Deep learning for glioblastoma segmentation using preoperative magnetic resonance imaging identifies volumetric features associated with survival

• Published in: Acta neurochirurgica 2020-12, Vol.162 (12), p.3067-3080
• Main Authors: Wan, Yizhou, Rahmat, Roushanak, Price, Stephen J.
• Format: Article
• Language: English
• Subjects: Aged; Automation; Brain cancer; Brain Neoplasms - diagnostic imaging; Brain Neoplasms - pathology; Brain Neoplasms - surgery; Brain tumors; Deep Learning; Female; Glioblastoma; Glioblastoma - diagnostic imaging; Glioblastoma - pathology; Glioblastoma - surgery; Humans; Image processing; Image Processing, Computer-Assisted - methods; Interventional Radiology; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Male; Medicine; Medicine & Public Health; Middle Aged; Minimally Invasive Surgery; Neural networks; Neural Networks, Computer; Neuroimaging; Neurology; Neuroradiology; Neurosurgery; Original - Brain Tumors; Original Article - Brain Tumors; Segmentation; Surgical Orthopedics; Survival; Treatment Outcome
• ISSN: 0001-6268; 0942-0940
• DOI: 10.1007/s00701-020-04483-7

Background Measurement of volumetric features is challenging in glioblastoma. We investigate whether volumetric features derived from preoperative MRI using a convolutional neural network–assisted segmentation is correlated with survival. Methods Preoperative MRI of 120 patients were scored using Visually Accessible Rembrandt Images (VASARI) features. We trained and tested a multilayer, multi-scale convolutional neural network on multimodal brain tumour segmentation challenge (BRATS) data, prior to testing on our dataset. The automated labels were manually edited to generate ground truth segmentations. Network performance for our data and BRATS data was compared. Multivariable Cox regression analysis corrected for multiple testing using the false discovery rate was performed to correlate clinical and imaging variables with overall survival. Results Median Dice coefficients in our sample were (1) whole tumour 0.94 (IQR, 0.82–0.98) compared to 0.91 (IQR, 0.83–0.94 p  = 0.012), (2) FLAIR region 0.84 (IQR, 0.63–0.95) compared to 0.81 (IQR, 0.69–0.8 p  = 0.170), (3) contrast-enhancing region 0.91 (IQR, 0.74–0.98) compared to 0.83 (IQR, 0.78–0.89 p  = 0.003) and (4) necrosis region were 0.82 (IQR, 0.47–0.97) compared to 0.67 (IQR, 0.42–0.81 p  = 0.005). Contrast-enhancing region/tumour core ratio (HR 4.73 [95% CI, 1.67–13.40], corrected p  = 0.017) and necrotic core/tumour core ratio (HR 8.13 [95% CI, 2.06–32.12], corrected p  = 0.011) were independently associated with overall survival. Conclusion Semi-automated segmentation of glioblastoma using a convolutional neural network trained on independent data is robust when applied to routine clinical data. The segmented volumes have prognostic significance.