Image segmentation of organs in medical images using model-based approaches requires a priori information which is often given by manually tagging landmarks on a training set of shapes. This is a tedious, time-consuming, and error prone task. To overcome some of these drawbacks, several automatic methods were devised. Identification of the same homologous set of points in a training set of object shapes is the most crucial step in Active Shape Modelling, which has encountered several challenges. The most crucial among these are: (C1) defining and characterizing landmarks; (C2) obtaining landmarks at the desired level of detail; (C3) ensuring homology; (C4) generalizing to n>2 dimensions; (C5) achieving practical computations. This thesis proposes several novel modelling techniques attempting to meet C1-C5. In this process, this thesis makes the following key contributions: the concept of local scale for shapes; the idea of allowing level of detail for selecting landmarks; the concept of equalization of shape variance for selecting landmarks; the idea of recursively subdividing shapes and letting the sub-shapes guide landmark selection, which is a very general n-dimensional strategy; the idea of virtual landmarks, which may be situated anywhere relative to, not necessarily on, the shape boundary; a new compactness measure that considers both the number of landmarks and the number of modes selected as independent variables. The first of three methods uses the c-scale shape descriptor, based on the new concept of curvature-scale, to automatically locate mathematical landmarks on the mean of the training shapes. The landmarks are propagated to the training shapes to establish correspondence among shapes. Since all shapes of the same family do not necessarily present exactly the same shape features, another novel method was devised that takes into account the real shape variability existing in the training set and that is guided by the strategy of equalization of the variance observed in the training set for selecting landmarks. By incorporating the above basic concepts into modelling, a third family of methods with numerous possibilities was developed, taking into account shape features, and the variability among shapes, while being easily generalized to the 3D space. Its output is multi-resolutional allowing landmark selection at any lower resolution trivially as a subset of those found at a higher resolution. The best strategy to use within the family will have to be determined according to the clinical application at hand. All methods were evaluated in terms of compactness on two data sets - 40 CT images of the liver and 40 MR images of the talus bone of the foot. Further, numerous artificial shapes with known salient points were also used for testing the accuracy of the proposed methods. The results show that, for the same number of landmarks, the proposed methods are more compact than manual and equally spaced annotations. Besides, the accuracy (in terms of false positives and negatives and the location of landmarks) of the proposed shape descriptor on artificial shapes is considerably superior to a state-of-the-art scale space approach to finding salient points on shapes.

The studies presented in this thesis investigated the neural correlates of attention in touch. In particular, the electrophysiology of exogenous tactile processing and inhibition of return (IOR) - an area previously unexplored. In all studies a variation of the Posner cue-target paradigm was used. Typically, a cue was presented to the left or right hand. Following a stimulus onset asynchrony of 800 ms, a target would appear at the same or opposite hand. Behavioural results consistently demonstrated IOR when employing a simple target detection task, showing that IOR is a reliable phenomenon in touch. The concurrently recorded event related potentials (ERPs) demonstrated an early attention modulation of the N80 in all studies presented in this thesis, regardless of the presence or absence of IOR. This early component likely reflects processing of the exogenous lateralized cues. Following the N80, the attention modulations varied across studies. The conclusion to be drawn from this thesis is that not one particular ERP component is directly associated with IOR. Analysis of endogenous tactile attention (Chapter V) demonstrated modulations at the N140 and Nd components. Moreover, correlation analysis showed that larger ERP attention modulation was associated with a larger behavioural effect, demonstrating a novel relationship between ERP modulations and response time effects. Analysis of the cue-target interval has previously only been investigated during endogenous orienting. Here, and for the first time, an anterior directing attention negativity (ADAN) was demonstrated during exogenous orienting. This ADAN was unaffected by varying posture suggesting exogenous tactile attention and IOR are somatotopically coded. Indications of an external frame of reference were only demonstrated during shifts of endogenous attention, as indicated by the presence of a late directing attention positivity (LDAP) (endogenous counter-predictive task presented in Chapter V). The final study of this thesis (Chapter VI) demonstrated that varying visual perceptual load influenced tactile processing. Specifically, high perceptual load led to elimination of IOR. Moreover, the P100 for irrelevant tactile stimuli was significantly reduced in high versus low load condition. This suggests perceptual load may suppress irrelevant tactile stimuli relatively early (around 100 ms post stimuli onset) during tactile processing. Taken together, this thesis presents a series of experiments which map out effects of endogenous and exogenous attention and how these mechanisms interact, both through behaviour and underlying neural correlates.

