Machine learning-based automatic estimation of cortical atrophy using brain computed tomography images

Abstract

Cortical atrophy is measured clinically according to established visual rating scales based on magnetic resonance imaging (MRI). Although brain MRI is the primary imaging marker for neurodegeneration, computed tomography (CT) is also widely used for the early detection and diagnosis of dementia. However, they are seldom investigated. Therefore, we developed a machine learning algorithm for the automatic estimation of cortical atrophy on brain CT. Brain CT images (259 Alzheimer's dementia and 55 cognitively normal subjects) were visually rated by three neurologists and used for training. We constructed an algorithm by combining the convolutional neural network and regularized logistic regression (RLR). Model performance was then compared with that of neurologists, and feature importance was measured. RLR provided fast and reliable automatic estimations of frontal atrophy (75.2% accuracy, 93.6% sensitivity, 67.2% specificity, and 0.87 area under the curve [AUC]), posterior atrophy (79.6% accuracy, 87.2% sensitivity, 75.9% specificity, and 0.88 AUC), right medial temporal atrophy (81.2% accuracy, 84.7% sensitivity, 79.6% specificity, and 0.88 AUC), and left medial temporal atrophy (77.7% accuracy, 91.1% sensitivity, 72.3% specificity, and 0.90 AUC). We concluded that RLR-based automatic estimation of brain CT provided a comprehensive rating of atrophy that can potentially support physicians in real clinical settings.

Figure 1

A scheme of framework. Tensors are indicated as boxes while arrows denote computational operations. Number of channers is indicated beneath each box. Input and output of this network are CT slice with Label slice pairs and segmented CT images (segCT) slice. The classification conducted by threefold cross-validation was performed by randomly assigned the subject into three subgroups. BET Brain extraction, ReRU rectified linear unit activation, segCT Segmented CT, RLR regularized logistic regression, FA frontal atrophy, PA Parietal atrophy, MTAR medial temporal atrophy, right, MTAL medial temporal atrophy, left, Pos positive, Neg negative.

Figure 2

Sample of segmentation result (A) Alzheimer’s dementia (B) Normal control. GCA global cortical atrophy, CSF cerebrospinal fluid. Green = white matter, blue = gray matter, red = CSF.

Figure 3

ROC Curve for binary classification of atrophy. FA frontal atrophy, PA Parietal atrophy, MTAR medial temporal atrophy, right, MTAL medial temporal atrophy, left.

Figure 4

Feature importance of visual rating scale. FA frontal atrophy; PA parietal atrophy; MTAR medial temporal atrophy, right; MTAL medial temporal atrophy, left; Ven2D (the area ratio of ventricle); GMWMR2D (the area ratio of the sum of gray matter [GM] and white matter [WM]); WMR2D (the area ratio of WM); GMR2D (the area ratio of GM); Ven3D (the volume ratio of the ventricle); GMWMR3D (the volume ratio of the sum of GM and WM); WMR3D (the volume ratio of WM); GMR3D (the volume ratio of GM).