Investigating the potential of segmented echo-planar imaging-based susceptibility-weighted imaging for glioma grading

Susceptibility-weighted imaging (SWI) has shown potential in glioma grading using features like intratumoral susceptibility signals (ITSS). Segmented echo-planar imaging-based SWI (SWI-EPI) sequence offers good spatial coverage and a high signal-to-noise ratio.

To evaluate segmented echo-planar imaging-based susceptibility-weighted imaging (SWI-EPI) for glioma grading utilizing ITSS volume (ITV) and a proposed new metric, normalized ITV (nITV), calculated as the ratio of ITV to total tumor volume, to improve grading accuracy.

ITSS were segmented from the SWI-EPI sequence. ITV was computed and normalized by whole tumor volume to derive nITV. Statistical analysis included Shapiro-Wilk, Mann-Whitney, Kruskal-Wallis, and receiver-operating-characteristic (ROC) curve analyses to assess the effectiveness of both the metrics in glioma grading.

The proposed nITV metric significantly enhanced glioma grading accuracy. The area under the ROC curve (AUC) for nITV was consistently higher than for ITV when differentiating between grades 2 vs. 3 (0.886 vs. 0.818), 2 vs. 4 (0.969 vs. 0.939), and 3 vs. 4 (0.832 vs. 0.788). For differentiating low- vs. high-grade gliomas, nITV achieved an AUC of 0.95. Confusion matrices confirmed nITV led to fewer misclassifications.

ITSS quantification from SWI-EPI is effective for tumor evaluation. The proposed nITV metric provides superior glioma grading efficiency compared to ITV.