Click here to

Session: Imaging for Treatment Planning [Return to Session]

Quantifying the Accuracy of T1- and T2-Weighted MRI for Detecting Whole Brain Tumor Burden

P Chen*, D Healey, S Mehta, C Quarles, M Scarpelli, Purdue University, West Lafayette, IN


TU-A-202-6 (Tuesday, 7/12/2022) 7:30 AM - 8:30 AM [Eastern Time (GMT-4)]

Room 202

Purpose: Conventional anatomic MRI scans are the most commonly used radiographic methods for brain tumor localization, treatment planning, and response assessment. This study rigorously evaluated these MRI methods by performing a spatial comparison between the MRI-defined brain tumor burden and the ‘true’ brain tumor burden derived from ex vivo fluorescence images of optically cleared brain tissue.

Methods: Six glioblastoma patient-derived xenograft (PDX) models were implanted orthotopically in nude rats (N=13 animals total). All tumor cells were transduced with tdTomato, enabling fluorescent imaging. Once tumors were established in vivo, T1-weighted and T2-weighted MRI scans were acquired, including with/without gadolinium to assess contrast enhancement. Next, brains were excised, optically cleared, and scanned ex vivo with fluorescence imaging. A previously developed method was used to register the ex vivo fluorescent images with the in vivo MR images. MRI-derived tumor burden was manually segmented using all MR images as a reference. The ‘true’ tumor burden was defined by manually segmenting the fluorescently labelled tumor regions on the ex vivo images.

Results: The MRI-defined tumor burden (median volume 63mm³) was smaller than the ‘true’ tumor burden (median volume 86mm³). The sensitivity of MRI for detecting ‘true’ tumor burden was lower for non-enhancing tumors (median 52%) than gadolinium enhancing tumors (median 74%; P=0.05). Similarly, the median dice coefficient between MRI-defined tumor burden and ‘true’ tumor burden was lower for non-enhancing tumors (median 0.58) than enhancing tumors (median 0.77; P=0.01). The MRI performed the worst for the GB7 PDX, where non-enhancing tumor invasion of the contralateral brain hemisphere went undetected on MRI.

Conclusion: In these glioblastoma PDXs, the performance of MRI was significantly worse in non-enhancing tumors than in gadolinium enhancing tumors, with approximately half the tumor going undetected in most non-enhancing tumors. These results support development of improved imaging methods for assessing non-enhancing brain tumors.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email