Donald Doherty's blog

Category: Brain Science

  • Epilepsy EEG, MRI, and Post-Operational Assessment Online Data Repository

    Epilepsy is a relatively common brain disorder that can be severely debilitating and sometimes lead to death. Several drugs are available that effectively help to control epileptic seizures. However, about a third of the population of patients with epilepsy have drug-resistant forms where the only known treatment is to remove the brain tissue that initiates the seizures.

    Figure 1. A record from the EEG.PL Epilepsy data repository. From “Open Database of Epileptic EEG with MRI and Postoperational Assessment of Foci—a Real World Verification for the EEG Inverse Solutions“. By Piotr Zwolinski, Marcin Roszkowski, Jaroslaw Zygierewicz, Stefan Haufe, Gido Nolte, and Piotr J. Durka. Neuroinformatics Volume 8, December 2010.

    A major goal stated in the paper “Open Database of Epileptic EEG with MRI and Postoperational Assessment of Foci—a Real World Verification for the EEG Inverse Solutions” (published December 2010 in Neuroinformatics) is to reduce the need for complicated surgical procedures to figure out the location of the brain tissue responsible for kicking off epileptic seizures. To accomplish this goal the authors report on research to improve the accuracy of epilepsy source localization from scalp electroencephalgram (EEG) recordings using data from their publicly available repository .

    The EEG.PL Epilepsy data repository contains records of 23 patients with severe epilepsy between the ages of 1 to 18 years old. All patients were diagnosed with and operated on for drug-resistant epilepsy. In each case, surgery was performed. During surgery electrodes were placed directly on the exposed surface of the brain to collect electrocorticography (ECoG) data, which enabled precise localization of the origin of the seizure activity. If brain tissue was removed, its pathology was extensively examined.

    Each EEG.PL Epilepsy data repository record (see Figure 1 above) includes:

    • Clinically relevant EEG epochs collected during the pre-surgical period.
    • Scanned printouts of EEG epochs containing the epileptogenic structures explicitly marked by the epileptologist.
    • Magnetic Resonance Imaging (MRI) brain scans of the patient.
    • Identification and structural outline of the epileptogenic zone marked on pre-surgical MRI scans in transverse, sagittal and coronal projections. The placement was verified by ECoG and post-operational results.
    • A textual description of each case containing demographic data of each patient, short medical history of epilepsy, essential symptomatology of epileptic fits, additional necessary diagnostic test results, and a concordance report for the epilepsy surgery decision.

    The paper under review uses these data to investigate how different computational procedures performed on the EEG data affect the prediction of where the epileptic focus is located in the patient’s brain. For instance, the individual geometry of a patient’s head is an important variable that affects EEG recordings. Using the MRI data included in each EEG.PL Epilepsy data repository record, the patient’s individual head geometry may be taken into account. Algorithms can be used to segment an MR image into volumes representing skin, skull and brain tissues. The EEG.PL Epilepsy data repository appears to be a rich data source that may ultimately benefit patients with epilepsy.

  • ttime: An R Package for Comparing and Contrasting Brain Development Event Timing Across Mammals

    Clinicians, drug companies, and scientists rely on data gathered from non-human brains. Only a relatively small set of data is gathered from human brains and the kinds of human brain data are limited. These reasons alone make a compelling case to understand the similarities and differences in developmental event timing amongst brains across mammalian species. The recent paper “ttime: an R Package for Translating the Timing of Brain Development Across Mammalian Species” (published October 2010 in Neuroinformatics) introduces the current version of a free, open source, and publicly available software package named ttime that runs in R, a free, open source, and publicly available software environment for statistical computing and graphics.

    The authors point out that we have a relative abundance of data on brain development for some mammals such as rats, mice, and macaque monkeys but far less for others such as rabbits, cats, and humans. To get a more complete picture they set out to build software that would translate brain development timing data from one mammal, a rat for instance, into a model of brain development timing into another mammal, a human for instance. Previous research has demonstrated that event timing sequences in brain development are conserved across mammalian species so their main task was to capture, as best they could, the transformation rules that take time sequences from the context of the developing brain in one mammal and place them into the context of another.

    Here’s a brief tutorial (based on the paper) to get you up and running:

    Note: Here are the details for installing the ttime R package on a Macintosh computer (simply because that’s what I’m using). Go to R’s Packages & Data menu and select Package Installer. Use the CRAN (binaries) repository and search on ttime. Select ttime in the Package list and check the install dependencies check box and then click on the Install Selected button.

    Now you should be able to load the ttime library by typing in the following (> is the R prompt):

    >library(ttime)

    The ttime package is provided with a test data set named event_data. This data set includes data representing 106 brain development events from 10 different mammals: hamster, mouse, rat, rabbit, spiny mouse, guinea pig, ferret, cat, macaque, and human. The authors pulled these data from published peer reviewed sources. Load the test data set by entering the following:

    >data(event_data)

    The ttime function translate() uses primate cerebral cortical and limbic interaction terms to predict unknown brain development event timing across mammalian species. While using this model you must pass the translate() function the name of the data file and the number of non-primate species (npsp) represented in the data file. Enter the following to test the translate() function:

    >npsp<-8; >results<-translate(event_data, npsp); A scatter plot should appear like in Figure 1 below.

    Figure 1. This scatter plot shows values of the entire data set of 106 brain development event times from 10 different mammals (x-axis) in relation to times predicted by ttime for other species (y-axis). The events and species are not labeled here.

    Note: The ttime package includes the ability to investigate phylogenetic proximity amongst species.

    The ttime package is a powerful tool that provides the user flexibility to analyze your own data sets as long as the data are compiled into the proper format. A brief reference manual is available here. However, you’ll probably need to read the paper under review and perhaps some of the team’s older papers to get more use out of the package. For those who do not need the flexibility of running and potentially modifying the code and who don’t need to access their own data sets, the team runs a service called Translating Time at www.translatingtime.net. Even those who use the service may find that access to the source code is useful or even critical.

  • Unambiguous Determination of the Direct Inputs to a Single Neuron

    Identifying neural circuits and figuring out how they process signals has been a prominent approach in attempts to understand how the brain works. However, the clear identification of neural circuit components is nearly always impossible. Even when a particular neuron is indisputably identified the rest of the circuit may usually only be guessed at using neural anatomical principles gained over the past century. A new technique reported in the paper “Targeting Single Neuronal Networks for Gene Expression and Cell Labeling In Vivo” (published August 26, 2010 in Neuron) may go a long way towards changing this situation.

    Using the new technique, a single living neuron and the neurons that form synapses onto it may be unambiguously labeled. This is accomplished in part by using a modified virus that spreads from the original infected neuron through synapses to the neurons that formed the synapses. The first neuron is infected using a technique called electroporation. The virus only moves backwards (retrograde) and only crosses one synapse. The virus does not cross the infected neuron’s synapses onto other cells.

    The labels depend on what are put into the genes. The electroporated neurons in this paper were labeled with yellow florescence while the neuron’s that synapse onto them were labeled with red florescence. One alternative labeling scheme suggested in the paper would be to use Brainbow labeling of the monosynaptically connected neurons so the synapses may be identified by color. (Brainbow labeling would cause each neuron to have a distinct randomly expressed florescent color.) Wouldn’t this be exciting while recording the electrical activity of the labeled neurons?