Alcoholic brain injury via 8-layer deep convolutionalneural network

Abstract

Alcohol can act quickly in the human body and alter mood and behavior. If people drink too much alcohol,it will accumulate in the liver and brain. To a certain extent, the symptoms of alcoholism will appear. So far, the main method of diagnosis of alcoholic brain injury is through MRI images by radiologists. However,this is a very subjective diagnosis. Radiologists could be influenced by external factors, resulting in diagnostic errors, such as physical discomfort, inattention, lack of rest, etc.. In the paper, we introduced a 8-layer deep convolutional neural network structure for alcoholic brain injury detection. Three fully connected layers, five pooling layers, and five convolution layers formed the proposed neural network structure. We proposed three improvements in this paper, (i) An automatic diagnosis method of alcoholic brain injury based on deep learning was proposed; (ii) We introduced Dropout to the proposed structure to improve robustness; (iii) Compared with other seven advanced approaches, the proposed 8-layer deep convolutional neural network structure is more efficient. The experimental results demonstrated that the specificity, sensitivity, accuracy, precision, FMI MCC and F1 were 96.20 ± 1.47, 96.14 ± 1.99, 96.17 ± 1.55,95.98 ± 1.54, 96.06 ± 1.62 93.34 ± 3.11, 96.05 ± 1.62, respectively. Based on the comparison results,our method had the excellent performance. The proposed method can be used as one of the methods to detect alcoholic brain injury based on MRI images.