Abstract
BACKGROUND: During tuberculosis treatment, multiple chest computed tomography (CT) scans are required for follow-up, each with an effective radiation dose (ED) of approximately 7 mSv. If treatment outcomes are poor or complications arise, additional scans may be necessary, thereby increasing the cumulative radiation burden. This study aimed to evaluate the effects on image quality of low-dose CT using spectral purification technology, in combination with varying levels of sinogram-affirmed iterative reconstruction (SAFIRE) algorithms, with the goal of identifying the optimal SAFIRE level for imaging in follow-up of active secondary pulmonary tuberculosis in adults. METHODS: A total of 90 patients diagnosed with active secondary pulmonary tuberculosis between January and May 2024 undergoing follow-up chest CT scans after treatment were randomly assigned to three groups: conventional dose scanning (Group A), optimal tube voltage scanning with intelligence (Group B), and low-dose scanning (Group C). In Groups A and B, images were reconstructed using SAFIRE level 3 (A3 and B3), while Group C images were reconstructed with SAFIRE levels 1 through 5 (C1 to C5). The overall image quality was evaluated, double-blind, independently by two senior radiologists. Radiation doses, subjective assessments of image quality, and objective image quality metrics were analyzed and compared across the groups. RESULTS: The ED was significantly lower in Group C compared to Groups A and B (0.28±0.05 vs. 2.79±0.65 and 1.75±0.47 mSv respectively, P<0.001). Within Group C, image noise decreased and both the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) increased with higher SAFIRE levels. SNR was significantly higher in the C5 group compared to the C1 to C4 groups (P<0.05). Subjective assessments indicated significant differences in the visibility of lung fissures, peripheral bronchi, adjacent pulmonary vessels, pleura, and subpleural regions between different SAFIRE levels (P<0.05). The best subjective scores for visualization of parenchymal lesions, including patches and nodules, were at SAFIRE level 2, while images reconstructed at level 5 scored significantly lower (P<0.05). Consistency in subjective evaluations between the C1 to C5 groups was high (Kappa =0.682). CONCLUSIONS: The combination of spectral purification technology with SAFIRE reconstruction effectively reduced radiation exposure and image noise, and maintained diagnostic adequacy during follow-up for patients with active secondary pulmonary tuberculosis.