Segmenting the Liver Tumor from Computed Tomography Using 3D-UNet with IDFOA Model

Cancer of the liver is a primary cause of mortality worldwide. Recognising cancerous tumour tissue is a difficult and time-consuming process. When compared to hepatoblastoma and other types of liver cancer, cholangiocarcinoma receives far less attention and statistical analysis. In this study, we analyse images of these two cancers specifically. This research will employ three different designs from the 3D UNet modifications to perform CT-scan segmentation: 3D UNet is the most widely suggested segmentation method for bio-medical pictures. In addition, the classification accuracy is enhanced by the use of hyper-parameter tuning performed by the Improved Dragonfly Optimisation Algorithm (IDFOA). The proposed approach is validated by experimental analysis utilising two datasets, namely 3DIRCADb and the Liver Tumour Segmentation Challenge (LITS). On the first dataset, the suggested model achieved an accuracy of 88%, while on the second dataset, it achieved an accuracy of 92.75.