An Innovative Machine Learning Approach to Diagnose Cancer at an Early Stage

For high level application perceiving the low‐level objects in an image has a great deal. Multiscale representation of the image can be used to detect the various cancers whereas the cancers appear in various ways based on their scale. Several cancer detection method and its applications are described in this chapter and they are compared fairly. The comparison defines about various cancers and its own strength and short coming. Automatic detection of cancers from data sets of images is a significant step in analyzing the huge scale of datas. Tumor positions in Magnetic resonance imaging or computerized tomography scan information, nuclear organization in refined colony and homogeneous sections in geophysical information can be represented by cancers. Before probing into cancer detection, initially few definitions are provided. Cancer is well‐defined as a region associated by minimum of one limited extreme, maximum to minimum no. of local extreme, to represent bright or dark cancer. In the image intensity function, the saddle point limits the spatial level of a cancer. The saddle point is point at which the intensity breaks diminishing and jumps aggregating for optimistic cancers and for dim cancers.