Functional Neuromarkers for Psychiatry

Applications for Diagnosis and Treatment

Juri D. Kropotov

• Publisher: Academic Press
• Year: 2016
• ISBN:
  9780124105201
• Topic: Biological Sciences
• Subtopic: Neuroscience
• Index: Biological Sciences

Functional Neuromarkers for Psychiatry explores recent advances in neuroscience that have allowed scientists to discover functional neuromarkers of psychiatric disorders. These neuromarkers include brain activation patterns seen via fMRI, PET, qEEG, and ERPs. The book examines these neuromarkers in detail—what to look for, how to use them in clinical practice, and the promise they provide toward early detection, prevention, and personalized treatment of mental disorders.

The neuromarkers identified in this book have a diagnostic sensitivity and specificity higher than 80%. They are reliable, reproducible, inexpensive to measure, noninvasive, and have been confirmed by at least two independent studies. The book focuses primarily on the analysis of EEG and ERPs. It elucidates the neuronal mechanisms that generate EEG spontaneous rhythms and explores the functional meaning of ERP components in cognitive tasks. The functional neuromarkers for ADHD, schizophrenia, and obsessive-compulsive disorder are reviewed in detail. The book highlights how to use these functional neuromarkers for diagnosis, personalized neurotherapy, and monitoring treatment results.

• Identifies specific brain activation patterns that are neuromarkers for psychiatric disorders
• Includes neuromarkers as seen via fMRI, PET, qEEG, and ERPs
• Addresses neuromarkers for ADHD, schizophrenia, and OCD in detail
• Provides information on using neuromarkers for diagnosis and/or personalized treatment

Part 1

Methods of assessing neuromarkers

Chapter 1.1: Theory of Measurement

Chapter 1.2: Psychometrics and Neuropsychological Assessment

Chapter 1.3: Functional Magnetic Resonance Imaging

Chapter 1.4: Positron Emission Tomography

Chapter 1.5: Spontaneous Electroencephalogram

Chapter 1.6: Event-Related Potentials

Chapter 1.1

Theory of Measurement

Abstract

Any measurement is made with errors. Sources of errors in neuromarker measures are the subject variations in mood, fatigue, stress, the subject bias in motivation, cheating, the variations of the environment, test administration bias, etc. The error effects reliability of measurements. In practice the test-retest reliability is used as the correlation between the scores at two different times on a selected population of subjects. Most of functional neuromarkers are normally or log-normally distributed. Z-scores are used to measure the individual deviation from the mean value of the population. When a neuromarker is used for diagnosis the test outcome can be positive or negative for a tested subject and may or may not match the subject’s actual status. Sensitivity of neuromarker is defined as probability of the positive test given that the subject is ill. Specificity is probability of the negative test given that the subject is healthy. For any test, there is a trade-off between sensitivity and specificity represented graphically in a receiver operating characteristic curve (ROC). The other measure of the difference between two groups of subjects is given by effect size - the difference between means in terms of the standard deviation. For being a real help for psychiatry the neuromarker must be: 1) reliable, 2) sensitive and 3) specific index of the brain functioning and/or dysfunctioning.

Keywords

neuromarker

reliability

sensitivity

specificity

Z-scores;

receiver operating characteristic curve (ROC)

True and observed scores, errors

Neuromarkers are measured quantitatively in each person selected from the general population. From our experience we know that any measurement is made with errors, not talking about mistakes. The question is how those errors can be estimated. Charles Spearman at the beginning of the 20th century was the first who figured out how errors can be estimated and who laid down foundations of the theory of measurement (Spearman, 1904).

In brief, classical test theory assumes that each person has a true score, T, that would be obtained if there were no errors in measurement. A person’s true score is defined as the score over an infinite number of independent applications of the test. Because the infinite number is never reached, a person’s true score is never obtained. So in any test only an observed score X is measured. It is assumed that observed score = true score plus some error E:

$X=T+E$

The error E is composed of two types of error: random E_r and systematic E_s.

