ReMIND: The Brain Resection Multimodal Imaging Database

Area of Science:

• Neurosurgery
• Medical Imaging
• Artificial Intelligence

Background:

• Maximal safe surgical resection is standard for brain tumors, aided by neuronavigation.
• Brain shift and poor tissue contrast (gliomas vs. healthy tissue) challenge surgical accuracy.
• Intraoperative MRI (iMRI) and ultrasound (iUS) aid visualization, with iUS offering faster integration but lower contrast than iMRI.

Purpose of the Study:

• To establish the largest publicly available database of intraoperative imaging for surgically treated brain tumors.
• To support research in brain shift correction and AI-based medical image analysis.
• To facilitate neurosurgical training in interpreting iUS and iMRI.

Main Methods:

• Compiled a database from 114 consecutive patients at a single institution.
• Included preoperative MRI, 3D intraoperative ultrasound (iUS), and intraoperative MRI (iMRI) series.
• Collected 356 segmentations alongside 369 preoperative MRI, 320 iUS, and 301 iMRI series.

Main Results:

• The database comprises data from 114 patients, including 92 gliomas, 11 metastases, and 11 other tumor types.
• Contains a significant volume of multimodal imaging data: 369 preoperative MRI, 320 3D iUS, and 301 iMRI series.
• Includes 356 corresponding segmentations for research and training.

Conclusions:

• The presented database is the largest public resource for brain tumor intraoperative imaging.
• It is expected to significantly advance AI research for brain shift and image analysis.
• The database will be invaluable for neurosurgical training and improving surgical outcomes.