Segmentation and 3D reconstruction of MRI images for breast cancer detection

Cancer is a dense and abnormal cells proliferation in the body tissue. Breast cancer is the most common in woman´s life. Fortunately, science evolution has led to the development of medical imaging techniques. The latter are efficiently used to detect any abnormality in breast parenchyma. Among these techniques, we can mention the MRI which is very relevant especially in terms of dubious image analysis by mammography and ultrasound. Our research addresses the problem of detecting this type of cancer and a three dimensional reconstruction of MRI images for breast cancer detection. We have segmented 2D MRI images and then make a 3D reconstruction. Segmentation allows us to locate the tumor. We have looked much more towards the elimination of false positives to obtain a clear segmented image. The used segmentation methods are based on a structural approach to isolate the breast edge and a region approach to extract the tumor. For segmenting the breast skinline, we have developed and proposed a method that browses all pixels of images to research those which belong to breast edge. After this extraction, the segmentation of the tumor is performed by K-means algorithm preceded filtering operations. To better visualize the tumor and understand its expansion, the 3D reconstruction is then performed by an indirect volume rendering method, the Marching Cubes and a direct volume rendering method, the Maximum Intensity Projection.