Noninvasive Glioma Grading with Deep Learning: A Pilot Study

Gliomas are the most common neuroepithelial brain tumors, different by various biological tissue types and prognosis. They could be graded with four levels according to the 2007 WHO classification. The emergence of non-invasive histological and molecular diagnostics for nervous system neoplasms can revolutionize the efficacy and safety of medical care and radically reduce healthcare costs. Our pilot study aimed to evaluate the diagnostic accuracy of deep learning (DL) in subtyping gliomas by WHO grades (I-IV) based on preoperative magnetic resonance imaging (MRI) from Burdenko Neurosurgery Center's database. A total of 707 MRI studies was included. A "3D classification" approach predicting tumor type for the entire patient's MRI data showed the best result (accuracy = 83%, ROC AUC = 0.95), consistent with that of other authors who used different methodologies. Our preliminary results proved the separability of MR T1 axial images with contrast enhancement by WHO grade using DL.

Preoperative Grading of Astrocytic Supratentorial Brain Tumors with Diffusion-Weighted Magnetic Resonance Imaging and Apparent Diffusion Coefficient

Background

Diffusion-weighted imaging (DWI) is a form of magnetic resonance imaging (MRI) based on measuring the random Brownian motion of water molecules within the biological tissues and is particularly useful in tumor characterization.

Objectives

The purpose of this study was to evaluate the diagnostic value of DW MRI and the apparent diffusion coefficient (ADC) for preoperative grading of astrocytic supratentorial brain tumors.

Patients and Methods

Twenty-three patients (14 females, 9 males, mean age 43 years) with astrocytic supratentorial brain tumors underwent preoperative conventional MR imaging and DW MRI. The minimum, maximum and mean ADC values and the minimum, maximum and mean DWI signal intensities of each tumor were taken by placing several regions of interest in the tumor on DWI images and ADC maps. To assess the relationship between these values and the tumor grade, we used the Mann-Whitney U test and the Spearman correlation. Receiver operating characteristic (ROC) analysis was used to determine the cutoff value of the minimum, maximum and mean ADC values and the minimum, maximum and mean DWI signal intensities that had the best composition of sensitivity and specificity for differentiating low-grade and high-grade astrocytic brain tumors.

Results

According to the pathology reports, 10 patients had low-grade astrocytomas (grades I, II) and 13 patients had high-grade astrocytomas (grades III, IV). The minimum ADC value showed a significantly inverse correlation with astrocytic tumor grade (P = 0.006). The correlation between the maximum ADC value and the maximum DWI signal intensity with tumor grade was direct (P = 0.013, P = 0.035). According to the ROC analysis, the cutoff values of 0.843 × 10^-3 mm²/s, 2.117 × 10^-3 mm²/s and 165.2 for the minimum ADC, maximum ADC and maximum DWI respectively, obtained the best combination of sensitivity and specificity for distinguishing low-grade and high-grade astrocytomas.

Conclusion

Measuring minimum ADC, maximum ADC and maximum DWI signal intensity can provide valuable information for grading of astrocytic supratentorial brain tumors before surgery.