Detection and Classification of Cervical Cancer in Pap Smear Images using EETCM, EEETCM & CFE methods based Texture features and Various Classification Techniques
S. Athinarayanan, M. V. Srinath, R. Kavitha · Zenodo (CERN European Organization for Nuclear Research) · 2016
Cervical cancer is the second most common gynaecologic cancer worldwide. Unlike the other cancers it does not show any symptoms in its earlier stage which causes mortality among women. It takes 8 to 10 years to develop from precancerous to severe stage. The important reasons for higher cervical cancer in developing countries are lack of resources, lack of effectual screening programs and inadequately organized health system aimed for detecting precancerous condition before they progress to persistent cancer and also 80% of cervical cancers are incurable at the time of detection due to their advanced stage. Therefore, early detection of cervical cancer is more important for reducing the mortality rate of the women. Thus, the aim of this paper is to investigate about the classification of cervical cell as Normal Cell or Abnormal Cell by using individual feature extraction method and joining individual feature extraction features method with the classification technique. In this paper, we proposed three novel feature extraction methods. From that three, two were individual feature extraction methods, they are Extending Enriched Texton Co-Occurrence Matrix (EETCM) and Effective Extending Enriched Texton Co-Occurrence Matrix (EEETCM) and the remained one was joining individual feature extraction features method named as Concatenated Feature Extraction (CFE). The CFE method represents joining all the individual feature extraction methods of EETCM, EEETCM features into one feature to assess their joint performance. Then these three feature extraction methods are tested over two classifiers: Kernel Support Vector Machine (K-SVM) and Support Vector Machine (SVM). This Examination was conducted over a set of single cervical cell based pap smear images. The dataset contains two classes of images, with a total of 952 images. The distribution of number of images per class is not uniform. Then, the performance of the proposed system was evaluated in terms of the statistical parameters of sensitivity, specificity & accuracy in both the individual feature extraction & classification combinations and joining all the individual feature extraction features method and classification combinations. Hence, the performance of individual combination method described, the proposed EEETCM features with Kernel SVM Classifier combination had given the better results than the other combinations such as EEETCM with SVM Classifier, EETCM with Kernel SVM Classifier, EETCM with SVM Classifier. Then the performance of joining all the individual feature extraction method and classification combination described, proposed Concatenated Feature Extraction (CFE) with Kernel SVM Classifier had given the better results than CFE with SVM Classifier and all other individual feature extraction and classification combinations.