Application of Separate Modal Analysis and Scale-Invariant Feature Transform on Clinical Data for the Screening of Breast Cancer

Public health data indicate that breast cancer is a major health problem among women worldwide, measured by the occurrence, mortality, and economic cost of diagnosis. Digital Image Elasto Tomography (DIET) system is a non-invasive breast cancer screening approach based on induced mechanical sinusoidal vibrations to the breast surface. Fiducial markers were applied on the breast, and an array of five digital cameras captures surface oscillations. Scale Invariant Feature Transform (SIFT) algorithm used to extract interest points from the image for estimating the frequency of motion. After removing the background of pendant breast, SIFT features were extracted using the skin texture. The movement of the breast surface was tracked by using the SIFT points instead. Separate modal analysis was applied to the SIFT points for detecting the presence and location of the tumor. In the second case, separate modal analysis was applied using both SIFT points and fiducial markers on given clinical and phantoms data. Clinical data of two healthy and two affected patients with breast cancer were used for cancer detection by extracting SIFT features from the bare skin. Results were matching when compared with a clinical summary of the patient's details. Given results indicate the tumor's presence and location and removes empty spaces compared to only markers based system.