Adaptive cluster analysis approach for functional localization using magnetoencephalography

Alikhanian, Hooman; Crawford, J. Douglas; DeSouza, Joseph F. X.; Cheyne, Douglas; Blohm, Gunnar

doi:10.3389/fnins.2013.00073

Cited by 16 publications

(21 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…However, since it requires to predefine the number of clusters, i.e., K , for which a cross-validation technique is usually applied in the literature, it is limited to use the K -means algorithm in practical applications. To this end, in this work, we use affinity propagation (Frey and Dueck 2007), which can automatically select the optimal number of clusters and has been successfully applied to a variety of applications (Dueck and Frey 2007; Lu and Carreira-Perpinan 2008; Wang 2010; Shi et al 2011; Alikhanian et al 2013). For the details of affinity propagation, please refer to Appendix and Frey and Dueck (2007).…”

Section: Methodsmentioning

confidence: 99%

Deep sparse multi-task learning for feature selection in Alzheimer’s disease diagnosis

2015

View full text Add to dashboard Cite

Recently, neuroimaging-based Alzheimer’s disease (AD) or mild cognitive impairment (MCI) diagnosis has attracted researchers in the field, due to the increasing prevalence of the diseases. Unfortunately, the unfavorable high-dimensional nature of neuroimaging data, but a limited small number of samples available, makes it challenging to build a robust computer-aided diagnosis system. Machine learning techniques have been considered as a useful tool in this respect and, among various methods, sparse regression has shown its validity in the literature. However, to our best knowledge, the existing sparse regression methods mostly try to select features based on the optimal regression coefficients in one step. We argue that since the training feature vectors are composed of both informative and uninformative or less informative features, the resulting optimal regression coefficients are inevidently affected by the uninformative or less informative features. To this end, we first propose a novel deep architecture to recursively discard uninformative features by performing sparse multi-task learning in a hierarchical fashion. We further hypothesize that the optimal regression coefficients reflect the relative importance of features in representing the target response variables. In this regard, we use the optimal regression co-efficients learned in one hierarchy as feature weighting factors in the following hierarchy, and formulate a weighted sparse multi-task learning method. Lastly, we also take into account the distributional characteristics of samples per class and use clustering-induced subclass label vectors as target response values in our sparse regression model. In our experiments on the ADNI cohort, we performed both binary and multi-class classification tasks in AD/MCI diagnosis and showed the superiority of the proposed method by comparing with the state-of-the-art methods.

show abstract

Section: Methodsmentioning

confidence: 99%

Deep sparse multi-task learning for feature selection in Alzheimer’s disease diagnosis

2015

View full text Add to dashboard Cite

show abstract

“…In the context of the present review, inverse localization refers to the process of identifying the cortical sources of electrical activity which are responsible for generating the electric potentials measured at the scalp by EEG sensors [42, 43]. Illustrations which convey the gist of this technique are available in a number of textbooks [44, 45], journal articles [46–48] and peer-reviewed online sources [49, 50], which the reader is encouraged to consult. Briefly, inverse localization is performed as follows.…”

Section: Recent Advances and Emerging Methodsmentioning

confidence: 99%

Epileptogenic focus localization in treatment-resistant post-traumatic epilepsy

Irimia

Horn

2015

Journal of Clinical Neuroscience

View full text Add to dashboard Cite

Pharmacologically intractable post-traumatic epilepsy (PTE) is a major clinical challenge for patients with penetrating traumatic brain injury, where the risk for this condition remains very high even decades after injury. Although over 20 anti-epileptic drugs (AED) are in common use today, approximately one-third of epilepsy patients have drug-refractory seizures and even more have AED-related adverse effects which compromise life quality. Simultaneously, there have been repeated recommendations by radiologists and neuroimaging experts to incorporate localization based on electroencephalography (EEG) into the process of clinical decision making regarding PTE patients. Nevertheless, thus far, little progress has been accomplished towards the use of EEG as a reliable tool for locating epileptogenic foci prior to surgical resection. In this review, we discuss the epidemiology of pharmacologically resistant PTE, address the need for effective anti-epileptogenic treatments, and highlight recent progress in the development of noninvasive methods for the accurate localization of PTE foci for the purpose of neurosurgical intervention. These trends indicate the current emergence of promising methodologies for the noninvasive study of post-traumatic epileptogenesis and for the improved neurosurgical planning of epileptic foci resection.

show abstract

“…It is based on a bootstrap estimation of the distance between map peaks in the two conditions that is assessed with a multivariate location test. Cluster-based methods have been developed previously that also capitalize on map peaks but their aim was rather to estimate a confidence volume for source location in individual conditions (Alikhanian et al, 2013;Gilbert et al, 2012).…”

Section: Applicability Of the Proposed Location-comparison Proceduresmentioning

confidence: 99%

Contrasting functional imaging parametric maps: The mislocation problem and alternative solutions

2018

View full text Add to dashboard Cite

A B S T R A C TIn the field of neuroimaging, researchers often resort to contrasting parametric maps to identify differences between conditions or populations. Unfortunately, contrast patterns mix effects related to amplitude and location differences and tend to peak away from sources of genuine brain activity to an extent that scales with the smoothness of the maps. Here, we illustrate this mislocation problem on source maps reconstructed from magnetoencephalographic recordings and propose a novel, dedicated location-comparison method. In realistic simulations, contrast mislocation was on average~10 mm when genuine sources were placed at the same location, and was still above 5 mm when sources were 20 mm apart. The dedicated location-comparison method achieved a sensitivity of~90% when inter-source distance was 12 mm. Its benefit is also illustrated on real brain-speech entrainment data. In conclusion, contrasts of parametric maps provide precarious information for source location. To specifically address the question of location difference, one should turn to dedicated methods as the one proposed here.

show abstract

Adaptive cluster analysis approach for functional localization using magnetoencephalography

Cited by 16 publications

References 36 publications

Deep sparse multi-task learning for feature selection in Alzheimer’s disease diagnosis

Deep sparse multi-task learning for feature selection in Alzheimer’s disease diagnosis

Epileptogenic focus localization in treatment-resistant post-traumatic epilepsy

Contrasting functional imaging parametric maps: The mislocation problem and alternative solutions

Contact Info

Product

Resources

About