Brain tumor segmentation using U-Net in conjunction with EfficientNet.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Cancer is the leading cause of death, with brain tumors like pleomorphic glioblastoma posing diagnostic challenges due to unclear boundaries.
• Accurate brain tumor segmentation is crucial for surgical planning to avoid damaging critical neural structures.
• Deep learning (DL) and artificial intelligence (AI) show promise in medical image analysis, particularly for image segmentation tasks.

Purpose of the Study:

• To evaluate the effectiveness of integrating EfficientNetV2 as an encoder within the U-Net architecture for brain tumor segmentation.
• To improve the accuracy and efficiency of AI-assisted brain tumor detection and measurement.

Main Methods:

• The study employed the U-Net convolutional neural network architecture, a standard for image segmentation.
• EfficientNetV2 was integrated as the encoder component of the U-Net model.
• The performance of the combined model was evaluated using metrics such as loss, accuracy, and the Dice similarity coefficient (DSC).

Main Results:

• The proposed model, U-Net with EfficientNetV2 encoder, achieved a high accuracy of 0.9977.
• The model demonstrated a Dice similarity coefficient (DSC) of 0.9133, indicating robust segmentation performance.
• The integration resulted in improved segmentation accuracy compared to standard U-Net implementations.

Conclusions:

• Combining EfficientNetV2 with U-Net significantly enhances brain tumor image segmentation performance.
• This AI-driven approach offers a valuable tool for clinicians, potentially reducing diagnostic time and improving surgical outcomes.
• The study highlights the potential of advanced deep learning models in neuro-oncology for precise tumor delineation.