Radiography of lungs was one of the first applications of X-rays in medicine. With conventional radiography it is not possible to see the airways whereas with computed tomography (CT), the images may reveal morphological abnormalities. This has made CT the principal method for preoperative staging of lung cancer. For imaging the airways in studies of lung function, shorter acquisition times as well as better spatial resolution and contrast are needed.

K-edge subtraction (KES) radiography and CT by synchrotron radiation, using stable Xe gas as the contrast agent, have addressed most of the problems of lung imaging by X-rays. The method is an extension of the ID17 coronary angiography CT programs [1], where KES is used with iodine or gadolinium as contrast agents. K-edge subtraction allows quantitative measurements of indicator (Xe) concentrations, whereas traditional CT or MRI do not. Imaging of the airways of the lungs by KES using Xe gas has been demonstrated earlier [2]. The present work aims at functional imaging of rabbit lungs in vivo as the first stage in the program development.

The goal is to see the morphological details of the bronchial tree and filling of alveoli during the breathing cycle with Xe in the airways. The distribution of Xe is observed in a series of KES images taken using two X-ray energies bracketing the Xe K-edge at 34.56 keV. The animal moves vertically through the 0.7 mm high fan beams, and an 80 x 120 mm image is acquired line by line with 1.3 ms sampling time of the dual-array Ge detector. The imaging sequence requires computer control of the breathing cycle and administration of the gas mixture. At the same time, physiological parameters such as heart rate, CO2 level in the lungs, and tracheal pressure are monitored.

Anaesthetised rabbits were imaged in both the radiography and CT modes. The animals were imaged in the vertical position, fixed on a rotation stage on the angiography/CT chair. The inspiration phase could either be a mixture of air and oxygen or oxygen and xenon. Images were obtained during various sequences of inspiration, rest and expiration. In these initial experiments, the imaging was not gated by the breathing or cardiac cycle.

When the alveoli were filling, the view of the bronchial tree became obstructed in the radiographs. This will generally be true for radiographs. However, a high-resolution CT image of the 2-dimensional cross section of the animal can be obtained by tomographic reconstruction. In the present case the animal is rotated about an axis perpendicular to the plane defined by the beam. Typically, the full rotation takes 1 to 2 seconds, and about 1000 projections are acquired during the rotation with 1 ms sampling time. The distribution of Xe is reconstructed from the difference of the sinograms corresponding to the two energies.

Figure 21 demonstrates that, with correct timing, the entire bronchial tree may be imaged by radiographic projection. The first bifurcation of the bronchial tree, as well as smaller bronchioles, is clearly seen along with the initial filling of alveoli. When the alveoli are filled, the details of lung structure are seen only in CT images. An example is shown in Figure 22. It corresponds to a transverse section in the upper part of the rabbit lungs. Absolute concentration of Xe can be calculated from the images, which is important for lung function studies.

The work was done in close collaboration between the ESRF staff, and teams working at the Medical Faculty of the Grenoble University and at the Central Hospital and the Department of Physics of the University of Helsinki.

[1] H. Elleaume, A.M. Charvet, P. Berkvens, G. Berruyer, T. Brochard, Y. Dabin, M. C. Dominguez, A. Draperi, S. Fiedler, G. Goujon, G. Le Duc, M. Mattenet, C. Nemoz, M. Perez, M. Renier, C. Schulze, P. Spanne, P. Suortti, W. Thomlinson, F. Esteve, B. Bertrand, J.F. Le Bas, Nucl. Instr. and Meth., A 428, 513 (1999).
[2] J. Giacomini, H. Gordon, R. O'Neil, A. VanKessel, B. Cason, D. Chapman, W. Lavender, N. Gmür, R. Menk, W. Thomlinson, Z. Zhong, E. Rubenstein, Nucl. Instr. Meth., A 406, 473 (1998).

G. Le Duc (a), S. Bayat (b), F. Grimbert (b), S. Fiedler (a), C. Nemoz (a), W. Thomlinson (a), L. Porra (c), P. Suortti (c), C-G. Standertskjöld-Nordenstam (c), A.R.A. Sovijärvi (c).

(a) ESRF
(b) University of Grenoble (France)
(c) University of Helsinki (Finland)