To the Editor: Computed tomography (CT) is increasingly used in Russia. Some CT departments are intensely busy, concentrating on outpatient cases. Free medical insurance covers only a small percentage of CT examinations, however; 1 many patients pay for the procedure themselves, and some also prescribe it for themselves without consulting a physician, i.e. practice self-referral. 2 Scanning is usually performed after following a standard protocol and takes several minutes; then the patient

more »

... s and the next one is consulted. Detailed examination and analysis of images is performed by a radiologist in the late afternoon or evening, after the technicians have finished their work. Consequently, a radiologist's working day can last 14 hours or more, which can impair diagnostic quality because of fatigue. Such a workflow does not permit immediate re-examination with enhanced resolution or modified scanning parameters, which can be indicated for an unclear lesion or a group of lesions -for example, a pneumonia-like infiltrate or dissemination in a lung. 3 One of the reasons for this workflow pattern is a shortage of upto-date literature and insufficient acquaintance with foreign experience. 4 Misleading statements can be found in some domestic publications; for example: 'CT has become three-dimensional, which excludes missing small pathologic lesions and structures' . 5 Such generalisations tend to maintain a belief in the unlimited capabilities of 'computer diagnostics' in order, among other things, to enhance the number of self-referred patients. As a result, an expensive procedure accompanied by X-ray exposure is sometimes performed without justifiable indications. Some excerpts from one of the most broadly used Russian manuals (translated verbatim): '... uninterrupted scanning cycle of the spiral CT allows image reconstruction at any level ... In the spiral CT, a level of reconstruction does not depend on the main scanning parameters such as speed of the table or gantry positioning. As the scanning occurs uninterruptedly, the level of each section and the distance between adjacent sections are selected optionally by an operator not before but after the scanning ... Possibility of optional positioning of sections along the scanning axis and arbitrary choice of the section width allow forming blocks of partially overlapping images, whereas degree of mutual overlapping is practically unlimited ... In conventional CT, a similar effect can be achieved only if the table feed per gantry rotation would be less than the thickness of a tomographic layer ... In spiral CT, mutual overlapping of the slides is independent of the scanning parameters being a post-processing procedure ... location of pathological lesions between the tomographic layers, as it can occur in conventional CT, is therefore excluded' . 6 The impression is created by such assertions that a computer can reconstruct an image that is representative of the morphological substrate, on any level from the first to the last scanning plane. This notion does not take into account that a computer disposes only of the data that have been obtained as a result of radiodensity averaging of the tissue layers encompassed by the X-ray beam. If the pitch is >1, 'the patient would have a candy-striped appearance with unmarked flesh between ribbons of paint' . 7 Another quotation (verbatim from the Russian): ' A result of the spiral scanning is an uninterrupted data volume, which can be arbitrarily subdivided into a required number of slides of optional thickness' . 6 Note that information flows uninterruptedly along a spiral line but not along the axis of table movement (z-axis), where information is intermittent and subdivided by intervals if the pitch is >1. A reconstruction algorithm interpolates the data from adjacent sections into the spaces between them. These spatial relationships can be illustrated by a plane cutting a helicoid. Overextension of the spiral, and excessive elevating of the pitch and/or collimation, can cause inadequate visualisation of pathological lesions. It is sometimes argued that spatial resolution is defined only by pixel size (Tiurin IE, personal communication). One must note that pixel (or voxel) size is a characteristic of the equipment, while spatial resolution is defined also by information density per volume unit of the scanned tissue, which in turn depends on the scanning parameters. With high pitch values, the volume of interest is under-sampled. 8 All the above applies also to the multi-slice spiral CT, although spatial relationships here are more complicated: collimation values of a single section and of the whole detector array are distinguished, with 2 corresponding pitch values. Therefore, information density (as well as X-ray exposure) depends on the table feed per gantry rotation and on the distance between adjacent sections. The issue of radiation exposure in multi-slice spiral CT is beyond the scope of this letter, but the question should be posed in principle: Under which conditions is the enhanced X-ray exposure caused by crossings of beam trajectories with repeated exposure of the same tissue areas compensated by the advantages of multi-slice spiral CT, such as isotropy and high resolution volumetric data? 9 CT technology is progressing -scanning time is being reduced and image quality improved. This is no reason, however, to discard the principle of image analysis that is common to radiological and microscopic methods: in the case of an unclear lesion, one should be able to go to a 'high magnification' i.e. to repeat an examination with necessary adjustment to scanning parameters. It is therefore advantageous, when all images are examined by a radiologist, if the patient is still in the office, so that secondary scanning can be performed, if indicated. The total time per patient would probably be lower than in delayed image viewing, because a radiologist would not have to delve twice into a case. To maintain the same productivity, reception hours and, correspondingly, the working time of technicians should be prolonged. The great difference between the length of the working day for technicians and radiologists seems to be specific to Russia, as well as the relatively low remuneration for radiologists (<10% of the amount paid by the patient or insurer). Additionally, it has been reported that CT productivity is enhanced when several technicians work simultaneously. 10 The workflow described in the first paragraph (scanning performed after a standard protocol, and images viewed after reception hours) can be justified for screening purposes. However, the usefulness of CT screening and examination to selfreferred patients is questionable. 2,11 A concluding point is that clinicians in modern hospitals have access to CT images via a computer network. 12 But in many Russian hospitals, they can obtain only images on an Xray film and a verbal description. Overall, however, there are grounds for optimism: the improved economy now makes it possible to acquire modern equipment and literature, while broadening international cooperation will attract foreign expertise into the country.