We live in a world which bombards us with a huge amount of sensory information, even if we are not always aware of it. To successfully navigate, function and ultimately survive in our environment we use all of the cues available to us. Furthermore, we actually combine this information: doing so allows us not only to construct a richer percept of the objects around us, but actually increases the reliability of our decisions and sensory estimates. However, at odds with our naturally multisensory awareness of our surroundings, the literature addressing unisensory processes has always far exceeded that which examines the multimodal nature of perception. Arguably the most salient and relevant stimuli in our environment are other people. Our species is not designed to operate alone, and so we have evolved to be especially skilled in all those things which enable effective social interaction – this could be engaging in conversation, but equally as well recognising a family member, or understanding the current emotional state of a friend, and adjusting our behaviour appropriately. In particular, the face and the voice both provide us with a wealth of hugely relevant social information - linguistic, but also non-linguistic. In line with work conducted in other fields of multisensory perception, research on face and voice perception has mainly concentrated on each of these modalities independently, particularly face perception. Furthermore, the work that has addressed integration of these two sources by and large has concentrated on the audiovisual nature of speech perception. The work in this thesis is based on a theoretical model of voice perception which not only proposed a serial processing pathway of vocal information, but also emphasised the similarities between face and voice processing, suggesting that this information may interact. Significantly, these interactions were not just confined to speech processing, but rather encompassed all forms of information processing, whether this was linguistic or paralinguistic. Therefore, in this thesis, I concentrate on the interactions between, and integration of face-voice paralinguistic information. In Chapter 3 we conducted a general investigation of neural face-voice integration. A number of studies have attempted to identify the cerebral regions in which information from the face and voice combines; however, in addition to a large number of regions being proposed as integration sites, it is not known whether these regions are selective in the binding of these socially relevant stimuli. We identified firstly regions in the bilateral superior temporal sulcus (STS) which showed an increased response to person-related information – whether this was faces, voices, or faces and voices combined – in comparison to information from objects. A subsection of this region in the right posterior superior temporal sulcus (pSTS) also produced a significantly stronger response to audiovisual as compared to unimodal information. We therefore propose this as a potential people-selective, integrative region. Furthermore, a large portion of the right pSTS was also observed to be people-selective and heteromodal: that is, both auditory and visual information provoked a significant response above baseline. These results underline the importance of the STS region in social communication. Chapter 4 moved on to study the audiovisual perception of gender. Using a set of novel stimuli – which were not only dynamic but also morphed in both modalities – we investigated whether different combinations of gender information in the face and voice could affect participants’ perception of gender. We found that participants indeed combined both sources of information when categorising gender, with their decision being reflective of information contained in both modalities. However, this combination was not entirely equal: in this experiment, gender information from the voice appeared to dominate over that from the face, exerting a stronger modulating effect on categorisation. This result was supported by the findings from conditions which directed to attention, where we observed participants were able to ignore face but not voice information; and also reaction times results, where latencies were generally a reflection of voice morph. Overall, these results support interactions between face and voice in gender perception, but demonstrate that (due to a number of probable factors) one modality can exert more influence than another. Finally, in Chapter 5 we investigated the proposed interactions between affective content in the face and voice. Specifically, we used a ‘continuous carry-over’ design – again in conjunction with dynamic, morphed stimuli – which allowed us to investigate not only ‘direct’ effects of different sets of audiovisual stimuli (e.g., congruent, incongruent), but also adaptation effects (in particular, the effect of emotion expressed in one modality upon the response to emotion expressed in another modality). Parallel to behavioural results, which showed that the crossmodal context affected the time taken to categorise emotion, we observed a significant crossmodal effect in the right pSTS, which was independent of any within-modality adaptation. We propose that this result provides strong evidence that this region may be composed of similarly multisensory neurons, as opposed to two sets of interdigitised neurons responsive to information from one modality or the other. Furthermore, an analysis investigating stimulus congruence showed that the degree of incongruence modulated activity across the right STS, further inferring neural response in this region can be altered depending on the particular combination of affective information contained within the face and voice. Overall, both behavioural and cerebral results from this study suggested that participants integrated emotion from the face and voice.

