Predicting the brain age of children with cerebral palsy using a two-dimensional convolutional neural networks prediction model without gray and white matter segmentation

Abstract

Background: Abnormal brain development is common in children with cerebral palsy (CP), but there are no recent reports on the actual brain age of children with CP.

Objective: Our objective is to use the brain age prediction model to explore the law of brain development in children with CP.

Methods: A two-dimensional convolutional neural networks brain age prediction model was designed without segmenting the white and gray matter. Training and testing brain age prediction model using magnetic resonance images of healthy people in a public database. The brain age of children with CP aged 5-27 years old was predicted.

Results: The training dataset mean absolute error (MAE) = 1.85, r = 0.99; test dataset MAE = 3.98, r = 0.95. The brain age gap estimation (BrainAGE) of the 5- to 27-year-old patients with CP was generally higher than that of healthy peers (p < 0.0001). The BrainAGE of male patients with CP was higher than that of female patients (p < 0.05). The BrainAGE of patients with bilateral spastic CP was higher than those with unilateral spastic CP (p < 0.05).

Conclusion: A two-dimensional convolutional neural networks brain age prediction model allows for brain age prediction using routine hospital T1-weighted head MRI without segmenting the white and gray matter of the brain. At the same time, these findings suggest that brain aging occurs in patients with CP after brain damage. Female patients with CP are more likely to return to their original brain development trajectory than male patients after brain injury. In patients with spastic CP, brain aging is more serious in those with bilateral cerebral hemisphere injury than in those with unilateral cerebral hemisphere injury.

Keywords: brain age; brain age gap estimation; cerebral palsy; convolutional neural networks; deep learning.

Figure 1

Model schematic. At the beginning of the model, a 2D convolution layer composed of three convolution kernels was used for feature extraction, followed by a skipping layer composed of nine skipping blocks to capture deeper features. Features were learned again through three groups of convolution layers designed by reference GoogLeNET, and the same skipping layer was used to capture deeper features. Finally, the brain age was regressed through the full connection layer.

Figure 2

Sample size distribution before and after stratified sampling. (A) The sample distribution before stratified sampling, and (B) the sample distribution after stratified sampling.

Figure 3

The brain age prediction model was trained 40 times. The MAE of the training and test sets decreased gradually and tended to be stable with increased training times.

Figure 4

Performance of brain age prediction model. (A) The physiological age of the training set and its predicted brain age (r = 0.99, p < 0.0001). (B) The physiological age of the test set and its predicted brain age (r = 0.95, p < 0.05).

Figure 5

MRI of an 11-year-old patient with bilateral spastic cerebral palsy vs. an 11- and 62-year-old healthy person. As shown in (A,B) the head MRI of an 11-year-old patient with bilateral spastic cerebral palsy, who has significant brain atrophy and the patient's brain age is 61.74 years; (C,D) the head MRI of an 11-year-old healthy person; (E,F) the head MRI of a 62-year-old healthy elderly person. Based on the pictures, we can see that the 11-year-old bilateral spastic cerebral palsy patient's degree of brain atrophy is relatively close to that of a 62-year-old healthy elderly person.

Figure 6

Physiological age and brain age of patients with cerebral palsy. As shown in (A), the data represented by the circle icon are the physiological age of the CP patient, and the triangle icon represents the corresponding actual brain age. As shown in (B), the brain age of patients with CP is older than the physiological age. As shown in (A), the data on the left are the physiological age of patients with CP, and the data on the right are the corresponding actual brain age. The brain age of patients with CP was greater than their physiological age.

Figure 7

Data analysis of brain age difference in patients with cerebral palsy. In (A) the BrainAGE of patients with CP is higher than that of healthy peers (p < 0.0001). In (B) the brain age difference of men is higher than that of women in patients with mixed cerebral palsy (p < 0.05). In (C) the brain age difference of female patients with spastic CP was higher than that of patients with mixed CP (p < 0.01). Furthermore, the BrainAGE of patients with bilateral spastic CP was higher than that of patients with unilateral spastic cerebral palsy (p < 0.05), as shown in (D). In (E) the BrainAGE of patients with non-asphyxia was higher than that of patients with asphyxia in female mixed CP (p = 0.12).

Figure 8

Relationship between BrainAGE difference and physiological age in patients with cerebral palsy. As shown in the figure, the BrainAGE of patients with CP negatively correlates with their physiological age (r = −0.14, p < 0.01).