Segmentation of Mammography Breast Images Using Automatic SEGMEN Adversarial Network with UNET Neural Networks

Abstract

Breast cancer is an extremely serious and lethal form of cancer. It is the second most common cancer among Indian women in rural areas. Early breast cancer detection can considerably enhance the effectiveness of treatment. Early detection of symptoms and signs is the most important technique to effectively treat breast cancer, as it enhances the odds of receiving an earlier, more specialist care. As a result, it has the possibility to significantly improve survival odds by delaying or entirely eliminating cancer. Mammography is a high-resolution radiography technique that is an important factor in avoiding and diagnosing cancer at an early stage. Automatic segmentation of the breast part using mammography pictures can help reduce the area available for cancer search, while also saving time and effort compared to manual segmentation. Previous studies have utilized convolutional and deconvolutional neural networks with autoencoder-like architecture (CN-DCNN) to perform automated breast region segmentation in mammography images. We present Automatic SegmenAN, an end-to-end unique adversarial neural network for the task of segmenting medical images, in this paper. Image segmentation necessitates extensive, pixel-level labeling, a standard GAN's discriminator's single scalar real/fake output may be inefficient in providing steady and appropriate gradient feedback for the networks. Our approach entails using a novel adversarial critic network and a multi-scale L1 loss function to prompt the critic and segmentor for acquiring local as well global features which encompass spatial relationships among pixels at both short and long ranges, instead of employing a segmentor which is a fully convolutional neural network. We demonstrate that an Automatic SegmenAN perspective is more up to date and reliable for segmentation tasks than any latest U-net segmentation technique.