Background Transfer learning (TL) with convolutional neural networks aims to improve performances on a new task by leveraging the knowledge of similar tasks learned in advance. It has made a major contribution to medical image analysis as it overcomes the data scarcity problem as well as it saves time and hardware resources. However, transfer learning has been arbitrarily configured in the majority of studies. This review paper attempts to provide guidance for selecting a model and TL approaches for the medical image classification task. Methods 425 peer-reviewed articles were retrieved from two databases, PubMed and Web of Science, published in English, up until December 31, 2020. Articles were assessed by two independent reviewers, with the aid of a third reviewer in the case of discrepancies. We followed the PRISMA guidelines for the paper selection and 121 studies were regarded as eligible for the scope of this review. We investigated articles focused on selecting backbone models and TL approaches including feature extractor, feature extractor hybrid, fine-tuning and fine-tuning from scratch. Results The majority of studies (n = 57) empirically evaluated multiple models followed by deep models (n = 33) and shallow (n = 24) models. Inception, one of the deep models, was the most employed in literature (n = 26). With respect to the TL, the majority of studies (n = 46) empirically benchmarked multiple approaches to identify the optimal configuration. The rest of the studies applied only a single approach for which feature extractor (n = 38) and fine-tuning from scratch (n = 27) were the two most favored approaches. Only a few studies applied feature extractor hybrid (n = 7) and fine-tuning (n = 3) with pretrained models. Conclusion The investigated studies demonstrated the efficacy of transfer learning despite the data scarcity. We encourage data scientists and practitioners to use deep models (e.g. ResNet or Inception) as feature extractors, which can save computational costs and time without degrading the predictive power.
IMPORTANCE Although the detrimental effects of alcohol on the brain are widely acknowledged, observed structural changes are highly heterogeneous, and diagnostic markers for characterizing alcohol-induced brain damage, especially in early abstinence, are lacking. This heterogeneity, likely contributed to by comorbidity factors in patients with alcohol use disorder (AUD), challenges a direct link of brain alterations to the pathophysiology of alcohol misuse. Translational studies in animal models may help bridge this causal gap.OBJECTIVE To compare microstructural properties extracted using advanced diffusion tensor imaging (DTI) in the brains of patients with AUD and a well-controlled rat model of excessive alcohol consumption and monitor the progression of these properties during early abstinence. DESIGN, SETTING, AND PARTICIPANTSThis prospective observational study included 2 cohorts of hospitalized patients with AUD (n = 91) and Marchigian Sardinian alcohol-preferring (msP) rats (n = 27). In humans cross-sectional comparison were performed with control participants (healthy men [n = 36]) and longitudinal comparisons between different points after alcohol withdrawal. In rats, longitudinal comparisons were performed in alcohol-exposed (n = 27) and alcohol-naive msP rats (n = 9). MAIN OUTCOMES AND MEASURES Fractional anisotropy and other DTI measures of white matter properties after long-term alcohol exposure and during early abstinence in both species and clinical and demographic variables and time of abstinence after discharge from hospital in patients. RESULTSThe analysis included 91 men with AUD (mean [SD] age, 46.1 [9.6] years) and 27 male rats in the AUD groups and 36 male controls (mean [SD] age, 41.7 [9.3] years) and 9 male control rats. Comparable DTI alterations were found between alcohol and control groups in both species, with a preferential involvement of the corpus callosum (fractional anisotropy Cohen d = −0.84 [P < .01] corrected in humans and Cohen d = −1.17 [P < .001] corrected in rats) and the fornix/fimbria (fractional anisotropy Cohen d = −0.92 [P < .001] corrected in humans and d = −1.24 [P < .001] corrected in rats). Changes in DTI were associated with preadmission consumption patterns in patients and progress in humans and rats during 6 weeks of abstinence. Mathematical modeling shows this process to be compatible with a sustained demyelination and/or a glial reaction.CONCLUSIONS AND RELEVANCE Using a translational DTI approach, comparable white matter alterations were found in patients with AUD and rats with long-term alcohol consumption. In humans and rats, a progression of DTI alterations into early abstinence (2-6 weeks) suggests an underlying process that evolves soon after cessation of alcohol use.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.