Dataset Information

A comparison of automated anatomical-behavioural mapping methods in a rodent model of stroke.

ABSTRACT: Neurological damage, due to conditions such as stroke, results in a complex pattern of structural changes and significant behavioural dysfunctions; the automated analysis of magnetic resonance imaging (MRI) and discovery of structural-behavioural correlates associated with these disorders remains challenging. Voxel lesion symptom mapping (VLSM) has been used to associate behaviour with lesion location in MRI, but this analysis requires the definition of lesion masks on each subject and does not exploit the rich structural information in the images. Tensor-based morphometry (TBM) has been used to perform voxel-wise structural analyses over the entire brain; however, a combination of lesion hyper-intensities and subtle structural remodelling away from the lesion might confound the interpretation of TBM. In this study, we compared and contrasted these techniques in a rodent model of stroke (n=58) to assess the efficacy of these techniques in a challenging pre-clinical application. The results from the automated techniques were compared using manually derived region-of-interest measures of the lesion, cortex, striatum, ventricle and hippocampus, and considered against model power calculations. The automated TBM techniques successfully detect both lesion and non-lesion effects, consistent with manual measurements. These techniques do not require manual segmentation to the same extent as VLSM and should be considered part of the toolkit for the unbiased analysis of pre-clinical imaging-based studies.

SUBMITTER: Crum WR

PROVIDER: S-EPMC3759848 | biostudies-literature | 2013 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A comparison of automated anatomical-behavioural mapping methods in a rodent model of stroke.

Crum William R WR Giampietro Vincent P VP Smith Edward J EJ Gorenkova Natalia N Stroemer R Paul RP Modo Michel M

Journal of neuroscience methods 20130529 2

Neurological damage, due to conditions such as stroke, results in a complex pattern of structural changes and significant behavioural dysfunctions; the automated analysis of magnetic resonance imaging (MRI) and discovery of structural-behavioural correlates associated with these disorders remains challenging. Voxel lesion symptom mapping (VLSM) has been used to associate behaviour with lesion location in MRI, but this analysis requires the definition of lesion masks on each subject and does not ...[more]

PMID: 23727124

Similar Datasets

Project description:BackgroundAutomated ICD coding on medical texts via machine learning has been a hot topic. Related studies from medical field heavily relies on conventional bag-of-words (BoW) as the feature extraction method, and do not commonly use more complicated methods, such as word2vec (W2V) and large pretrained models like BERT. This study aimed at uncovering the most effective feature extraction methods for coding models by comparing BoW, W2V and BERT variants.MethodsWe experimented with a Chinese dataset from Fuwai Hospital, which contains 6947 records and 1532 unique ICD codes, and a public Spanish dataset, which contains 1000 records and 2557 unique ICD codes. We designed coding tasks with different code frequency thresholds (denoted as [Formula: see text]), with a lower threshold indicating a more complex task. Using traditional classifiers, we compared BoW, W2V and BERT variants on accomplishing these coding tasks.ResultsWhen [Formula: see text] was equal to or greater than 140 for Fuwai dataset, and 60 for the Spanish dataset, the BERT variants with the whole network fine-tuned was the best method, leading to a Micro-F1 of 93.9% for Fuwai data when [Formula: see text], and a Micro-F1 of 85.41% for the Spanish dataset when [Formula: see text]. When [Formula: see text] fell below 140 for Fuwai dataset, and 60 for the Spanish dataset, BoW turned out to be the best, leading to a Micro-F1 of 83% for Fuwai dataset when [Formula: see text], and a Micro-F1 of 39.1% for the Spanish dataset when [Formula: see text]. Our experiments also showed that both the BERT variants and BoW possessed good interpretability, which is important for medical applications of coding models.ConclusionsThis study shed light on building promising machine learning models for automated ICD coding by revealing the most effective feature extraction methods. Concretely, our results indicated that fine-tuning the whole network of the BERT variants was the optimal method for tasks covering only frequent codes, especially codes that represented unspecified diseases, while BoW was the best for tasks involving both frequent and infrequent codes. The frequency threshold where the best-performing method varied differed between different datasets due to factors like language and codeset.

Dataset Information

A comparison of automated anatomical-behavioural mapping methods in a rodent model of stroke.

Publications

A comparison of automated anatomical-behavioural mapping methods in a rodent model of stroke.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets