A ResNet-50-UNet Hybrid with Whale Optimization Algorithm for Accurate Liver Tumor Segmentation

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Accurate segmentation of liver and liver tumors from 3D medical images is crucial for liver cancer diagnosis and treatment planning.
• Challenges in segmentation arise from organs with similar characteristics, leading to difficulties in precise delineation.

Purpose of the Study:

• To develop a hybrid deep learning model for enhanced segmentation of liver and liver tumors.
• To optimize deep learning model hyperparameters using the Whale Optimization Algorithm (WOA) for improved segmentation performance.

Main Methods:

• A hybrid model integrating a U-Net based structure with the Whale Optimization Algorithm (WOA) was proposed.
• WOA was employed to fine-tune the hyperparameters of the LiTS-Res-UNet architecture for optimal deep learning model performance.

Main Results:

• The LiTS-Res-Unet + WOA hybrid model achieved high accuracy (99.54%), Dice coefficient (92.38%), and Jaccard index (86.73%) on a benchmark dataset.
• The proposed model outperformed existing state-of-the-art methods in liver tumor segmentation tasks.

Conclusions:

• The WOA-based adaptive search effectively identified optimal hyperparameters, enhancing deep learning model convergence and accuracy.
• The hybrid model demonstrated robust performance and clinical applicability for precise liver tumor segmentation.