Liver Tumor Segmentation Using CBAM-U-NET

Segmentation is the most important algorithm in image processing. In order to detect tumor lumps or nontumor lumps or any wound in human body, it need to be diagnosed and treated. The inner organ diagnosis is done by a powerful device like MRI, CT or ultra sound. Liver tumor is the most dangerous and it is a challenge for life survival. In order to detect and treat liver tumor, segmentation must be done very accurately. We propose a novel approach technique that combining convolutional block attention module (CBAM) in the U-Net framework. We improve the accuracy of liver tumor segmentation by combining spatial and channel attention techniques and employing Adam optimizer. We use Intersection over Union (IoU) metrics and Dice coefficient to test its accuracy. The accuracy of the proposed method is 98.87%, and the existing method shows 97.61%, 96.86% and 95.46%, respectively. As a result, the proposed approach provides a better solution for liver tumor diagnosis.