Wednesday, December 2 | 9:40 a.m.-9:50 a.m. | W3-SSNR10-2 | Room E350
Multimodal, quantitative MRI used with 2.5D deep-learning (DL) could offer a noninvasive decision-support tool for preoperative molecular characterization, prognosis estimation, and individualized treatment planning for adult diffuse gliomas, a research team from China has found.
In this session, presenter Xin Ge, PhD, of Air Force Medical University in Xi'an, will share results from a study that included 1,008 adults with histologically confirmed diffuse gliomas. The researchers used quantitative MRI to generate six co-registered maps -- apparent diffusion coefficient, cerebral blood flow, T1, T2, proton density, and contrast-enhanced T1 map -- and assigned participants to a training cohort, an internal validation group, and three external test cohorts. They also developed five model families (VOI-mean, radiomics, and 2D, 2.5D, and 3D DenseNet-201 DL).
One of the single-map models achieved the highest external-test area under the receiver operating curve, outperforming four of the five model families for presurgical prediction of isocitrate dehydrogenase mutation -- a diagnostic marker for glioma.
Check out this session to learn more.
