The role of high resolution computed tomography in the diagnosis of interstitial lung disease.

Until recently the chest radiograph has been the only imaging technique used in the assessment of patients with suspected diffuse lung disease. In this context the chest radiograph is less than ideal. Problems arise with both false negative and false positive results. It is well established that the chest radiograph may appear entirely normal in up to 10% of patients with biopsy proved diffuse lung disease of various causes,1 and a poor quality chest radiograph, especially of an obese patient, may misleadingly raise the spectre of diffuse lung disease. The two dimensional nature of a chest radiograph dictates that there is superimposition of structures over the lungs and it has been estimated that up to 40% of the lungs is obscured in normal subjects. This obscuration will be further compounded by the presence of pleural thickening or an effusion. Even when the chest radiograph shows definite evidence of diffuse lung disease McCloud has pointed out that "the chest radiograph is often non-specific. Various radiographic patterns, together with their predominant location, correlate statistically with their pathologic entities but in individual cases the chest roentgenogram is rarely diagnostic. "' Until recently the main use for computed tomography has been detection of pulmonary nodules and, despite some pioneering work by Kreel in the early 1980s,3 it has had a limited role in the investigation of diffuse lung disease. Improvements in computed tomography scanner technology, notably in terms of spatial resolution and shorter scan times, have led to renewed interest in the application of computed tomography to show the fine morphological detail of the lung. Research in this area is particularly active and is reflected in the large volume of descriptive work on the computed tomographic appearances of many diffuse lung diseases that has appeared over the last five years4; there are no signs that this period of research is over. The technique of high resolution computed tomography, used to show the lung parenchyma,' is sufficiently different from conventional computed tomography to warrant a brief description. Conventional computed tomography of the thorax uses contiguous 1 cm sections; in this way the entire lungs are included in one study. This protocol is widely used when a comprehensive examination of the lungs is required-for example, in the search for metastases. The volume averaging that occurs within the 1 cm thickness of the scan, however, substantially reduces the ability of conventional computed tomography to resolve small structures. For high resolution computed tomography the section thickness is reduced to 1-3 mm and a different software reconstruction of the image is used to improve spatial resolution (figs la and lb). These scans are interspaced by at least 1 cm. Where 3 mm sections are taken 1 cm apart the radiation dose to the breast is reduced to about half that of conventional computed tomography. None the less, the radiation burden inherent in high resolution computed tomography is considerable and in a complete study using 3 mm sections every 1 cm the dose is roughly 40 times that received from a single chest radiograph.6 Because high resolution computed tomography includes only very short segments of pulmonary vessels these narrow sections may be misinterpreted as showing a nodular pattern. For this reason some radiologists advocate conventional computed tomography before high resolution computed tomography but this obviously adds to the radiation dose. Because, by definition, diffuse lung disease is widespread many radiologists now perform a limited number of high resolution computed tomography scans, perhaps as few as six, in such patients. In this way not only is the radiation dose restricted but the cost of the examination is not much more than that of a posteroanterior and lateral chest radiograph. It has been estimated in the United States that high resolution computed tomography used in this way costs $180, compared with $90 for a posteroanterior and lateral chest radiograph. In the same comparison a transbronchial biopsy with bronchoalveolar lavage costs $1500 and an open lung biopsy $5000 (R A Webb, personal communication). Computed tomography is becoming more widely available throughout the United Kingdom and most computed tomography machines installed over the last three years are able to provide high resolution images. The spatial resolution of the latest generDepartment of Radiology, Royal Brompton National Heart and Lung Hospital, London SW3 6HP D M Hansell I H Kerr