be considered as part of the object. In brain tumor

detection, the object we try to separate from the background is the tumor.

Thresholding methods can be roughly categorized as global thresholding

and local (adaptive) thresholding. Thresholding types are detailed in the fol-

lowing paragraphs.

Global Thresholding

In Global Thresholding, a single value is considered to be a threshold for an

entire image. It inputs a source image (src) and a threshold value (T) and

outputs an image (dst) based on the comparison of the pixel intensity at

the source pixel location (x, y) with the threshold value. Global Thresholding

is a widely used and easy-to-implement technique. The following paragraphs

discuss the various types of global thresholding methods.

Binary Thresholding is the most basic type of global thresholding. Pix-

els with an intensity value higher than the threshold will be set to maxV al

(white), while those lower than will be set to zero (black) (Equation 4.6) [38].

dst(x, y) =



maxV al

if src(x, y) > T

0

otherwise

(4.6)

The opposite form of binary thresholding is Inverse-Binary Thresholding,

in which the pixels whose intensity values are greater than the threshold will be

zero (black) and the pixels whose intensity values are lower than the threshold

will be maxV al (white) (Equation 4.7) [38].

dst(x, y) =



0

if src(x, y) > T

maxV al

otherwise

(4.7)

In Truncate Thresholding, while pixel values above the threshold are ad-

justed to the threshold value, other pixels remain same (Equation 4.8) [38].

dst(x, y) =

 T

if src(x, y) > T

src(x, y)

otherwise

(4.8)

The threshold-to-Zero technique will assign a value of zero to the source

pixels whose intensity values are lower than the threshold and will not modify

the values of the other pixels (Equation 4.9) [38].

dst(x, y) =



src(x, y)

if src(x, y) > T

0

otherwise

(4.9)

The inverted Threshold to Zero technique will do the opposite of the pre-

vious method and only change pixels whose intensity values are greater than

the threshold (Equation 4.10) [38].

dst(x, y) =

 0

if src(x, y) > T

src(x, y)

otherwise

(4.10)