Dementia, depression and the CT scan.

Hounsfield's Nobel prize for the invention of computed tomography (CT) was well-deserved recognition of a major advance in medical technology, an innovation which produced immediate rewards in the diagnosis and management of space-occupying lesions (SOL), and justified high capital costs (Jacobsen et al. 1976; Thomson, 1977). However, for psychiatry, which is more concerned with trophic changes than SOL, the benefits of CT have not been fully defined. In dementia syndromes, for example, CT constitutes a marked improvement on pneumoencephalography (PEG) because of a greater and more easily obtained yield of information, but the clinical value of demonstrating cerebral atrophy in demented patients has not been fully established. What information did PEG yield ? The answer is not straightforward mainly because of methodological deficiencies, some of which, such as the lack of normative data, were due to ethical constraints. Thus, Burhenne & Da vies (1963) admitted that the 'normal' appearance on PEG is necessarily what the experienced radiologist considers not to be abnormal. Information from PEG was further limited by the risk of exacerbating dementia by the procedure itself. Nevertheless, it was agreed that clinically demented patients showed significantly more radiological atrophy than the non-demented (Chodoff et al. 1948; Wagner, 1951; Trolle & Fog, 1951; Haug, 1962). In a follow-up study from a neurological hospital, Gosling (1955) showed that dementia was unlikely to be confirmed in the absence of radiological atrophy on clinical diagnosis. A consensus was also reached for the relative importance of ventricular enlargement over cortical (surface) atrophy (Chodoff et al. 1948; Wagner, 1951; Nielsen et al. 1966 a, b), an observation confirmed by Mann (1973) in a follow-up study. In general, these findings were not invalidated by evidence for age-related ventricular enlargement (Burhenne & Davies, 1963; Booker et al. 1969). An association between PEG atrophy and cognitive impairment was also demonstrated (McFie, 1960; Kiev et al. 1962; Nielsen et al. 1966a, b), but common psychological characteristics associated with cerebral atrophy could not be identified (Mathews & Booker, 1972). Prima facie direct visualization of the brain must be superior to merely outlining it with air, an opinion confirmed by the evidence. However, Jacoby et al. (1980) criticized the failure to realize the opportunity afforded by CT to obtain true normative data, a criticism based on the extensive use of patients found not to be abnormal rather than of healthy volunteers (Huckman et al. 1975; Gyldensted & Kosteljanetz, 1976; Synek et al. 1976; Haug, 1977). More recently, data from non-patient controls have been published (Ron et al. 1979 a; Jacoby et al. 1980). The clinical introduction of CT (Ambrose, 1973) was followed nastily by enthusiastic statements about its value in the diagnosis of dementia (Huckman et al. 1975; Fox et al. 1975; Menzer et al. 1975). Some of these authors have now confessed to unwarranted optimism (Fox et al. 1979). As with PEG, groups of demented patients exhibited significantly more atrophy than controls, but dementia occurred without atrophy and atrophy without dementia. In one study 16 % of controls were found on blind rating to have enlarged ventricles, whereas 25 % of dements had normal ventricles (Jacoby & Levy, 1980a). Overlap between the groups was confirmed by mean ventricle: skull ratio (dements 20-6 ± 5 0 S.D. cf. controls 14-2 ± 3-9 S.D.), although the difference between these means was highly significant (P < 0001). Discriminant function analysis yielded an optimum prediction table in which 17% of cases were incorrectly assigned to their diagnostic group. Although SOL tend to present to the neurologist, their exclusion in psychiatric patients is important as unforeseen tumours in psychiatric populations are well known (Waggoner & Bagchi, 1954; Kraft et al. 1965). Jacoby & Levy (1980a) found a corpus callosum glioma and a subdural haema-