LV-CadeNet: Long View Feature Convolution-Attention Fusion Encoder-Decoder Network for Clinical MEG Spike Detection

2024-12-12

arXiv.org

AI Summary

Existing automated interictal epileptiform discharge (IED) detection methods suffer from two key limitations: (1) insufficient modeling of long-range temporal context for clinical significance assessment, and (2) neglect of dipole potential distribution across adjacent sensors—a critical diagnostic feature. To address challenges inherent in clinical magnetoencephalography (MEG) data—including severe class imbalance, scarce expert annotations, and poor generalizability—we propose the Long View Morphological Modeling framework. Our approach introduces a novel encoder-decoder architecture that jointly integrates spatiotemporal convolution and self-attention mechanisms, and incorporates semi-supervised learning to mitigate domain shift between synthetic and real clinical data distributions. Evaluated on a real-world MEG dataset from Sanbo Brain Hospital, Capital Medical University, our method achieves an IED detection accuracy of 54.88%, substantially improving upon the prior state-of-the-art (42.31%). This work establishes a new paradigm for clinically interpretable and robust automated IED identification.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

It is widely acknowledged that the epileptic foci can be pinpointed by source localizing interictal epileptic discharges (IEDs) via Magnetoencephalography (MEG). However, manual detection of IEDs, which appear as spikes in MEG data, is extremely labor intensive and requires considerable professional expertise, limiting the broader adoption of MEG technology. Numerous studies have focused on automatic detection of MEG spikes to overcome this challenge, but these efforts often validate their models on synthetic datasets with balanced positive and negative samples. In contrast, clinical MEG data is highly imbalanced, raising doubts on the real-world efficacy of these models. To address this issue, we introduce LV-CadeNet, a Long View feature Convolution-Attention fusion Encoder-Decoder Network, designed for automatic MEG spike detection in real-world clinical scenarios. Beyond addressing the disparity between training data distribution and clinical test data through semi-supervised learning, our approach also mimics human specialists by constructing long view morphological input data. Moreover, we propose an advanced convolution-attention module to extract temporal and spatial features from the input data. LV-CadeNet significantly improves the accuracy of MEG spike detection, boosting it from 42.31% to 54.88% on a novel clinical dataset sourced from Sanbo Brain Hospital Capital Medical University. This dataset, characterized by a highly imbalanced distribution of positive and negative samples, accurately represents real-world clinical scenarios.

Problem

• Automating EEG/MEG spike analysis to reduce manual labor and expertise dependency
• Capturing long-view contextual patterns missed by short-window automated methods
• Detecting dipole patterns across sensors that experts use for diagnosis

Innovation

• Long-View morphological feature representation
• Hierarchical Encoder-Decoder network architecture
• Convolution-Attention blocks for spatiotemporal learning