Raidionics: an open software for pre- and postoperative central nervous system tumor segmentation and standardized...

Area of Science:

• Neuroimaging
• Medical Informatics
• Oncology

Background:

• Accurate analysis of magnetic resonance (MR) scans is crucial for central nervous system tumor management.
• Current lack of standardized, open-source tools for automatic tumor segmentation and reporting introduces subjectivity and potential risks in clinical decisions.

Purpose of the Study:

• To introduce Raidionics, an open-source software designed for automated brain tumor segmentation and clinical report generation.
• To provide a user-friendly solution for analyzing various central nervous system tumor types using state-of-the-art segmentation models.

Main Methods:

• Development of Raidionics software with a graphical user interface and processing backend.
• Inclusion of preoperative segmentation models for glioblastomas, lower grade gliomas, meningiomas, and metastases.
• Inclusion of an early postoperative glioblastoma segmentation model.

Main Results:

• Preoperative segmentation achieved high performance across tumor types (average Dice ~85%, recall/precision ~95%).
• Postoperative segmentation performance was lower (average Dice ~41%).
• Standardized clinical report generation, including segmentation and feature computation, takes approximately ten minutes on a standard laptop.

Conclusions:

• Raidionics is the first open-source solution for easy use of advanced segmentation models for major brain tumor types.
• The software facilitates preoperative and postsurgical standardized reporting, improving objectivity in patient care.
• Raidionics addresses the need for reliable, automated tools in neuro-oncology for improved prognosis estimation and treatment planning.