Multimodality Data Integration in Epilepsy

Otto Muzik, Diane C. Chugani, Guangyu Zou, Jing Hua, Yi Lu, Shiyong Lu, Eishi Asano, Harry T. ChuganiInternational Journal of Biomedical Imaging 2007 · DOI

TL;DR

This paper introduces a novel software environment and database for integrating multimodal brain imaging data (PET, MRI, EEG) in epilepsy patients. It enables quantitative assessment of the relationships between functional brain activity and electrophysiological signals, aiming to improve the understanding and localization of epileptic foci for better treatment outcomes.

What Problem Does This Paper Address?

Diagnosing and treating epilepsy, especially medically intractable forms, requires a deep understanding of brain function and its abnormalities. While various imaging techniques like PET (measuring metabolic activity) and EEG (measuring electrical activity) provide crucial information, integrating and quantitatively analyzing data from these disparate modalities has been a significant challenge. Existing methods often rely on qualitative assessments or struggle to precisely correlate findings across different data types, particularly in the complex context of epilepsy where precise localization of the seizure onset zone is critical for surgical planning. This paper addresses the need for a unified, quantitative framework to bridge the gap between anatomical, functional, and electrophysiological data, thereby enhancing our ability to understand epileptogenesis and guide treatment.

Methods

The core of this work lies in a novel software environment and a structured database designed for multimodality integration. Key methodological components include:

  • Landmark-Constrained Conformal Surface Mapping: This technique is used to create a standardized representation of each patient's cortical surface. By defining anatomical landmarks (like the central sulcus), the method allows for accurate mapping of individual brain surfaces onto a canonical spherical domain. This is crucial for comparing data across subjects and for ensuring that analyses are performed in the patient's native space, accommodating structural abnormalities.
  • Quantitative Assessment of PET Abnormalities: The software calculates a "severity index" to quantify deviations in PET tracer concentration (e.g., FDG for glucose metabolism) compared to a control group. This index is then mapped onto the cortical surface, providing a visual and quantitative representation of functional abnormalities.
  • Spatial Proximity Index (SPI): This novel index quantifies the spatial relationship between electrophysiological abnormalities (from EEG) and functional PET abnormalities. It measures the distance between seizure onset electrodes and areas of abnormal PET uptake, providing a continuous variable that can be used for data mining and to assess how well PET imaging guides EEG electrode placement.
  • Extendable XML-Based Database: The database is designed to be flexible and accommodate new modalities and data types. It uses an XML-based approach to allow for graceful integration of future data, ensuring scalability and adaptability. The database is structured hierarchically (patient, cluster, cortical element) to store clinical, PET, and EEG data.
  • Multimodality Display and Querying: The software provides a 3D visualization module that overlays PET functional data onto MR anatomical surfaces, with EEG electrodes color-coded by their activity. Users can interactively explore these integrated datasets and perform specific queries, such as identifying electrodes with high spike frequencies and calculating their corresponding SPI values.

Datasets

The study utilized data from twelve children (mean age 5.2 ± 4.3 years) with medically intractable partial epilepsy. These patients underwent high-resolution structural MRI, functional PET imaging (using FDG and FMZ tracers), and intracranial EEG monitoring. A control group of 15 young adults (mean age 27.6 ± 4.5 years) with normal MRI scans was used to establish normative PET tracer concentration patterns. No new benchmarks were introduced; rather, the work focuses on integrating existing data types.

Key Insights and Ideas

  • Bridging the Modality Gap: The most significant insight is the development of a practical framework that quantitatively links functional brain imaging (PET) with electrophysiological recordings (EEG). This moves beyond qualitative visual inspection to a data-driven approach.
  • Native Space Analysis is Key: Performing analyses in the patient's native space, rather than relying solely on atlas-based warping, is crucial for accurately capturing individual brain anatomy and abnormalities, especially in conditions like epilepsy.
  • SPI as a Predictive Metric: The Spatial Proximity Index (SPI) offers a powerful quantitative measure to assess the spatial overlap between seizure onset zones and functional abnormalities. This has direct implications for evaluating the effectiveness of PET in guiding the placement of EEG electrodes.
  • Database as a Foundation for Discovery: The extendable, XML-based database design is a forward-thinking approach, enabling future research by allowing seamless integration of new imaging techniques and data types as they emerge.

Key Results

The paper demonstrates the implementation of the software environment in twelve pediatric epilepsy patients. The experiments showed that the approach can lead to a better understanding of epileptogenesis. Specifically:

  • The software successfully integrated MRI, PET, and EEG data, providing a unified 3D visualization.
  • The Spatial Proximity Index (SPI) was calculated, showing quantitative relationships between EEG and PET findings. For example, lower SPI values indicated a closer spatial proximity between PET and EEG modalities, suggesting successful guidance of electrode placement by PET imaging.
  • The system allows for advanced data mining, such as association rule mining, to identify relationships between clinical variables (seizure duration, severity) and SPI values, potentially revealing patterns in epilepsy progression.

Limitations and Future Work

The authors acknowledge several limitations:

  • Control Group Age Mismatch: The use of young adult controls for establishing normative PET patterns, when analyzing pediatric patients, is a potential source of error, though they suggest FDG patterns might be similar.
  • Alignment Accuracy: Spatial alignment between MRI and PET volumes, and the manual definition of cortical landmarks for surface mapping, introduce potential misregistration errors, though they estimate the total effect to be less than 5 mm.
  • Reproducibility of Landmark Definition: The manual definition of cortical landmarks, while using well-defined features, can still contribute to misalignment.

Future work suggested includes:

  • Further validation of the method in larger patient cohorts and for other neurological disorders.
  • Incorporating additional modalities like diffusion tensor imaging (DTI) for brain connectivity and susceptibility-weighted imaging (SWI) for venous vasculature.
  • Developing more advanced graphical user interfaces for intuitive data exploration and meta-analysis of data patterns across different disorders.

Summary updated

Abstract

An important goal of software development in the medical field is the design of methods which are able to integrate information obtained from various imaging and nonimaging modalities into a cohesive framework in order to understand the results of qualitatively different measurements in a larger context. Moreover, it is essential to assess the various features of the data quantitatively so that relationships in anatomical and functional domains between complementing modalities can be expressed mathematically. This paper presents a clinically feasible software environment for the quantitative assessment of the relationship among biochemical functions as assessed by PET imaging and electrophysiological parameters derived from intracranial EEG. Based on the developed software tools, quantitative results obtained from individual modalities can be merged into a data structure allowing a consistent framework for advanced data mining techniques and 3D visualization. Moreover, an effort was made to derive quantitative variables (such as the spatial proximity index, SPI) characterizing the relationship between complementing modalities on a more generic level as a prerequisite for efficient data mining strategies. We describe the implementation of this software environment in twelve children (mean age 5.2 +/- 4.3 years) with medically intractable partial epilepsy who underwent both high-resolution structural MR and functional PET imaging. Our experiments demonstrate that our approach will lead to a better understanding of the mechanisms of epileptogenesis and might ultimately have an impact on treatment. Moreover, our software environment holds promise to be useful in many other neurological disorders, where integration of multimodality data is crucial for a better understanding of the underlying disease mechanisms.
BibTeX
@article{muzik2007multimodality,
  author = {Otto Muzik and Diane C. Chugani and Guangyu Zou and Jing Hua and Yi Lu and Shiyong Lu and Eishi Asano and Harry T. Chugani},
  title = {{Multimodality Data Integration in Epilepsy}},
  journal = {International Journal of Biomedical Imaging},
  year = {2007},
  doi = {10.1155/2007/13963},
}
Figure