$E=E_{\text{r}}+E_{\text{s}}$

The random error E_r is defined as a randomly varying effect on the same subject across different testing sessions. Sources of E_r are subject variations in mood, fatigue, stress, etc.; subject bias in motivation, cheating, etc.; variations of the environment in noise, temperature, lighting, seat comfort, etc.; test administration bias such as nonstandard instruction, scoring errors. When we are talking about brain activity a source of E_r could be spontaneous variations in the local brain state as well as errors of the device measurement. Spontaneous variation in the parameter can be decreased by averaging. In such case the signal-to-noise ratio increases as the square root of numbers of averaged trials. Errors in the device measurement, such as 1-μm error in most electroencephalogram (EEG) devices, is defined by manufacturers and cannot be decreased.

E_r affects the repeatability and reproducibility of measurements. Repeatability of measurements refers to variation in repeat measurements made on the same subject under identical conditions and over a short period (several hours or days) of time. Variability in measurements made under these conditions can then be ascribed only to measurement errors E.

Reproducibility refers to variation in measurements made on a subject under changing conditions. Changing conditions may be due to different measurement methods or instruments being used, measurements being made by different observers or raters, or measurements being made over a long (from several weeks to several months) period of time.

The systematic error E_s is that error which is consistent across test sessions such as poorly written or verbalized directions, inclusion of items unrelated to the content, etc. E_s affects validity, but not repeatability and reproducibility. In general, validity refers to how well a test measures what it is supposed to measure.

Reliability

The classical test theory deals with relations between T (true score), X (observed score), and E (error) measured in a population of subjects. The reliability of observed test scores X, which is denoted as ρ²_XT, is defined as the ratio of true score variance σ²_T to observed score variance σ²_X:

${\rho }_{XT}^2=\frac{{\sigma }_T^2}{{\sigma }_X^2}\text{;}$

which after simple transformations can be presented as:

${\rho }_{XT}^2=1-\frac{{\sigma }_E^2}{{\sigma }_X^2}$

From this equation one can see that (1) the reliability of test scores is lower than one, (2) the reliability becomes higher as the error variance σ²_E becomes lower, (3) the heterogeneity of subjects in the population, measured by σ²_X, affects the value of reliability: the higher the heterogeneity, the higher the reliability. It can be mathematically shown that the square root of reliability (called the reliability index or ρ_XT) equals the correlation between observed and true scores in the population of persons.

In practice the correlation between the observed X scores measured at two different times t₁ and t₂ on a selected population of subjects is used for assessing the reliability of the measured X value. This correlation measures test–retest reliability. The other measure of reliability is half-split reliability in which the whole sample is randomly divided in two equal sets and the Pearson coefficient is computed between these sets.

Validity

While reliability of the test is necessary for appropriate measurement, alone it is not sufficient. For any measurement to be reliable, it also needs to be valid. As was mentioned previously, validity is the extent to which a measurement is well founded and corresponds accurately to the real world. The word “valid” is derived from the Latin validus, meaning strong.

Distribution across population

Any parameter when measured in a given population varies from one person to another. EEG spectra (both absolute and relative) are not distributed normally. In all known databases EEG spectra are subjected to a normalization procedure by taking the log of corresponding values. In contrast to EEG spectra, the amplitudes of event-related potentials are normally distributed and are not subjected to any normalization procedure.

Percentiles and z scores

When describing the results of measurement within...

Table of contents

• Cover
• Title page
• Table of Contents
• Copyright
• Acknowledgments
• Introduction
• Part 1: Methods of assessing neuromarkers
• Part 2: Neuromarkers of cortical self-regulation
• Part 3: Information flow within the brain
• Part 4: Methods of neuro-modulation
• Part 5: Neuromarkers in psychiatry
• Part 6: Assessing functional neuromarkers
• Part 7: The state of the art: overview
• Postscriptum
• References
• Further Readings
• Subject Index