Background and Study Aims: The aim of the present study was to establi
sh a suitable method for virtual computed tomography (CT) gastroscopy.
Patients and Methods: Three-millimeter helical CT scans of a pig stom
ach were obtained after air insufflation and instillation of diluted d
iatrizoic acid (Gastrografin), and with double contrast, Bn addition,
three patients with gastric tumors were studied after ingestion of an
effervescent agent (Duplotrast, 6 g) and intravenous injection of hyos
cine butylbromide (Buscopan, 1 ml). Virtual endoscopy images were comp
uted on a Sun Sparc 20 workstation (128 megabytes of random access mem
ory, four gigabytes of hard disk space), using dedicated software (Nav
igator, General Electric Medical System Company), The endoscopy sequen
ces were compared with real endoscopic examinations and with anatomica
l specimens. Results: In the cadaver studies, the best results were ob
tained with plain air insufflation, whereas virtual CT gastroscopy wit
h diluted contrast and with double contrast showed artifacts simulatin
g polyps, erosions, and flat ulcers. Patient studies showed good corre
lation with the fiberoptic endoscopy findings, although large amounts
of retained gastric fluid substantially reduced the quality of the sur
face reconstruction. Conclusion: These preliminary results show that v
irtual CT gastroscopy is table to provide insights into the upper gast
rointestinal tract similar to those of fiberoptic endoscopy. However,
due tee the limited spatial resolution of the CT protocol used, as wel
l as inherent image artifacts associated with the Navigator program's
reconstruction algorithm, the form of virtual CT gastroscopy studied w
as East capable of competing with the imaging quality provided by fibe
roptic gastroscopy.