Clinical Value of Ultra-Low-Dose High-Resolution CT Combined with Deep Learning Image Reconstruction in CT-Guided Percutaneous Lung Biopsy

Objective To evaluate the effects of ultra-low-dose computed tomography (CT) combined with a 1024 × 1024 reconstruction matrix and a deep learning image reconstruction algorithm on image quality and radiation dose in CT-guided percutaneous transthoracic needle biopsy (PTNB).Methods A total of 50 patients who underwent CT-guided thoracic needle biopsy were prospectively enrolled. Multiple repeated local CT scans were performed during the biopsy procedure. Three scanning protocols were applied sequentially: a conventional-dose protocol (Group A: 120 kVp, 180 mA, 512 matrix, ClearView 60%), a low-dose protocol (Group B: 120 kVp with automatic tube current modulation using O-Dose, noise index 1.0, 1024 matrix, ClearInfinity reconstruction at 40%, 60%, and 80%), and an ultra-low-dose protocol (Group C: O-Dose noise index 0.7, 1024 matrix, ClearInfinity reconstruction at 40%, 60%, and 80%). Other scan parameters were kept identical across groups. CT attenuation values and standard deviations (SD) of the ascending aorta and chest wall fat were measured to calculate signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). Subjective image quality, including lesion margin delineation, density homogeneity, and artifact severity, was independently assessed using a five-point Likert scale. CT dose index volume (CTDIvol), dose–length product (DLP), effective dose (ED), and size-specific dose estimate (SSDE) were recorded and analyzed. Intergroup comparisons were performed using one-way ANOVA or non-parametric tests as appropriate.Results Inter-observer agreement for subjective image assessment was good (Kappa=0.705~0.865，P<0.05). Compared with the conventional-dose protocol, CTDIvol and SSDE were reduced by 61.8% and 80.0% in Groups B and C, respectively. In both low- and ultra-low-dose protocols, images reconstructed with ClearInfinity at 60% and 80% demonstrated significantly higher SNR and CNR than those of the conventional protocol (P < 0.05). Image quality reconstructed with ClearInfinity at 60% was comparable to that of the conventional-dose group in terms of lesion margin visibility, density uniformity, and artifact control (P > 0.05) and was superior to other reconstruction strengths within the same dose group.Conclusion With an approximately 80% reduction in radiation dose, high-resolution reconstruction using a 1024 matrix combined with ClearInfinity deep learning reconstruction at 60% strength provides image quality sufficient for CT-guided lung biopsy. This technique represents a safe and feasible imaging solution for ultra-low-dose CT-guided percutaneous lung biopsy.