Deliberate self-harm is a significant phenomenon amongst people in the general community, and is particularly prevalent amongst patients being treated in mental health inpatient settings. Views that staff hold towards patients who self-harm could have an impact upon the care and interventions that are delivered. The focus of this thesis was deliberate self-harm within inpatient mental health settings. The literature review presented in Chapter I provides an overview of previous reviews of studies that have evaluated psychological interventions for deliberate self-harm. All recent relevant evaluative studies of psychological interventions predominantly aimed at reducing deliberate self-harm, or treating self-harm as part of the symptomatology of Borderline Personality Disorder, arnongst inpatient enviromnents are then critically reviewed. The results of this are discussed along with clinical implications for practitioners working in mental health inpatient settings and recommendations for future research. The focus moves to staff attributions towards deliberate self-hann in inpatient settings in Chapter 2. This chapter presents empirical findings, reporting on adaptations of attributional and knowledge measures, and analyses of responses to these measures provided by qualified and unqualified nursing staff participants working in an inpatient setting. Questions regarding training needs were also posed, and participants were given the opportunity to comment on working with people who self-harm. Chapter 3 presents a reflective paper, incorporating references from literature, poetry and music, providing reflections on producing the first two chapters of the thesis, and on the overall experience of completing research for the clinical psychology doctorate course.

This thesis consists of three papers: a literature review, an empirical paper and a reflective paper. The literature review considers the validity of measures of premorbid intellectual functioning with people with varying levels of dementia severity. It aims to be of use to British clinicians by focussing specially on tests that were designed for use with a British English speaking population. No single measure of pre-morbid intellectual functioning is found to be valid for use with people of all levels of dementia severity. The use of multiple measures for estimating pre-morbid intellectual functioning are indicated. The empirical paper investigates the relationship between a person’s pre-morbid level of intellectual functioning and the amount of cognitive decline that they experience prior to gaining a diagnosis of dementia. It uses a retrospective correlational design to analyse data from existing memory clinic files. A significant positive Pearson’s correlation coefficient is found between pre-morbid intellectual functioning and amount of cognitive decline at the point of diagnosis, in all three of the diagnostic groups. This suggests that people with high pre-morbid functioning have to undergo greater cognitive decline before they are given a diagnosis of dementia. Changes to current clinical practice are indicated which take account of relative amounts of cognitive decline in the diagnostic process. Finally, a reflective account is presented which is based on the author’s thoughts and experiences of carrying out research in the field of dementia. This focuses on the challenges in accessing retrospective clinical data. Suggestions are made regarding future research and clinical practice.

Ascending pathways in the spinal cord are very important in transmitting sensory information from the periphery to the higher brain centres. The spinothalamic tract represents an important component of these ascending pathways, and it has been traditionally described as the main pathway for conveying nociceptive and thermoceptive information. Spinothalamic neurons are widely distributed within the grey matter. Lamina I represents an important nociceptive zone and provides a significant source of projection neurons, some of which project to the thalamus. A projection from cervical superficial dorsal horn to the posterior triangular thalamic nucleus (PoT) in the rat has recently been identified. The PoT is located at the caudal end of the thalamus and was not included in injection sites in many previous quantitative retrograde tracing studies of the spinothalamic tract. Therefore, one of the reasons to conduct the present study was to estimate the number of the spinothalamic cells in lamina I in rat cervical and lumbar enlargements following injections that target the PoT with or without other thalamic nuclei known to receive input from lamina I. Neurons in this lamina are also known to project to the lateral parabrachial nucleus (LPb) and the periaqueductal grey matter (PAG). Other aims of the study were to quantify neurons in lamina I in the cervical enlargement that project to the LPb and PAG, to determine the proportion of lamina I spinothalamic neurons in lumbar and cervical enlargements that could be labelled from LPb and PAG, and to investigate morphological differences between different projection populations. Recent investigations have identified a group of neurons in lamina I of rat lumbar spinal cord that had large numbers of puncta that were immunoreactive for the glycine receptor-associated protein, gephyrin, and have a very high density of input from glutamatergic axons that contain vesicular glutamate transporter 2 (VGLUT2). These "large gephyrin-coated cells" in the lumbar cord are known to project to the LPb, but it is not yet known whether they project to thalamus and PAG. Therefore, another aim was to determine whether these cells project to these areas and also to analyse the projection pattern of cells of this type in the cervical enlargement. Previous studies have identified a population of large neurons in laminae III and IV of rat spinal cord that express the neurokinin 1 receptor (NK1r) and have prominent dorsal dendrites that enter the superficial laminae. A substantial body of evidence points to the involvement of this population of cells in processing various types of noxious stimulus. Neurons of this type in lumbar enlargement are projection cells and form a major route through which nociceptive information reaches the brain. The proportion of these neurons that project to thalamus was not previously known, and the projection pattern of cells of this type in cervical enlargement has not yet been investigated. Therefore, an additional aim was to elucidate more on the projection patterns of these cells in both enlargements. Various tracers (cholera toxin B subunit, Fluorogold or fluorescent latex microspheres) were injected stereotaxically into thalamus (25 rats), into thalamus and LPb (3 rats), or into thalamus and PAG (4 rats). Rats were perfused after three days and sections from the spinal cord (cervical and lumbar enlargements) were processed immunocytochemically to reveal tracer(s) in lamina I and lamina III/IV neurons, the NK1r, neuronal nuclei and, in some cases, the glycine receptor-associated protein gephyrin. Sections from brains were processed to visualise the injection sites. Results of this study showed that: 1) most lamina I spinothalamic neurons in the C7 and L4 segments could be labelled from injections centred on the PoT; 2) the estimated total numbers of spinothalamic cells in lamina I on the contralateral side of the C7 and L4 segments are 91 and 16 cells, respectively, and this constitutes 2-3% and 0.2% of the total neuronal population in lamina I in the C7 and L4 segments, respectively; 3) the C7 segment contained fewer lamina I spinoparabrachial cells, but a similar number of spino-PAG cells, compared to L4; 4) virtually all spinothalamic lamina I neurons at both cervical and lumbar levels were labelled from LPb and between a third and a half were labelled from PAG; 5) spinothalamic lamina I neurons differed from those labelled only from LPb in that they were generally larger, more often multipolar and (in cervical enlargement) had stronger NK1r-immunoreactivity; 6) ~39% of "large gephyrin-coated cells" in L5 project to the thalamus and this accounts for ~21% of the total thalamic projection from lamina I in this segment, even though these cells constitute only ~2.5% of projection neurons in lamina I; 7) the great majority of "large gephyrin-coated cells" in C6 project to thalamus and LPb, and at both segmental levels, some project to both of these areas; 8) only few "large gephyrin-coated cells" in L5 and some of those in C6 project to PAG; 9) ~84% of the lamina III/IV NK1r-immunoreactive neurons in C6 and C7 and 17 28% of those in L4 and L5 belong to the spinothalamic tract, and these apparently project exclusively to the caudal thalamus, including PoT; 10) most of the large NK1r-immunoreactive lamina III/IV cells at both levels project to LPb, but few were labelled from PAG, and at both segmental levels, some project to both thalamus and LPb. Findings from the present study indicate that the PoT is one of the major targets for neurons in lamina I as well as to the population of the NK1r-immunoreactive neurons in laminae III and IV. Since the PoT projects to the second somatosensory and insular cortices, the present results suggest that these are major targets for information conveyed by both these populations of spinothalamic neurons. In addition, these results confirm that projection neurons have extensive collateral projections, and suggest that different sub-populations of lamina I cell have characteristic patterns of supraspinal projection.

