Abstract:
Radiologist' visual observation of lung x-ray films is limited. This observation is more complicated
by the effect of noise from the background on image brightness. The aim of this research is to develop an
enhancement and segmentation techniques through digital image processing that can identify, localize and
characterize lung nodules. This method can help radiologists in identifying early lung tumors. Steps of
development process for chest x-ray image are as follows: (i) Preparation of chest x-ray image; (ii)
Convertion of images to gray scale; (iii) The enhancement process which is varied using four paths: (1)
spatial domains, (2) average specific frequency domains, (3) spatial domains then average specific
frequency, and (4) average specific frequency domains then spatial; (iv) Process segmentation; (v)
Characterization of objects; and (vi) Optimization of object attribute values. The result of the study shows
only three paths namely: path of (1), (3) and (4) can identify and localize the object of the tumor. Then do the
optimization of the results of measurement of the object area in all three paths methods, namely by changing
the value of the threshold variable, the variable divider smoothing process and the value of the contour
variable. Based on the optimization of the object area, the average special frequency domain method that
continued with the spatial domain is the best technique. This study has successfully developed digital image
processing techniques that can identify, localize and characterize lung tumors.
