Abstract: Objective: Ultralow tube voltage scanning was combined with a deep-learning-based image reconstruction algorithm (ClearInfinity, CI). The impacts of this method on the image quality and organ radiation dose in low-dose chest computed tomography (CT) were investigated. Methods: A Lungman PH-1 chest phantom was scanned using two protocols: 120 and 70 kV standard- and low-dose protocols, respectively. Images were reconstructed using three algorithms, filtered back-projection (FBP); adaptive iterative reconstruction (ClearView, CV) at 20%, 40%, 60%, and 80% strengths; and CI at 20%, 40%, 60%, and 80% strengths, totaling nine reconstruction schemes. Objective image quality metrics for the soft tissue thoracic regions were evaluated, including the signal-to-noise ratio (SNR) for 100 HU solid nodules, contrast-to-noise ratio (CNR) for nodules with varying densities (100 HU, −630 HU, −800 HU), figure of merit (FOM), and the SD of the CT values. The nodule detection rates were analyzed using an AI-assisted diagnosis system. Four radiomic texture parameters were extracted for −630 HU ground-glass nodules: sum of squares (SumSquares), difference, contrast, and correlation. Image quality was subjectively assessed using a five-point Likert scale, focusing on noise and clarity in high-artifact regions near the cervicothoracic junction. The doses to the breast, thyroid, and thymus were estimated using Monte Carlo simulations. Image quality metrics were compared using one- or multiway ANOVA or the Scheirer-ray-hare test. Results: CI reconstruction at 60%–80% significantly reduced image noise as well as increased the nodule SNR, CNR, and FOM compared with the baseline method. The performance of CI 80% ranked first overall. The AI model detected 100% of the lung nodules at CI reconstruction strengths of 60% or higher with the 70 kV low-dose protocol. The texture of the nodules was most accurately reconstructed with 80% CI under the 70 kV low-dose protocol, with 18.04%, 1.59%, and 13.54% mean deviations in the SumSquares, difference entropy, and contrast, respectively, from those of the 120 kV FBP reference. Correlation restoration was highest at CI 40%, with a mean deviation of −16.36%. The image quality did not subjectively significantly differ between the 70 kV CI and 120 kV FBP images when the CI strength exceeded 40%, and these images were deemed diagnostically acceptable. The doses to the thyroid, breast, and thymus in the 70 kV low-dose protocol were 86.32%, 81.79%, and 81.90% lower, respectively, than in the 120 kV standard-dose protocol. Conclusion: CI reconstruction, particularly at 60%–80% strength, with ultralow tube voltage chest CT considerably enhances image quality, reduces noise, increases the accuracy AI-based nodule detection, and preserves radiomic texture features compared with the baseline method. The 70 kV low-dose protocol combined with CI reconstruction ≥60% balances image quality and radiation dose, demonstrating the potential for clinical application with low-dose chest CT.