In the first part of my PhD I explored the link between childbirth and mood disorders in a retrospective sample of over 1500 parous women with mood disorders, recruited as part of ongoing molecular generic studies. Around two thirds of participants reported at least one episode of illness during pregnancy or the postpartum. Women with bipolar I disorder reported an approximately 50% risk of a perinatal major affective episode. Risks were lower in recurrent major depression and bipolar II disorder at around 40%. The majority of perinatal episodes occurred within 4 weeks of childbirth. Episodes of mania or psychosis had an earlier onset than those of depression. For bipolar II disorder, onsets of psychiatric episodes were more spread out over the perinatal period with more onsets in pregnancy and later in the postpartum. Moreover, childbirth did not seem to be a specific trigger for the majority of perinatal episodes of bipolar II disorder. Primiparity was associated with postpartum mania/psychosis and unipolar postpartum depression in women who experienced their first lifetime episode within 6 weeks postpartum. My findings raise the possibility of a relationship between postpartum mood disorders and other disorders influenced by parity, such as pre-eclampsia. In the second part of my PhD I designed and piloted a prospective study aimed i) to replicate and ii) to extend the findings on the retrospective sample, exploring the influence of a range of variables on the vulnerability to develop an episode of severe illness in pregnancy or the postpartum. Over 14 months of recruitment 19 women completed the follow-up assessment. To capture the clinical complexity of bipolar disorder in pregnancy and the postpartum period very large scale longitudinal studies are needed. These studies must be based on a strong collaboration with the NHS.

