High-Resolution Ultrasound Data for AI-Based Segmentation in Mouse Brain Tumor

Area of Science:

• Neuro-oncology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Glioblastoma multiforme (GBM) is an aggressive brain cancer requiring effective treatments.
• Preclinical mouse models, like the GL261 glioma model, are crucial for studying GBM and developing therapies.
• Ultrasound imaging offers non-invasive, real-time monitoring and therapeutic potential in preclinical research.

Purpose of the Study:

• To address the challenges in real-time segmentation of GL261 tumors during preclinical surgical procedures.
• To develop a high-quality, annotated ultrasound dataset to advance AI-driven tumor segmentation.
• To support the development of more accurate and efficient tumor resection techniques in preclinical GBM research.

Main Methods:

• Creation of the first publicly available ultrasound dataset specifically for GL261 glioblastomas.
• Annotation of 1,856 ultrasound images to facilitate precise tumor boundary delineation.
• Dataset designed to support the training and validation of AI models for automated segmentation.

Main Results:

• Introduction of a novel, comprehensive ultrasound dataset for GL261 glioblastoma research.
• The dataset provides 1,856 annotated images, enabling robust AI model development.
• This resource is expected to significantly improve the accuracy and efficiency of automated tumor segmentation.

Conclusions:

• The developed dataset is a foundational resource for advancing AI in preclinical GBM research.
• It facilitates the development of improved surgical guidance and tumor resection techniques.
• Bridging preclinical findings with clinical practice, this dataset aids in creating more effective GBM therapies.