A subregion-based RadioFusionOmics model discriminates between grade 4 astrocytoma and glioblastoma on multisequence MRI
Ruili Wei, Songlin Lu, Shengsheng Lai, Fangrong Liang, Wanli Zhang, Xinqing Jiang, Xin Zhen, Ruimeng Yang
Abstract
Abstract Purpose To explore a subregion-based RadioFusionOmics (RFO) model for discrimination between adult-type grade 4 astrocytoma and glioblastoma according to the 2021 WHO CNS5 classification. Methods 329 patients (40 grade 4 astrocytomas and 289 glioblastomas) with histologic diagnosis was retrospectively collected from our local institution and The Cancer Imaging Archive (TCIA). The volumes of interests (VOIs) were obtained from four multiparametric MRI sequences (T 1 WI, T 1 WI + C, T 2 WI, T 2 -FLAIR) using (1) manual segmentation of the non-enhanced tumor (nET), enhanced tumor (ET), and peritumoral edema (pTE), and (2) K -means clustering of four habitats (H 1 : high T 1 WI + C, high T 2 -FLAIR; (2) H 2 : high T 1 WI + C, low T 2 -FLAIR; (3) H 3 : low T 1 WI + C, high T 2 -FLAIR; and (4) H 4 : low T 1 WI + C, low T 2 -FLAIR). The optimal VOI and best MRI sequence combination were determined. The performance of the RFO model was evaluated using the area under the precision-recall curve (AUPRC) and the best signatures were identified. Results The two best VOIs were manual VOI 3 (putative peritumoral edema) and clustering H 34 (low T 1 WI + C, high T 2 -FLAIR (H 3 ) combined with low T 1 WI + C and low T 2 -FLAIR (H 4 )). Features fused from four MRI sequences ( $${F}_{seq}^{\mathrm{1,2},\mathrm{3,4}}$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msubsup> <mml:mi>F</mml:mi> <mml:mrow> <mml:mi>seq</mml:mi> </mml:mrow> <mml:mrow> <mml:mrow> <mml:mn>1</mml:mn> <mml:mo>,</mml:mo> <mml:mn>2</mml:mn> </mml:mrow> <mml:mo>,</mml:mo> <mml:mrow> <mml:mn>3</mml:mn> <mml:mo>,</mml:mo> <mml:mn>4</mml:mn> </mml:mrow> </mml:mrow> </mml:msubsup> </mml:math> ) outperformed those from either a single sequence or other sequence combinations. The RFO model that was trained using fused features $${F}_{seq}^{\mathrm{1,2},\mathrm{3,4}}$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msubsup> <mml:mi>F</mml:mi> <mml:mrow> <mml:mi>seq</mml:mi> </mml:mrow> <mml:mrow> <mml:mrow> <mml:mn>1</mml:mn> <mml:mo>,</mml:mo> <mml:mn>2</mml:mn> </mml:mrow> <mml:mo>,</mml:mo> <mml:mrow> <mml:mn>3</mml:mn> <mml:mo>,</mml:mo> <mml:mn>4</mml:mn> </mml:mrow> </mml:mrow> </mml:msubsup> </mml:math> achieved the AUPRC of 0.972 (VOI 3 ) and 0.976 (H 34 ) in the primary cohort ( p = 0.905), and 0.971 (VOI 3 ) and 0.974 (H 34 ) in the testing cohort ( p = 0.402). Conclusion The performance of subregions defined by clustering was comparable to that of subregions that were manually defined. Fusion of features from the edematous subregions of multiple MRI sequences by the RFO model resulted in differentiation between grade 4 astrocytoma and glioblastoma.