MGMT methylation status is an important criterion in decision making process of GBM treatment. MGMT methylated tumors benefit from a drug called “Temozolomide” which may double survival time. Currently the only definitive diagnosis can be achieved by tissue analysis which is obtained by invasive procedures. Furthermore, it is an epigenetic property which implies spatial and temporal heterogeneity which may require additional surgeries during treatment. We proposed a classifier model which can make that diagnosis using MRI and deep learning. The model makes its prediction with confidence scores and gradient based explanations. We reached high accuracy scores in our experimental model and locally deployed our application in some of the hospitals to observe experimental results in real life.
ÇAĞAN KOSKA
16 years
| Stand | 25 |
| Project | Computing-06 |
| Country | Türkiye |
