OVARY TUMOR PREDICTION USING R – CNN

The ovary, a complex organ that displays major structural and functional changes in the female reproductive system over recurrent periods. Medical images obtained from different medical imaging methods are used in the early identification and diagnosis of multiple diseases. The most serious and dangerous ovary tumor leads to a very short life expectancy. Magnetic Resonance Imaging ( MRI) is a widely used imaging method for the examination of these tumors, although the large amount of evidence produced by MRI prohibits manual segmentation within a realistic timeline and limits the use of precise quantitative measures in clinical practice. In this paper, an automated segmentation approach based on the Recursive Convolution Neural Network (R-CNN) is a kind of deep neural network generated by applying a region-based method for recognition, by integrating recursive connections into each convolutionary layer.. This approach includes primarily Filter-driven segmentation applied to whole slide histological images, driven on a convolutionary neural network (CNN). The use of small kernels allows the creation of a deeper architecture, in addition to having a beneficial effect on overfitting due to the lower weights in the network. The use of strength normalisation as a pre-processing stage was also explored, which, while not common in R-CNN-based segmentation methods, proved to be very effective for ovary tumor segmentation in MRI images in line with data increase