A Deep Learning-Based Ensemble Method for Early Diagnosis of Alzheimer's Disease using MRI Images.

Recently, the early diagnosis of Alzheimer's disease has gained major attention due to the growing prevalence of the disease and the resulting costs imposed on individuals and society. The main objective of this study was to propose an ensemble method based on deep learning for the early diagnosis of AD using MRI images. The methodology of this study consisted of collecting the dataset, preprocessing, creating the individual and ensemble models, evaluating the models based on ADNI data, and validating the trained model based on the local dataset. The proposed method was an ensemble approach selected through a comparative analysis of various ensemble scenarios. Finally, the six best individual CNN-based classifiers were selected to combine and constitute the ensemble model. The evaluation showed an accuracy rate of 98.57, 96.37, 94.22, 99.83, 93.88, and 93.92 for NC/AD, NC/EMCI, EMCI/LMCI, LMCI/AD, four-way and three-way classification groups, respectively. The validation results on the local dataset revealed an accuracy of 88.46 for three-way classification. Our performance results were higher than most reviewed studies and comparable with others. Although comparative analysis showed superior results of ensemble methods against individual architectures, there were no significant differences among various ensemble approaches. The validation results revealed the low performance of individual models in practice. In contrast, the ensemble method showed promising results. However, further studies on various and larger datasets are required to validate the generalizability of the model.

Auto-Segmentation and Classification of Glioma Tumors with the Goals of Treatment Response Assessment Using Deep Learning Based on Magnetic Resonance Imaging.

Glioma is the most common primary intracranial neoplasm in adults. Radiotherapy is a treatment approach in glioma patients, and Magnetic Resonance Imaging (MRI) is a beneficial diagnostic tool in treatment planning. Treatment response assessment in glioma patients is usually based on the Response Assessment in Neuro Oncology (RANO) criteria. The limitation of assessment based on RANO is two-dimensional (2D) manual measurements. Deep learning (DL) has great potential in neuro-oncology to improve the accuracy of response assessment. In the current research, firstly, the BraTS 2018 Challenge dataset included 210 HGG and 75 LGG were applied to train a designed U-Net network for automatic tumor and intra-tumoral segmentation, followed by training of the designed classifier with transfer learning for determining grading HGG and LGG. Then, designed networks were employed for the segmentation and classification of local MRI images of 49 glioma patients pre and post-radiotherapy. The results of tumor segmentation and its intra-tumoral regions were utilized to determine the volume of different regions and treatment response assessment. Treatment response assessment demonstrated that radiotherapy is effective on the whole tumor and enhancing region with p-value ≤ 0.05 with a 95% confidence level, while it did not affect necrosis and peri-tumoral edema regions. This work demonstrated the potential of using deep learning in MRI images to provide a beneficial tool in the automated treatment response assessment so that the patient can obtain the best treatment.

Early diagnosis of Alzheimer's disease based on deep learning: A systematic review.

BACKGROUND: The improvement of health indicators and life expectancy, especially in developed countries, has led to population growth and increased age-related diseases, including Alzheimer's disease (AD). Thus, the early detection of AD is valuable to stop its progress at an early stage. METHOD: This study systematically reviewed the current state of using deep learning methods on neuroimaging data for timely diagnose of AD. We reviewed different deep models, modalities, feature extraction strategies, and parameter initialization methods to find out which model or strategy could offer better performance. RESULTS: Our search in eight different databases resulted in 736 studies, from which 74 studies were included to be reviewed for data analysis. Most studies have reported the normal control (NC)/AD classification and have shown desirable results. Although recent studies showed promising results of utilizing deep models on the NC/mild cognitive impairment (MCI) and NC/early MCI (eMCI), other classification groups should be taken into consideration and improved. DISCUSSION: The results of our review indicate that the comparative analysis is challenging in this area due to the lack of a benchmark platform; however, convolutional neural network (CNN)-based models, especially in an ensemble way, seem to perform better than other deep models. The transfer learning approach also could efficiently improve the performance and time complexity. Further research on designing a benchmark platform to facilitate the comparative analysis is recommended.

Development and use of a clinical decision support system for the diagnosis of social anxiety disorder.

BACKGROUND: Mental disorders, according to the definition of World Health Organization, consist of a wide range of signs, which are generally specified by a combination of unusual thoughts, feelings, behavior, and relationships with others. Social anxiety disorder (SAD) is one of the most prevalent mental disorders, described as permanent and severe fear or feeling of embarrassment in social situations. Considering the imprecise nature of SAD symptoms, the main objective of this study was to generate an intelligent decision support system for SAD diagnosis, using Adaptive neuro-fuzzy inference system (ANFIS) technique and to conduct an evaluation method, using sensitivity, specificity and accuracy metrics. METHOD: In this study, a real-world dataset with the sample size of 214 was selected and used to generate the model. The method comprised a multi-stage procedure named preprocessing, classification, and evaluation. The preprocessing stage, itself, consists of three steps called normalization, feature selection, and anomaly detection, using the Self-Organizing Map (SOM) clustering method. The ANFIS technique with 5-fold cross-validation was used for the classification of social anxiety disorder. RESULTS AND CONCLUSION: The preprocessed dataset with seven input features were used to train the ANFIS model. The hybrid optimization learning algorithm and 41 epochs were used as optimal learning parameters. The accuracy, sensitivity, and specificity metrics were reported 98.67%, 97.14%, and 100%, respectively. The results revealed that the proposed model was quite appropriate for SAD diagnosis and in line with findings of other studies. Further research study addressing the design of a decision support system for diagnosing the severity of SAD is recommended.