Semantic Segmentation of Kidney Tumors Using Variants of U-Net Architecture

Abstract

Kidney Cancer is one of the most prevalent diseases that is more common in men than in women. Detecting kidney tumors at an early stage has been found to increase survival rates of patients. It is therefore important to accurately segment tumors in Computed Tomography(CT) images. To assist in early detection of kidney tumors in CT images, we present a method for segmenting kidney tumors using deep convolutional neural networks. Predicted models using U-Net and Attention U-Net architectures are ensemble for effective tumor segmentation. Experimental and visual results obtained using the KiTS2019 dataset clearly demonstrate the enhanced Intersection Over Union(IoU) score of the ensemble model.

Segmentation of Retinal Images Using Improved Segmentation Network, MesU-Net

“…The performance of this architecture outperformed U-Net by 10.15 percent, 5.0 percent, 2.6 percent, 1.4 percent, and 0.6 percent, respectively. TM G et al [24] utilized attention-based U-Net for segmentation of tumor. This model accomplished the ensemble by weighted averaging and showing promising results with an IOU score of 0.9591.…”

Noise Invariant Convolution Neural Network for Segmentation of Multiple Sclerosis Lesions from Brain Magnetic Resonance Imaging

“…The improvised model achieves a better MS lesion segmentation outcome than the standard CNN model [32,33]. In [28] showed that the UNET architecture are very good in performing segmentation of tumor in Kidney using CT images. However, tumor segmentation using brain MRI [29] is challenging due to presence of motion artifacts; thus, requires efficient preprocessing technique [30,31].…”