Editorials

High resolution computed tomography of the lungs

BMJ 1995; 310 doi: http://dx.doi.org/10.1136/bmj.310.6980.616 (Published 11 March 1995) Cite this as: BMJ 1995;310:616
  1. Sylvia Worthy
  1. Senior registrar in radiology Royal Victoria Infirmary, Newcastle upon Tyne NE1 4LP

    Better than chest radiography for some conditions

    What you see on high resolution computed tomography of the lungs is as detailed as what you see when you look at a gross pathological specimen.1 It is the most accurate non-invasive method of evaluating lung parenchyma2 and has improved our understanding of the patterns and pathology of many pulmonary diseases.

    In conventional computed tomography slices are 10 mm thick and scans are obtained at 10 mm intervals; in high resolution computed tomography slices are 1-2 mm thick and scans are obtained at 10-40 mm intervals. The images therefore represent only one tenth or less of the volume of the lung. This sampling is suited to diffuse processes affecting extensive areas of the lung but will clearly miss small lesions lying between slices. High resolution computed tomography can resolve an object of 0.5 mm diameter and is ideally suited for use in the lungs, where a high contrast between tissue and air exists.2

    High resolution computed tomography comes into its own in showing the secondary pulmonary lobule, which is the site of characteristic changes in many lung diseases. Although chest radiography remains the preferred imaging technique in the assessment of patients with lung disease, two features limit its sensitivity and specificity for detecting subtle lung disease. Firstly, small differences in attenuation between normal and abnormal lung tissue are difficult to observe in a chest radiograph so minor parenchymal abnormalities are not readily detected. Secondly, many structures, both normal and abnormal, are superimposed in a radiograph, making diagnosis difficult.3 By contrast, high resolution computed tomography provides more accurate information about the anatomical site of abnormalities which is less subject to technical variations and variations in interpretation than information in plain chest radiographs.

    Several studies have shown that high resolution computed tomography is better than conventional computed tomography in assessing patients with chronic diffuse infiltrative lung disease.4 5 One study found that, although both modalities were equivalent in recognising nodules, masses, nodular irregularities of the interfaces, large cystic air spaces, and architectural distortion, high resolution computed tomography showed fine bronchial and parenchymal lesions better and allowed the assessment of changes in ground glass attenuation (important in identifying the early treatable stage of fibrosis).6

    Figure1

    Idiopathic pulmonary fibrosis. High resolution computed tomography through the lung bases in a 64 year old man shows honeycombing in a typical subpleural distribution and interlobular septal thickening

    Concern has been expressed over the relatively high doses of radiation used in computed tomography: with conventional imaging of the thorax patients receive a radiation dose 100 times that received in standard chest radiography7; high resolution computed tomography of the thorax delivers a radiation dose 10-20 times that received in standard chest radiography,8 and further reduction in radiation dose is possible without substantially reducing the resolution of the image.9 In patients with chronic diffuse lung disease a limited number of high resolution computed tomographic sections can be combined with a low dose technique, providing more diagnostic accuracy than a chest radiograph with no increase in the radiation dose.10

    High resolution computed tomography is particularly valuable in assessing patients with a normal chest radiograph but clinical symptoms and abnormalities of lung function indicating diffuse infiltrative lung disease.11 Comparative studies in patients with disease proved by biopsy have shown that high resolution computed tomography has a sensitivity of about 94% for the detection of chronic infiltrative lung disease, compared with 80% for chest radiography.12 It has a valuable role in patients with questionable radiographic abnormalities because confidently identifying normal subjects is easier than with chest radiography.13 High resolution computed tomography defines the pattern and anatomical location of lung parenchymal abnormalities, which in many cases are sufficiently characteristic to provide a specific diagnosis. These include lymphangitis carcinomatosa, sarcoidosis, lymphangioleiomyomatosis, and pulmonary fibrosis.14 15

    High resolution computed tomography is now the preferred diagnostic test for bronchiectasis, having replaced bronchography. Its sensitivity in detecting emphysema approaches that of gross pathological examination.16 In patients with fibrosing alveolitis it can distinguish active, potentially reversible or treatable disease from irreversible fibrosis. Two distinct patterns of abnormality are seen: ground glass opacity (corresponding to a cellular histological appearance at biopsy) and a reticular pattern (corresponding to fibrotic areas on histological examination).17 18 Muller et al compared findings on high resolution computed tomography with pathological determinants of disease activity in 12 patients with cryptogenic fibrosing alveolitis and found that high resolution computed tomography correctly identified all five patients with appreciable disease activity and five out of seven patients with mild disease activity.17

    In patients with fibrosing alveolitis changes in appearance have also been correlated with improvement in pulmonary function after steroid treatment.18 High resolution computed tomography provides a more accurate assessment of disease activity and the probable prognosis in patients with fibrosing alveolitis and frequently obviates more invasive procedures such as open lung biopsy, thereby reducing the risks to the patient and the costs of investigation.

    Parenchymal disease

    The new technique is also proving increasingly useful in acute lung disease. It may show parenchymal disease in the presence of normal or questionable radiographic findings, particularly in immunocompromised patients. For example, about 10% of patients with pneumocystis pneumonia have a normal chest radiograph at presentation,19 and in this condition high resolution computed tomography can show bilateral perihilar ground glass opacity. In patients with AIDS this is sufficiently characteristic to allow a presumptive diagnosis without further invasive procedures.20

    In patients with AIDS pulmonary Kaposi's sarcoma produces a characteristic distribution of nodules on high resolution computed tomography.21 Invasive aspergillosis may be recognised by the presence of nodules with a halo of ground glass attenuation.22

    Abnormal high resolution computed tomographic scans in patients with normal chest radiographs have also been described in drug induced lung disease and radiation pneumonitis.12 In acute lung diseases several patterns of abnormality are seen that are not sufficiently specific to permit a diagnosis without further investigation. Rapid advances are coming from studies of correlations between tomographic and pathological appearances.23

    As access to computed tomographic scanners becomes more widely available it is imperative that high resolution computed tomography of the lungs is used to its full potential. It may enable earlier diagnosis and more accurate follow up and can guide diagnostic procedures such as bronchoscopy and open lung biopsy to the most auspicious sites. It may also obviate invasive procedures. In some situations high resolution computed tomography of the chest will probably replace conventional imaging techniques.

    References

    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    13. 13.
    14. 14.
    15. 15.
    16. 16.
    17. 17.
    18. 18.
    19. 19.
    20. 20.
    21. 21.
    22. 22.
    23. 23.