Background & objectives: Preoperative localization of parathyroid lesions is essential for improving the results in patients with primary hyperparathyroidism. The purpose of this study was to evaluate retrospectively the value of technetium-99m (Tc-99m) methoxyisobutylisonitrile (MIBI) single photon emission computed tomography (SPECT) and to compare the diagnostic accuracy of iterative reconstruction (IR) and filtered backprojection (FBP) reconstruction algorithms about localization of parathyroid lesions.
Methods: Forty four patients with primary hyperparathyroidism, in whom histopathological correlation could be performed, were included in the study. Dual-phase Tc-99m parathyroid scintigraphy was performed 20 and 120 min after injection of 740 MBq Tc-99m MIBI in all patients. Tomographic images were acquired 120 min after the administration of radiopharmaceutical. The SPECT data were evaluated using an IR as well as a FBP algorithm. In 23 of 44 patients, SPECT acquisitions were performed in 64×64 matrix; in the remaining 21 patients, tomographic data were collected in 128×128 matrix. The imaging results were compared with pathological findings and sensitivities of both reconstruction algorithms, and planar views were calculated.
Results: Using planar MIBI scans, abnormal parathyroid glands were correctly localized in 75 per cent of the cases. Sensitivity increased to 77 per cent using SPECT with FBP and to 84 per cent with IR. When the sensitivities were calculated according to the acquisition matrix, these were 95 per cent (20/21) and 85 per cent (18/21) for IR and FBP, respectively in patients in whom 128×128 matrix was used. The sensitivities were lower in patients who were imaged with 64×64 matrix; these were calculated as 74 per cent (17/23) and 70 per cent (16/23) with IR and FBP, respectively.
Interpretation & conclusions: Our findings showed that compared to planar scintigraphy, Tc-99m MIBI SPECT was more sensitive diagnostic modality in the detection of abnormal parathyroid tissues. Image quality and sensitivity may be improved further when larger matrices with IR are used instead of FBP algorithm.