Convolutional neural networks for automatic tuber segmentation and quantification of tuber burden in tuberous sclerosis complex

Abstract

Objective: This study was undertaken to develop a fully automated algorithm for tuber segmentation and quantification of tuber volume that performs similarly to the gold standard human neuroradiologist.

Methods: We used brain magnetic resonance imaging (MRI) from patients with tuberous sclerosis complex (TSC) to train and validate a convolutional neural network (CNN), which was evaluated on segmentation with the Dice-Sørensen similarity coefficient (DSSC) and on tuber burden quantification with Spearman correlation coefficient against a neuroradiologist's gold standard in the test set.

Results: We collected 263 MRIs from 196 patients (57% males) with median (25th percentile-75th percentile) age of 4.3 (3.0-10.1) years: 176 MRIs in the train set, 39 in the validation set, and 48 in the test set. The final model achieved in the test set a DSSC of .820 (95% confidence interval [CI] = .799-.840) in the whole brain and in the different lobes the following: .831 (95% CI = .804-.850) in left frontal, .827 (95% CI = .799-.853) in right frontal, .817 (95% CI = .779-.842) in left temporal, .834 (95% CI = .812-.849) in right temporal, .821 (95% CI = .783-.856) in left parietal, .840 (95% CI = .810-.865) in right parietal, .832 (95% CI = .808-.851) in left occipital, and .856 (95% CI = .838-.871) in right occipital. CNN tuber volume quantification nearly perfectly correlated (Spearman correlation coefficient) with the neuroradiologist's across the whole brain (.984, 95% CI = .971-.991) and in the different lobes: .966 (95% CI = .940-.981) in left frontal, .973 (95% CI = .952-.985) in right frontal, .936 (95% CI = .888-.964) in left temporal, .967 (95% CI = .942-.982) in right temporal, .989 (95% CI = .980-.994) in left parietal, .983 (95% CI = .970-.990) in right parietal, .992 (95% CI = .985-.995) in left occipital, and .982 (95% CI = .968-.990) in right occipital (all p < .00001).

Significance: We generated, trained, validated, and made publicly available a CNN that achieves a near-perfect correlation with a neuroradiologist gold standard quantification of tuber burden, allows for objective tuber segmentation, and increases rigor and reproducibility in TSC research across institutions.

Keywords: convolutional neural network; deep learning; neuroradiology; tuber burden; tuberous sclerosis complex.

Figure 1.. The Dice-Sørensen similarity coefficient (DSSC) measures the spatial overlap between the gold-standard segmentation (red, first column) and the predicted segmentation (green, second column).

A. No overlap between gold-standard and predicted segmentation: DSSC = 0. B. Partial overlap between gold-standard and predicted segmentation: 0 < DSSC < 1. C. Complete overlap between gold-standard and predicted segmentation: DSSC = 1.

Figure 2.. Schematic representation of TSCCNN3D_dropout, the CNN architecture with the highest Dice-Sørensen similarity coefficient in the validation set, and, therefore, the architecture that was selected in the final model.

CNN: Convolutional neural network. ReLU: Rectified linear unit. 3D: three-dimensional

Figure 3.. Examples of the TSCNN3D segmentation results in 64×64×64×3 blocks.

On visual inspection, the segmentations predicted by the CNN appear quite similar to the gold-standard manual segmentation and appear to represent the tuber tissue well. Each example (A, B, and C) provides the T1-weighted, T2-weighted, and FLAIR, the gold-standard manual neuroradiologist segmentation, the TSCCNN3D_dropout segmentation, and the comparison of the gold-standard and TSCCNN3D_dropout segmentation (correct segmentation in green: overlap between gold-standard and TSCCNN3D_dropout segmentations, under-segmentation in red: areas of gold-standard segmentation that the TSCCNN3D_dropout did not segment, and over-segmentation in blue: areas not segmented in the gold-standard segmentation that the TSCCNN3D_dropout did segment).

Figure 4.. Examples of the TSCNN3D segmentation results in the 256×256×256×3 full MRIs.

On visual inspection, the segmentations predicted by the CNN appear quite similar to the gold-standard manual segmentation and appear to represent the tuber tissue well. Each example (A, B, and C) provides the T1-weighted, T2-weighted, FLAIR, the gold-standard manual neuroradiologist segmentation, the TSCCNN3D_dropout segmentation, and the comparison of the gold-standard and TSCCNN3D_dropout segmentation (correct segmentation in green: overlap between gold-standard and TSCCNN3D_dropout segmentations, undersegmentation in red: areas of gold-standard segmentation that the TSCCNN3D_dropout did not segment, and oversegmentation in blue: areas not segmented in the gold-standard segmentation that the TSCCNN3D_dropout did segment).

Figure 5.. Correlation between the tuber volume based on the gold-standard manual segmentation and the tuber volume based on the TSCCNN3D_dropout segmentation.

Each green dot represents a patient. The dashed diagonal line represents perfect correlation between predicted and gold-standard tuber volume. A. Whole brain. Spearman correlation coefficient: 0.984 (95% confidence interval: 0.971–0.991), p-value <0.00001. B. Gray matter. Spearman correlation coefficient: 0.988 (95% confidence interval: 0.978–0.993), p-value <0.00001.