Liver Tumor Segmentation and Classification Using Deep Learning

The high variation in the size of tumor, location, depth, intimate boundaries of other organs around liver, liver tumor segmentation is the most important yet challenging task. The size and other characteristics of liver computed tomography (CT) scan vary, including the number of layers in each one. Thus, obtaining liver CT scans with varying parameters is a difficult task. As a result, the project includes an effective method for segmenting liver tumors using 3D Unet, a Convolutional Neural Network. By generating suggestions about the tumor region, the proposed method, which is based on the RESNET approach, efficiently identifies tumors in the liver. In the preprocessing step, the proposed method creates a cardinal outline of the liver by converting CT volumes to normalized CT image slices. For feature extraction, normalized CT images are then fed to Resnet. A 3D FRN method is used to classify the data. The dataset, which includes 13000 CT scans of various patients taken at various clinical sites, is freely accessible on the LiTS challenge website. The CT image dataset is used to train the proposed method. This quickly classifies and diagnoses liver tumors by combining image processing techniques with Convolutional Neural Network (CNN) algorithms.