B. Janerotsjoberg et al., QUANTITATIVE DIGITAL EVALUATION OF MYOCARDIAL EXERCISE TL-201 SINGLE-PHOTON EMISSION TOMOGRAPHY IN POSTMENOPAUSAL WOMEN, Clinical physiology, 18(3), 1998, pp. 169-177
Quantitative computerized analysis of data from myocardial thallium-20
1 (Tl-201) single-photon emission tomography (SPET) may improve the di
agnostic accuracy of coronary heart disease. The reference ranges for
post-menopausal women are, however, limited and obtained mainly from p
atients. To compare reference values from healthy postmenopausal women
and to improve the quantitative analysis, 20 women (10 patients with
coronary heart disease and previous infarction and 10 age-matched heal
thy volunteers) were examined immediately post exercise and after a de
lay. A nine-segment 'bull's-eye' model was used for analysis. At visua
l evaluation, reproducibility was high (93%), no false-positive result
s were obtained and in 70% of the patients the SPET was interpreted as
abnormal. Using reported reference values for quantitative analysis,
all the healthy women had an abnormal result. New reference values bas
ed on three different methods of 'normalization' were calculated: the
relative activity of segment 3 set to 100%, the segment with the highe
st activity set to 100% and a least-squares method. They all differed
significantly from those that had previously been reported. The freque
ncies of agreement between visual and quantitative analysis were 84-92
% and were highest when segment 3 was used as a reference, but in thi
s case only 40% of the patients with coronary heart disease had an abn
ormal SPET. Using the least-squares method for handling digital inform
ation, the SD of the normal values decreased and 90% of the patients w
ith coronary heart disease were accurately diagnosed. These results pr
ovide quantitative digital reference values for healthy postmenopausal
women. They verify that quantitative analysis is in diagnostic agreem
ent with visual evaluation, stress the need for local verification of
reference ranges and suggest a least-square normalization method for t
he analysis.