Automatic Assessment of ASPECTS Using Diffusion-Weighted Imaging in Acute Ischemic Stroke Using Recurrent Residual Convolutional Neural Network

Abstract

The early detection and rapid quantification of acute ischemic lesions play pivotal roles in stroke management. We developed a deep learning algorithm for the automatic binary classification of the Alberta Stroke Program Early Computed Tomographic Score (ASPECTS) using diffusion-weighted imaging (DWI) in acute stroke patients. Three hundred and ninety DWI datasets with acute anterior circulation stroke were included. A classifier algorithm utilizing a recurrent residual convolutional neural network (RRCNN) was developed for classification between low (1–6) and high (7–10) DWI-ASPECTS groups. The model performance was compared with a pre-trained VGG16, Inception V3, and a 3D convolutional neural network (3DCNN). The proposed RRCNN model demonstrated higher performance than the pre-trained models and 3DCNN with an accuracy of 87.3%, AUC of 0.941, and F1-score of 0.888 for classification between the low and high DWI-ASPECTS groups. These results suggest that the deep learning algorithm developed in this study can provide a rapid assessment of DWI-ASPECTS and may serve as an ancillary tool that can assist physicians in making urgent clinical decisions.

Quad Convolutional Layers (QCL) CNN Approach for Classification of Brain Stroke in Diffusion Weighted (DW) - Magnetic Resonance Images (MRI)

“…The proposed method is also compared to Do et al [31] research, which similarly classified stroke images. In stroke care, the proposed early diagnosis and rapid quantification of acute ischemic lesions are critical.…”

“…The results of these experiments are written in Table 6. Do et al [31] adopted the VGG16 and Resnet, employing 12 convolution layers, and the accuracy result shows 92.31 percent. As for Zhu, H et al [32], which employs seven convolution layers, the accuracy result is 91.78 percent.…”

Deep Learning Applications for Acute Stroke Management

“…36 To perform early detection and rapid quantification of acute ischemic lesions using magnetic resonance (MR) images, Do et al (2020) developed a classifier algorithm using a recurrent residual convolutional neural network (RRCNN) to distinguish between diffusionweighted imaging (DWI) MRI slices belonging to low (1-6) and high (7-10) DWI-ASPECTS groups. 39 An RRCNN contains a residual unit, which allows the network to train deep architectures by incorporating skip connections, or shortcuts to surpass certain layers; in theory, the recurrent residual convolutional layers permit feature accumulation on temporal tasks, which should improve performance on segmentation tasks. 55 These CNNs benefit from the reduction in imaging parameters as a result of data preprocessing and normalization, maximizing image contrast and permitting the models to fully benefit from multi-contrast MRI 39,40 in the detection and diagnosis of AIS.…”

“…DL models have been applied to facilitate surveillance and analysis of acute stroke neuroimages, offering solutions for lesion segmentation and quantification, 11,17,[22][23][24][25][26][27][28][29][30][31][32][33][34] early stroke detection, 18,[35][36][37][38][39][40][41][42][43][44][45][46] selection of candidates for therapeutic intervention, 9,[47][48][49] and prediction of short-and long-term functional outcomes. 9,[47][48][49][50] Existing automated applications for clinical settings, such as Viz.ai and RapidAI, have been developed for a variety of tasks, including identifying large vessel occlusions (LVOs), diagnosing ischemic and hemorrhagic stroke, and assessing salvageable brain tissue.…”

Deep Learning for Prediction of Mechanism in Acute Ischemic Stroke Using Brain MRI

“…The diffusion imaging lesion pattern, which provides useful information for early diagnosis of acute ischemic stroke, has been reported to be closely related to the stroke subtype 8,9 . To diagnose acute ischemic stroke in brain MRI images, various deep learning algorithms based on convolutional neural networks (CNNs) have been proposed [10][11][12][13][14][15][16][17][18][19][20][21] . These studies have shown that deep learning can detect stroke lesions more accurately than traditional machine learning techniques and can extract meaningful features for severity evaluation or prognosis prediction.…”