MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Using Nosologic Imaging Algorithms to Quantitatively Standardize Grade Differentiation in Adult Primary Glioma

E Diller1,3*, R Lober2 , J Parker3 , (1) Purdue University, West Lafayette, IN, (2) Dayton Children's Hospital, Dayton, Ohio, (3) Indiana University School of Medicine, Indianapolis, IN

Presentations

(Sunday, 7/29/2018) 3:30 PM - 4:00 PM

Room: Exhibit Hall | Forum 9

Purpose: Nosologic imaging algorithms (NIAs) reduce large quantities of MR data to 3D maps of tumor type and grade. Frequently focused on MR spectroscopy, NIAs often fail to utilize routine clinical MR data. This work developed a multimodal NIA using five common clinical contrasts (ADC, FLAIR, T1, T1+, and T2) across high (GBM) and lower grade gliomas (LGG). We hypothesized that GBM would demonstrate an increased percentage of tumor necrotic voxels compared to LGG, and that survival time would have a positive correlation to the percentage of voxels classified as suspicious.

Methods: The NIA classified voxels based on a weighted k Nearest Neighbor machine-learning algorithm that trained on physician annotations of five abnormal tissue types (tumor, necrosis, edema, pure cyst, suspicious for tumor) across five MR contrasts (ADC, FLAIR, T1, T1+, T2) and predicted abnormal voxels as one of the five diseased classifications. Prognosis and MR data for 7 GBM and 10 LGG biopsy diagnosed patients were registered to the patients respective T1+ data and scaled to normal appearing white matter. The percent abnormal classification was analyzed in SPSS using a corrected multivariate general linear model (α=0.1) and correlated to prognosis by a two-tail bivariate correlation test and Cox regression(α=0.05).

Results: The GBM patients had a higher percent of their tumor marked necrotic (p=0.079) compared to LGG, while LGG had a greater percent marked suspicious compared to GBM (p=0.07). Additionally, Percent suspicious had a positive correlation with survival time (R=0.579, p=0.015).

Conclusion: For a small cohort, the NIA demonstrates promising results consistent with WHO diagnostic standards. These results provide preliminary support for our hypothesis and demonstrate that NIA may hold the potential to increase the accuracy, efficiency, and standardization of diagnosis in a clinical setting.

Keywords

Not Applicable / None Entered.

Taxonomy

Not Applicable / None Entered.

Contact Email