<script async src="//pagead2.googlesyndication.com/pagead/js/adsbygoogle.js"></script>
<!-- martinchinwe -->
<ins class="adsbygoogle"
style="display:inline-block;width:728px;height:90px"
data-ad-client="ca-pub-3432567292388333"
data-ad-slot="3193770573"></ins>
<script>
(adsbygoogle = window.adsbygoogle || []).push({});
</script>
2.1.Image Classification
The classification algorithm uses the concept of the ratio of the number of white and black pixels in binary image [80]. It is the process of using techniques to detect and isolate various desired features of an image [22]. The features to be classified are expected to carry enough information about the image. Classification is the last stage in the image abnormality determination. In lung cancer detection, the most important feature to be considered during lung image classification is the feature that represents the spiculated mass characteristics of the lung image [31]. Spiculation is the small spikes of masses that are present in malignant mass data indicating excessive growth of cells. There are two approaches to image classification, thus:
· The binarization/ pixels percentage
· The masking
2.1.1. Binarization method
The binarization of an image involves counting the number of black pixels in an image after converting the image to a binary image. The binarization technique was extensively used in the work done by [22]. In lung cancer detection, the binarization approach depends on the fact that the number of black pixels is more than the number of white pixels in the normal lung image [11]. The binarization technique is based on the fact that number of white pixels increases as the cancer spread; the number of white pixels increases with the stage of the cancer development. Binarization involves counting the number of black pixels and comparing it with the established threshold to determine the abnormality or normality of the image. In image classification, the classification threshold is used.
When the number of black pixels in an image is greater than , the image is normal, but when the number of black pixels is less than , the image is abnormal. If Tc is much greater than the number of black pixels in lung image, the lung infection is at its advanced stage. However, if Tc is insignificantly larger than the number of black pixels in lung image, the cancer is at the early stage.
2.1.2. Masking method
The mask is created using mathematical morphological functions [81]. The masking method applies the principles of object extraction from an image to reduce the complexity of the classification. The complexity of the image is reduced since only the region of interest (ROI) is extracted. The ROI of an image is a region that has the attributes (local colour, global colour, texture, and structure) needed in object extraction. The extraction of the features from an image can be done using a variety of image processing techniques [82]. One of the outstanding methods of mask creation is the used of the active contour method [46]. The contour can have various shapes such as square, rectangle, polygon and ellipse. The active contour method uses gradient-based optimal method defined by an energy function to separate the image into object regions.
The extraction of each region of identical feature is an important step in lung cancer classification. However, the number of regions must be minimized to ensure better classification since a large number of regions give unstable results [81]. When a lung is infected with cancer, the area of the dark region is much less than the area of white region since the cancer destroys some of the dark area. The masking method depends on the fact that the masses appear as white connected areas in the lung (ROI) [22], [83]. As the white interconnection area increases, the percentage of cancer presence increases. The combination of the two binarization and masking method will conclusively show the normality or abnormality of the image [22].