No disease-modifying compounds are available to halt disease progression in Alzheimer’s disease (AD). Immunotherapy offers promising possibilities for the manipulation of Aβ levels which the amyloid cascade hypothesis proposes as the causative factor in AD. However, antiAβ antibodies have caused inflammation in vivo. An alternative antibody (2B3) which targets the β-secretase cleavage site of the amyloid precursor protein (APP) from which Aβ is cleaved has been shown to downregulate Aβ in human cell lines. The approach is thought unlikely to cause inflammation as the immune system is not relied upon for Aβ clearance. It was hypothesised that the administration of 2B3 to aged transgenic APP mutation mice would lower Aβ levels through the inhibition of Aβ production, with an associated lowering of cognitive deficits. Two murine APP mutation models [London APP(V717I) and Indiana PDAPP(V717F)] were characterised in order to identify cognitive deficits against which the ability of 2B3 to reduce deficits could be assessed. APP(V717I) mice were assessed in the marble burying task, the elevated plus maze and a foraging task assessing spatial working memory at 3, 6 and 19 months of age. The radial arm water maze was carried out at 10 months, before the T-maze non-matching to position task was administered at 11 months. Object recognition memory was assessed at 18 months. Similarly, PDAPP mice were assessed using the marble burying and elevated plus maze (9.5months), the T-maze (9-10 months), the foraging task (11 and 14 months) and the object recognition task (12 months). Whilst aged transgenic PDAPP mice displayed disrupted spatial working memory, no evidence of agerelated cognitive decline was observed in APP(V717I) transgenic mice despite increases in Aβ pathology with age. 2B3 did not alter Aβ levels or spatial working memory in PDAPP(V717F) mice in a pilot study, whilst the in vitro downregulation of Aβ was successfully replicated in primary murine neurons. The findings indicate that transgenic models of neurodegenerative disease require thorough characterisation to optimise their use in pre-clinical research. Furthermore, the use of alternative immunotherapy in the treatment of AD remains a promising, but early stage avenue of study.

Introduction: Values have a strong tradition in social psychology, but until recently they have been largely neglected in mental health literature. More recently, the importance of values has been recognised by some psychological therapies (e.g. Acceptance and Commitment Therapy); however, the relative importance of values in mental health problems has not been empirically tested. Aims: The current research aimed to investigate the value priorities of people with anxiety and eating disorders, and to assess the relationship between value discrepancies and distress, and in doing so to draw upon Schwartz’s (1992) model of values and Higgins’s (1987) self-discrepancy theory. More specifically, it investigated whether people with anxiety and eating disorders differ from people without mental health problems in the values that they hold, and the level of value discrepancies in these values, and additionally whether these discrepancies were associated with anxiety and depression. Methods: A cross-sectional questionnaire-based design was employed, with data being collected from 122 participants (an anxiety disorder group, n=30; eating disorder group, n=31; and reference group n=61). Multivariate statistics, paired sample t-tests and Pearson’s correlations were used to test the hypotheses. All participants completed a measure assessing values and discrepancies in values (adapted PVQ), and the mental health groups also completed a measure assessing psychological distress (HADS). Results: The reference group rated particular values (e.g. self-direction, stimulation, hedonism) as more important than did the mental health groups, apart from the achievement value, which the eating disorders group rated as more important. The mental health groups had higher value discrepancies than the reference group. Actual-Ideal and Actual-Ought value discrepancies were found to be related to anxiety and depression. However, unexpectedly, depression was found not to be specifically associated with Actual-Ideal discrepancies and anxiety was found not to be specifically associated with Actual-Ought discrepancies. Conclusions: This study provides empirical support and evidence for considering the values that people with mental health problems hold and the role that values has in relation to the psychological distress experienced by people. The results are discussed with reference to existing literature and the strength and limitations of the research were outlined. In addition, the clinical limitations were discussed and ideas for future research were outlined.

Work towards the development of a new computer-assisted methodology for psychological study and intervention is described. This is referred to as the Personal Knowledge Methodology since it focuses on the elicitation and presentation of personal knowledge. Personal knowledge includes the knowledge individuals have of their life history, their behaviours, their moods, their relationships, their ambitions, and so on. Principles and techniques used in Knowledge Engineering form the basis of the design of the Personal Knowledge Methodology. At the heart of the methodology is the use of a suite of knowledge acquisition and modelling techniques. These are referred to as Personal Knowledge Techniques. Based on a review of a wide-range of literature, eight techniques were selected to be assessed for their possible use as Personal Knowledge Techniques. These included interview-based techniques, repertory grid techniques and diagram-based techniques. Two in-depth studies took place involving 18 participants and a total of 100 knowledge acquisition sessions. The results revealed that each of the eight techniques showed promise at efficiently capturing and structuring aspects of an individual’s personal knowledge. In addition, the techniques showed potential for providing help in allowing reflection and revealing insights. In particular, a technique based on the construction and use of a state transition network was found to be the most highly rated by the participants. A content analysis of the knowledge acquired formed the basis of an ontology of personal knowledge that would underpin many uses of the Personal Knowledge Methodology. The empirical work and analysis led to a number of ideas for future developments of the methodology and uses for the Personal Knowledge Techniques.