1 / 47

Silvana Galderisi University of Naples , SUN

Silvana Galderisi University of Naples , SUN. THE CONTRIBUTION OF NEUROSCIENCE TO THE DIAGNOSIS OF MENTAL DISORDERS. Diagnosis in Medicine. Medical history Physical signs and symptoms Laboratory and diagnostic tests (e.g. x-ray, biopsy or endoscopy). Diagnosis in Psychiatry.

kitra
Download Presentation

Silvana Galderisi University of Naples , SUN

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Silvana GalderisiUniversityofNaples, SUN THE CONTRIBUTION OF NEUROSCIENCE TO THE DIAGNOSIS OF MENTAL DISORDERS

  2. Diagnosis in Medicine • Medical history • Physical signs and symptoms • Laboratory and diagnostic tests (e.g. x-ray, biopsy or endoscopy)

  3. Diagnosis in Psychiatry • Personal development history, psychiatric and medical history • Subject’s behavior • Psychiatrists’ interpretation of individual’s report of his/her thoughts, feelings and experience • Referencing signs and symptoms to psychiatrist’s prototype for different disorders or to international classification systems, such as ICD-10 or DSM-IV

  4. 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 -0.4 Diagnosis in Psychiatry The research effort has been huge, but translation in clinical practice very limited LRP Galderisiet al, 1994; 2002

  5. Focus of the presentation Improvingdiagnosticaccuracy PredictingDiseaseOnset • Structural and functional neuroimaging • Electrophysiology • Neuropsychology • Genetics

  6. Structural NeuroimagingPattern recognition methods • Selection of a predefined set of features (e.g. anatomic regions) for classification using some prior knowledge, e.g. medial temporal lobe or hippocampus pathology in AD (Davatzikos et al., 2008, 2009; Duchesnay et al., 2007; Fan et al., 2006, 2007; Klöppel et al., 2008, 2009; Koutsouleriset al., 2009; Lerchet al., 2008; Teipelet al., 2007; Vemuriet al., 2008)

  7. Structural NeuroimagingPattern recognition methods • With only two potential diagnoses available such as AD (the most common form of dementia) and cognitively normal subjects, classifiers have been shown to outperform radiologists (Klöppel et al., 2008a) • Classifiers are sufficiently sensitive to separate patients with AD or mild cognitive impairment (MCI) (Petersen et al., 2001) from cognitively normal persons (Davatzikos et al., 2008 a,b)

  8. Structural NeuroimagingAlzheimer’s Disease • Recently, it has been recommended that updated diagnostic criteria include AD biomarkers in the diagnostic scheme • Evidence of medial temporal atrophy on structural MRI was one of the major biomarkers included in theserecommendations Dubois et al. 2010; Albert et al., 2011; Jack et al., 2011; McKhann et al., 2011; Sperling et al., 2011)

  9. Structural NeuroimagingSchizophrenia • Promising data were reported; however, it remains unclear whether MRI-based pattern recognition methods trained to dichotomize between HC and SCZ would achieve the level of sensitivity and specificity needed to be integrated intoclinicalreal-worldscenarios Davatzikoset al. 2005; Kawasaki et al., 2007; Ardekaniet al., 2010)

  10. Structural NeuroimagingSchizophrenia • Recentlyproposedsemi-supervisedmachine learning algorithms may provide an alternative to fully supervised machine learning methods • The methodallowsautomatical discovery of novel categories in the dataset by employing a clustering technique • It might deconstruct the heterogeneity of schizophrenia by modelling the hidden neurobiological clustering within this patient population Filipovych et al., Neuroimage. 2011

  11. Conclusions and Significance: The structural neuroanatomy of depression shows high predictive potential for clinical response to antidepressant medication, while its diagnostic potential is more limited. The present findings provide initial steps towards the development of neurobiological prognostic markers for depression.

  12. Pattern RecognitionMethods Will translationtoclinicalpracticebe the nextstep? • Diagnosticaccuracyisbelow 100% in mostcases • The costoferroneouslymisclassifying someone ill as healthy may be higher than the cost of misclassifying someone healthy as ill • Algorithms providing excellent sensitivity but only good specificity should be preferred to those providing excellent specificity but only good sensitivity

  13. BiologicalPsychiatry, 1994 Quantitative descriptors of resting electroencephalogram (QEEG) and event-related potentials (QERP) … were obtained from normal subjects and 94 chronic schizophrenic patients on medication, 25 chronic schizophrenics off medication, and 15 schizophrenics with no history of medication. These schizophrenic groups showed a high incidence of neurometric features that were significantly deviant from normative values. Multivariate discriminant analysis using these features successfully separated the schizophrenic patients from normals with high accuracy in independent replication. … Newly developed algorithms were used for objective selection of the most effective set of variables for clustering and the optimum number of clusters to be sought. Five clusters were obtained, containing roughly equivalent proportions of the sample with markedly different QEEG profiles.

