Content-based classification of images for liver cancer diagnosis

One well known liver disease is cancer of the liver. Since the current manual process of diagnosing the disease take time and is prone to subjectivity, the group aimed to perform a comparative analysis on the different image processing techniques and identity those that is/are suitable for liver cancer detection. The inputs used were histopathological images and the three (3) computational steps were allowed: processing, feature extraction and diagnosis. Each step included different image processing techniques. The output was a report containing the diagnosis and features identified in the image as a result of the image processing techniques used. The features extracted from the input images were the morphological, topological, fractal, intensity and textural features. The result of the experimental showed that the extraction of intensity feature of the image is the best among the five (5) features since the use of the said feature enabled the system to produce a %TP of 92.31% and a %TN of 93.33%