Deep Relational Reasoning for the Prediction of Language Impairment and Postoperative Seizure Outcome Using Preoperative DWI Connectome Data of Children With Focal Epilepsy

Banerjee, Soumyanil; Dong, Ming; Lee, Min Hee; O'Hara, Nolan; Juhász, Csaba; Asano, Eishi; Jeong, Jeong‐Won

doi:10.1109/tmi.2020.3036933

Cited by 13 publications

(16 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The present study utilized the dilated CNN + RN (Banerjee et al, 2020; Banerjee et al, 2021; Santoro et al, 2017) to objectively predict language scores, where the dilated CNN maps the sparse input matrix, A m,n , into a set of features, which are modeled as objects in the RN to reason about the relationship between objects, leading to a more accurate prediction of an output: expressive or receptive language score for individual patients (Figure 1). Briefly, the dilated CNN is a generalized CNN operator to provide exponential expansion of the receptive field without resolution loss.…”

Section: Methodsmentioning

confidence: 99%

“…In our RN, relational reasoning is performed between every pair of possible objects (i.e., dilated CNN features) regardless of their spatial arrangements in Ω expressive(receptive) . That is, for accurate prediction of language score: y k , the function: g is applied on each object combination to calculate the relation of every object pair (Banerjee et al, 2020; Banerjee et al, 2021; Santoro et al, 2017).

y_{k} = RN ((), O) = f_{φ} ((), 1 / normalN \sum_{normali, normalj} g_{θ} ((),, {normalo}_{normali}, {normalo}_{normalj}))

where o i and o j are a possible object pair obtained from the feature maps and O is the set of all objects.…”

Section: Methodsmentioning

confidence: 99%

“…An adaptive learning rate approach for stochastic gradient descent using the Adam optimizer (Kingma & Ba, 2014) minimizes the loss function (i.e., the mean square error for language score prediction). Detailed architecture and parameters of the dilated CNN + RN were available in our recent work (Banerjee et al, 2020; Banerjee et al, 2021).…”

Section: Methodsmentioning

confidence: 99%

“…To determine whether the pDWIC imaging features of two language networks: Ω expressive and Ω receptive can accurately predict later expressive and expressive language impairment/scores of in individual children with early LLE, the present study performed the prediction tasks with an end‐to‐end deep learning network, called “dilated convolution neural network (CNN) with a relational network (RN) (Banerjee et al, 2020; Banerjee et al, 2021; Santoro et al, 2017), that is a special graph network whose computations explicitly focus on relational reasoning.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Deep reasoning neural network analysis to predict language deficits from psychometry‐driven DWI connectome of young children with persistent language concerns

Jeong

Banerjee

Lee

et al. 2021

Human Brain Mapping

Self Cite

View full text Add to dashboard Cite

This study investigated whether current state‐of‐the‐art deep reasoning network analysis on psychometry‐driven diffusion tractography connectome can accurately predict expressive and receptive language scores in a cohort of young children with persistent language concerns (n = 31, age: 4.25 ± 2.38 years). A dilated convolutional neural network combined with a relational network (dilated CNN + RN) was trained to reason the nonlinear relationship between “dilated CNN features of language network” and “clinically acquired language score”. Three‐fold cross‐validation was then used to compare the Pearson correlation and mean absolute error (MAE) between dilated CNN + RN‐predicted and actual language scores. The dilated CNN + RN outperformed other methods providing the most significant correlation between predicted and actual scores (i.e., Pearson's R/p‐value: 1.00/<.001 and .99/<.001 for expressive and receptive language scores, respectively) and yielding MAE: 0.28 and 0.28 for the same scores. The strength of the relationship suggests elevated probability in the prediction of both expressive and receptive language scores (i.e., 1.00 and 1.00, respectively). Specifically, sparse connectivity not only within the right precentral gyrus but also involving the right caudate had the strongest relationship between deficit in both the expressive and receptive language domains. Subsequent subgroup analyses inferred that the effectiveness of the dilated CNN + RN‐based prediction of language score(s) was independent of time interval (between MRI and language assessment) and age of MRI, suggesting that the dilated CNN + RN using psychometry‐driven diffusion tractography connectome may be useful for prediction of the presence of language disorder, and possibly provide a better understanding of the neurological mechanisms of language deficits in young children.

show abstract

Section: Methodsmentioning

confidence: 99%

y_{k} = RN ((), O) = f_{φ} ((), 1 / normalN \sum_{normali, normalj} g_{θ} ((),, {normalo}_{normali}, {normalo}_{normalj}))

where o i and o j are a possible object pair obtained from the feature maps and O is the set of all objects.…”

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Deep reasoning neural network analysis to predict language deficits from psychometry‐driven DWI connectome of young children with persistent language concerns

Jeong

Banerjee

Lee

et al. 2021

Human Brain Mapping

Self Cite

View full text Add to dashboard Cite

show abstract

“…GNNs and its variants have been employed for numerous applications such as recommendation systems ( Wu, Sun, Zhang, & Cui, 2020 ), protein interface prediction ( Fout, Byrd, Shariat, & Ben-Hur, 2017 ) and semantic segmentation ( Qi, Liao, Jia, Fidler, & Urtasun, 2017 ). For example, a combination of Convolution Neural Networks (CNNs) and graph relation networks have been used for brain connectivity network analysis ( Banerjee et al, 2020 ).…”

Section: Related Workmentioning

confidence: 99%

Spatial–Temporal Synchronous Graph Transformer network (STSGT) for COVID-19 forecasting

2022

Self Cite

View full text Add to dashboard Cite

Application value of molecular imaging technology in epilepsy

Rong

Zhang

Wang

et al. 2021

Ibrain

View full text Add to dashboard Cite

Epilepsy is a common neurological disease with various seizure types, complicated etiologies, and unclear mechanisms. Its diagnosis mainly relies on clinical history, but an electroencephalogram is also a crucial auxiliary examination. Recently, brain imaging technology has gained increasing attention in the diagnosis of epilepsy, and conventional magnetic resonance imaging can detect epileptic foci in some patients with epilepsy. However, the results of brain magnetic resonance imaging are normal in some patients. New molecular imaging has gradually developed in recent years and has been applied in the diagnosis of epilepsy, leading to enhanced lesion detection rates. However, the application of these technologies in epilepsy patients with negative brain magnetic resonance must be clarified. Thus, we reviewed the relevant literature and summarized the information to improve the understanding of the molecular imaging application value of epilepsy.

show abstract

Deep Relational Reasoning for the Prediction of Language Impairment and Postoperative Seizure Outcome Using Preoperative DWI Connectome Data of Children With Focal Epilepsy

Cited by 13 publications

References 50 publications

Deep reasoning neural network analysis to predict language deficits from psychometry‐driven DWI connectome of young children with persistent language concerns

Deep reasoning neural network analysis to predict language deficits from psychometry‐driven DWI connectome of young children with persistent language concerns

Spatial–Temporal Synchronous Graph Transformer network (STSGT) for COVID-19 forecasting

Application value of molecular imaging technology in epilepsy

Contact Info

Product

Resources

About