  14. Electrophysiology and diagnosis in Psychiatry Source analysis Pascual-Marqui, 2002 Notwithstanding the factthatelectrophysiologicalabnormalitieswereshowntoberelatedtodiagnosticsubtypes, riskfactors, symptomdimensions and prognosis, electrophysiologicalmethods are stilloflimited impact in clinicalsettings, and theirapplicationisconfinedto the exclusionof “organic” brainpathology Neuralsynchrony Uhlhaas and Singer, 2010

  15. TranslationtoClinicalSettings Can weexpect a levelofdiagnosticaccuracyhigherthantraditionalclinicalmethods ?

  16. TranslationtoClinicalSettings Highlyunlikely: • The diagnostic algorithm is based on the distinction between patients and controls in the training data, which in turn relies on traditional clinical assessment • The diagnostic application of machine learning techniques to neuroimaging data could be very useful in a forensic setting for example, as an objective means of reducing controversy in evaluations of mental insanity and minimisingerrors in detectingmalingering Sartori et al., CurrOpinNeurol 2011

  17. ContributionsfromNeuropsychology

  18. DoesNeuropsychologycontributetodiagnosis in Psychiatry? • Neuropsychological research in psychiatric disorders provided a large amount of data documenting their presence often before the onset of overt symptomatology and during remission phases • However, so far translation into diagnostic applications is limited to disorders first diagnosed during childhood (mental retardation, learning disorders, communication disorders) and to dementia

  19. MeanNeurocognitiveEffectSizesOrderedByMagnitude and Correctedfor Sample Size

  20. Inclusion of cognitive impairment criteria in Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition) (DSM-V) would not provide a major advancement in discriminating schizophrenia from bipolar disorder and affective psychoses. Therefore, while cognitive impairment should be included in DSM-V, it should not dictate diagnostic specificity. Cognitive impairment as a specifier, rather than an inclusion criterion? Cognitive impairment as a dimension? Bora et al, Schizophrenia Bulletin vol. 36 no. 1 pp. 36–42, 2010

  21. Predicting the diseaseOnset

  22. Predicting Cognitive Decline and PDAT • Identificationofcognitivelynormalsubjects who will convert to MCI in the future (Davatzikos et al., 2009) • Multivariate classification methods based on sMRI that have accuracy of 80% in identifying MCIs who later convert to AD (Teipel et al., 2007; Costafreda et al, 2011) • DTI data seemtoprovide a higherclassificationaccuracyof 98.4%, discriminating between stable MCI patients and those who would suffer progressive decline (Haller et al, 2010)

  23. Structural NeuroimagingSchizophrenia Grey-matter probability maps for baseline comparison of people who developed psychosis with those who did not • The developmentof predictive methods in high-risk, prodromal or first-episode populations appears as a highly promising application Koutsouleris et al., 2009; Sun et al., 2009 Panteliset al, THE LANCET • Vol 361 • January 25, 2003

  24. Disease Prediction in the At-Risk Mental State for Psychosis Using Neuroanatomical Biomarkers: Results From the FePsy Study Koutsouleriset al, Schizophrenia Bulletin, 2011 A positive likelihoodratioof 6.5 in the converters vsnonconverters analysis indicated a 40% increase in diagnostic certainty by applying the biomarker to an at-risk population with a transition rate of 43%. Reliable high-probability voxels contributing to the average ARMS-T vs ARMS-NT ensemble decisionmainly mapped to (1) the dorsomedial, rostromedial, and cingulate cortex, bilaterally, with extensions to the medial orbitofrontal, precuneal, and premotor areas; (2) the dorsolateralprefrontalGMand WM; (3) the right para-hippocampal and inferiortemporalcortex; aswellas (4) the thalamus, bilaterally.

