Cervical cancer detection from MR images based on multiresolution wavelet analysis

Abstract

The demand for an automated system for diagnosis of cervical cancer images is steadily increasing in light of the ever increasing number of patients and the challenges involved in manual segmentation and classification of Magnetic Resonance Image (MRI) scans. This paper presents an experimental study aimed at developing an automatic image classification system for medical images by classifying Region of Interest (ROI) using the proposed framework. It is a challenge to identify abnormalities and quantify cervical tumor grading using just the shape, size and gray-level information of a patient's cervix. Multiresolution wavelet analysis of images by using wavelet transform is the heart of our proposed framework. In this review paper, we present a system to discriminate abnormality on the basis of procedure using image acquisition, preprocessing, segmentation, feature extraction, classification from normal patients based on their MRI scans obtained by TCIA. The tests are performed by using wavelet transforms of the images and results compared and analyzed.