3D brain tumor segmentation based on a novel nettree merging

Area of Science:

• Medical Imaging
• Computational Biology
• Radiology

Background:

• Brain tumors pose significant health risks, impacting neurological function and patient well-being.
• Accurate segmentation of 3D brain tumors is crucial for treatment planning and understanding tumor- GTV relationships.
• Existing 3D segmentation methods face limitations with clinical data and direct 3D image manipulation.

Purpose of the Study:

• To develop and validate a novel 3D brain tumor segmentation method.
• To address the limitations of current 3D segmentation techniques, particularly on clinical datasets.
• To accurately segment the Gross Tumor Volume (GTV) in 3D MRI brain tumor images.

Main Methods:

• A 3D brain tumor segmentation approach based on nettree merging is proposed.
• The method utilizes topological determination to establish relationships between body targets.
• A nettree structure is constructed based on pathological topology and merged according to structure and intensity.

Main Results:

• The proposed method achieved Dice scores of approximately 0.824 on a clinical dataset and 0.873 on the BraTS dataset.
• HD95, Precision, and Recall metrics demonstrated strong performance on both datasets.
• Experimental validation on public and clinical datasets confirmed the method's effectiveness.

Conclusions:

• The nettree merging method offers an effective solution for 3D brain tumor segmentation.
• The approach shows promise for improving the accuracy and reliability of GTV segmentation in clinical practice.
• This technique provides a valuable tool for analyzing brain tumor characteristics in 3D MRI data.