  25. Early Recognition and Disease Prediction in the At-Risk Mental States for Psychosis UsingNeurocognitive Pattern Classification Disease transition was mainly predicted by executive and verballearningimpairments Koutsouleriset al, Schizophrenia Bulletin Advance Access, 2011

  26. ContributionsfromGenetics

  27. The “Missing” HeritabilityofPsychiatricDisorders Genesexplainonly part of the riskofdeveloping a mentaldisorder

  28. HeritabilityofPsychiatricDisorders • Classification in Psychiatry is not informed by the wealth of genetic research in major psychiatric disorders • Important limitations are represented by the high complexity of psychiatric phenotypes, that are much more difficult to objectively “measure”, in comparison to other diseases, and by frequent misclassification of cases, that greatly diminishes the study power Genesexplainonly part of the riskofdeveloping a mentaldisorder

  29. Characteristics • Associatedwithillness • Heritable • Family co-segregation with illness • Found in some unaffected relatives • State independent

  30. The future of genetics in psychology and psychiatry:Microarrays, genome-wide association, and non-coding RNA The greatest value of DNA lies in its ability to predict genetic risk which can lead to preventative interventions. … Genetics can help to target children at genetic risk who are most likely to profit from interventions, which is important because successful prevention programmes usually require extensive and intensive, and thus expensive, interventions Plomin & Davis J Child Psychol Psychiatry 2009

  31. Conclusions • Translationtoclinicalpracticeisverylimited, so far • Diseaseprediction in at risksubjects and earlydiagnosisis a highlypromisingapplicationofneurosciencemethods • The fast developmentofinvestigation and data analysismethods in the fieldofneurosciencepromisesnewavenuesfortranslationalresearch • Diagnosticboundariesmightrequireimportantrevisionstofoster progress in translationalneuroscienceresearch

  32. The future of genetics in psychology and psychiatry:Microarrays, genome-wide association, and non-coding RNA Quantitative genetic research such as twin studies will continue to make importantdevelopmental, multivariate and genotype-environmentcontributionsto psychopathology. … Correlations and interactions between genes and environment will be addressed with much greater precision when specific genes are identified in molecular genetic research Non-coding RNA has changed what we mean by the word gene and underlines the need for genome-wide approaches to gene hunting Genome-wide association (GWA) has come to dominate gene hunting … The first wave of GWA studies indicates that for common disorders and quantitative traits the largest associations are very small The challenge is to identify sets of genes, each of small effect, that can be useful in predicting and preventing problems in childhood Plomin & Davis, J Child Psychol Psychiatry 2009

  33. PredictingResponseto Treatment

  34. Addington & Rapoport, Journal of Child Psychology and Psychiatry 2011

  35. Hyde et al, Trends in Cognitive Sciences September 2011, Vol. 15, No. 9

  36. Potential setting for a clinical application. Imaging data is collected in a specialized imaging center (upper panel) and gold-standard diagnosis is established using labor-intensivemethods such as post-mortem histological examination as gold-standard. Classifier is trained on the imaging data using the gold-standard labels. Trained algorithm is distributed electronically to another clinic (lower panel) and applied to patient data. S. Klöppel et al. / NeuroImage xxx (2011) xxx–xxx

  37. Structural NeuroimagingPattern recognition methods • XXXXXXXXXXXXXXXXX • (Mulleret al., 2001; Rätsch, 2004; Schölkopf and Smola, 2001; Shawe-Taylor and Cristianini, 2004; Vapnik, 1998) and several deal with the specific application to neuroimaging (Bles and Haynes, • 2008; Lemm et al., 2011),

  38. Structural NeuroimagingPattern recognition methods • Training data set • Acquire a sufficient number of training data sets from individual subjects with well-characterisedclinicalproperties (e.g. clinicaldiagnosis, pathologicalmeasures), which can be used as the gold standard for the classification problem at hand • The training data set has to be large enough to reliably express the disease effect against the “noise” of inter-subjectvariability

  39. Structural NeuroimagingPattern recognition methods • Feature extraction from raw data and dimensionality reduction: • The input data (e.g. the anatomical or functional characteristics of the disease process) have to be useful for classification or in other words, they need to be meaningful in the context of the disease and comparable (stable) across subjects • The input data can be as coarse as the total intracranial volume (TIV) or as fine as the amount of grey matter in a very small anatomical region, i.e. a voxel Kloppelet al, NeuroImage 2011

  40. Structural NeuroimagingPattern recognition methods • Feature extraction from raw data and dimensionality reduction: • Oftenbutnot always the numbers of input measures are reduced using various dimensionalityreductionmethods • Dimensionality reduction methods aim at generating a compact set of discriminative “features” that can be used for training the classification model instead of the original input data Kloppelet al, NeuroImage 2011

  41. Structural NeuroimagingPattern recognition methods In the training phase, an algorithm is developed which captures the key differences between groups (e.g. patients vs. controls). In the “testing” phase, the algorithm is used to determine the group to which a new observation belongs to Orrùet al, Neuroscience and BiobehavioralReviews 2012

  42. Structural Neuroimaging • Classifiers can be applied clinically to predict the course of disease in individuals and possibly even the combination of symptoms in the individual patient (Klöppel, 2009).

More Related