Search

Background: In recent years, neuroimaging with deep learning (DL) algorithms have made remarkable advances in the diagnosis of neurodegenerative disorders. However, applying DL in different medical domains is usually challenged by lack of labeled data. To address this challenge, transfer learning (TL) has been applied to use state-of-the-art convolution neural networks pre-trained on natural images. Yet, there are differences in characteristics between medical and natural images, also image classification and targeted medical diagnosis tasks. The purpose of this study is to investigate the performance of specialized and TL in the classification of neurodegenerative disorders using 3D volumes of 18F-FDG-PET brain scans. Results: Results show that TL models are suboptimal for classification of neurodegenerative disorders, especially when the objective is to separate more than two disorders. Additionally, specialized CNN model provides better interpretations of predicted diagnosis. Conclusions: TL can indeed lead to superior performance on binary classification in timely and data efficient manner, yet for detecting more than a single disorder, TL models do not perform well. Additionally, custom 3D model performs comparably to TL models for binary classification, and interestingly perform better for diagnosis of multiple disorders. The results confirm the superiority of the custom 3D-CNN in providing better explainable model compared to TL adopted ones. [ABSTRACT FROM AUTHOR]

Introduction: There is a growing body of evidence that subtle deficits in instrumental activities of daily living (IADL) may be present in mild cognitive impairment (MCI). However, it is not clear if there are IADL domains that are consistently affected across patients with MCI. In this systematic review, therefore, we aimed to summarize research results regarding the performance of MCI patients in specific IADL (sub)domains compared with persons who are cognitively normal and/or patients with dementia. Methods: The databases PsycINFO, PubMed and Web of Science were searched for relevant literature in December 2013. Publications from 1999 onward were considered for inclusion. Altogether, 497 articles were retrieved. Reference lists of selected articles were searched for potentially relevant articles. After screening the abstracts of these 497 articles, 37 articles were included in this review. Results: In 35 studies, IADL deficits (such as problems with medication intake, telephone use, keeping appointments, finding things at home and using everyday technology) were documented in patients with MCI. Financial capacity in patients with MCI was affected in the majority of studies. Effect sizes for group differences between patients with MCI and healthy controls were predominantly moderate to large. Performance-based instruments showed slight advantages (in terms of effect sizes) in detecting group differences in IADL functioning between patients with MCI, patients with Alzheimer's disease and healthy controls. Conclusion: IADL requiring higher neuropsychological functioning seem to be most severely affected in patients with MCI. A reliable identification of such deficits is necessary, as patients with MCI with IADL deficits seem to have a higher risk of converting to dementia than patients with MCI without IADL deficits. The use of assessment tools specifically designed and validated for patients with MCI is therefore strongly recommended. Furthermore, the development of performance-based assessment instruments should be intensified, as they allow a valid and reliable assessment of subtle IADL deficits in MCI, even if a proxy is not available. Another important point to consider when designing new scales is the inclusion of technology-associated IADL. Novel instruments for clinical practice should be time-efficient and easy to administer. [ABSTRACT FROM AUTHOR]