Mayfly Optimization and Fully Convolutional Networks for Liver Tumor Classification

In recent decades, liver tumors have been considered as the leading and most serious causes of death worldwide.One of the most well-known Computer Vision (CV) is medical imaging (MI) which supports healthcare professionals in the early detection and diagnosis of liver tumors.To diagnose patients, professionals read hundreds of images using manual or Computed Tomography (CT).Consequently, a completely automated technique that makes use of CT scans is a common and extensively utilized imaging modality which requires an early diagnosis and identification of tumor.To overcome the above-stated issues, Mayfly Optimization (MFO) and Fully Convolutional Networks (FCN) are proposed for the classification of liver tumors.Image resizing approach is used for preprocessing and Adding Inception Module-UNet (AIM-UNet) is used for image segmentation.The suggested MFO algorithm is used for feature selection which takes the input from the extracted features using ResNet50.FCN is used as a classifier to classify liver tumors.In comparison to the existing techniques like EfficientNetB0 and InceptionV3, the proposed FCN achieves commendable accuracies of 99.5% and 99.98%, respectively, on the LiTS and CHAOS datasets.