Sci-Sat AM(1): Imaging-06: Proximity-based modification to an automatic method for tumor delineation using MRSI

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Oncology

Background:

• The Choline-to-NAA Index (CNI) is a standard method for biologically identifying brain tumors using Magnetic Resonance Spectroscopic Imaging (MRSI).
• The original CNI method assumes minimal variation in Choline (Cho) to N-Acetylaspartate (NAA) ratios within normal brain tissue.
• Advancements in MRSI sequences reveal greater variability in Cho-to-NAA levels in normal tissue, challenging the specificity of the existing CNI method.

Purpose of the Study:

• To modify the Choline-to-NAA Index (CNI) to enhance its specificity in brain tumor delineation.
• To address the increased uncertainty in Cho-to-NAA levels within normal brain tissue observed with advanced MRSI techniques.

Main Methods:

• Introduced a modified CNI method that defines a high-certainty tumor volume and an adjacent uncertainty region.
• Segmented voxels within the uncertainty region as either tumor or normal tissue based on proximity to high-certainty tumor areas.
• Moved away from an arbitrary CNI threshold to a more nuanced approach for tumor boundary definition.

Main Results:

• Preliminary results indicate that the modified CNI method reduces the number of false positives compared to the original CNI method.
• The new approach effectively segments voxels in the uncertainty region, improving diagnostic accuracy.
• The proposed modification enhances the biological identification of brain tumors by accounting for tissue variability.

Conclusions:

• The modified CNI method offers improved specificity for brain tumor delineation in MRSI.
• This approach is essential for accurate tumor identification given the increasing detection of variability in normal brain tissue.
• The method shows promise in decreasing false-positive diagnoses in neuro-oncology.