Artificial intelligence model-based iterative reconstruction for lung ultra-low-dose CT: image quality, ground-glass nodules detectability, and Lung-RADS evaluation

Artificial intelligence model-based iterative reconstruction for lung ultra-low-dose CT: image quality, ground-glass nodules detectability, and Lung-RADS evaluation | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Artificial intelligence model-based iterative reconstruction for lung ultra-low-dose CT: image quality, ground-glass nodules detectability, and Lung-RADS evaluation Ying Chen, Jing Deng, Zhuo Zhu, Xi Cai, Jiashuai Li, Wei Xu, Chenglin He, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7206057/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 6 You are reading this latest preprint version Abstract Objective: To assess the effect of artificial intelligence model-based iterative reconstruction (AIIR) on image quality of lung ultra-low-dose CT (ULDCT), as well as its influence on the detection and diagnostic classification of GGNs. Methods: Fifty-three patients diagnosed with GGNs underwent both lung standard-dose CT (SDCT) and ultra-low-dose CT (ULDCT) scans. SDCT hybrid iterative reconstruction (SDCT-HIR) images, ULDCT hybrid iterative reconstruction (ULDCT-HIR) images, and ULDCT-AIIR images were generated using a lung sharpness algorithm. Image noise measurements were performed. Image quality was independently scored by Radiologists 1 and 2. Separately, GGNsdetectability and Lung-RADS classification were independently evaluated by Radiologists 3 and 4. Results: ​Two patients were excluded due to respiratory motion artifacts and one due to raw data errors, resulting in 50 patients with 81 GGNs for final analysis. Compared to SDCT, ULDCT achieved a remarkable 93.7% reduction in radiation dose​(SDCT: 6.27 ± 0.98 mSv vs. ULDCT: 0.40 ± 0.13 mSv, P < 0.001). Moreover, ULDCT-AIIR images exhibited the lowest noise levels (P < 0.001). The image quality scores of ULDCT-AIIR were significantly superior to those of ULDCT-HIR (P 0.05). For GGNs detection, Radiologist 3 reported rates of 64.2% (ULDCT-HIR) vs. 95.1% (ULDCT-AIIR), while Radiologist 4 reported 67.9% (ULDCT-HIR) vs. 96.3% (ULDCT-AIIR). ULDCT-AIIR images demonstrated significantly higher detection rates than ULDCT-HIR images (P < 0.001). The consistency in the Lung-RADS classification was moderate between ULDCT-HIR and SDCT-HIR images (κ=0.343 and 0.411 for radiologist 3 and 4, respectively), but good between ULDCT-AIIR and SDCT-HIR images (κ=0.772 and 0.743 for radiologist 3 and 4, respectively). Conclusion: AIIR can significantly enhance image quality in ULDCT while maintaining excellent GGNs detection capability. Moreover, there is good consistency in Lung-RADS classification between ULDCT-AIIR and SDCT-HIR images. Artificial Intelligence Model-based Iterative Reconstruction Ultra-low Dose CT Lung Ground-glass Nodule Figures Figure 1 Figure 2 Figure 3 Introduction Lung cancer remains the most prevalent malignancy, and early detection and intervention can significantly enhance the 5-year survival rate of patients with non-small cell lung cancer, raising it to 80% [ 1 ]. The National Lung Screening Trial in the US revealed that lung low-dose CT(LDCT) scans decrease the mortality risk of lung cancer patients by 20% compared to chest X-rays [ 2 ]. Current guidelines suggest a maximum radiation dose of 3.0 mSv for LDCT in non-obese individuals, significantly exceeding the radiation exposure from chest X-rays, which falls between 0.03 to 0.1 mSv [ 3 – 4 ]. Hence, there is a need to further minimize the radiation dose associated with lung CT scans. Pulmonary ground-glass nodule (GGN), encompassed pure GGN (pGGN) and mixed GGN (mGGN), serve as the primary CT manifestation of early lung cancer, making it crucial to classify them based on Lung Imaging-Report and Data System (Lung-RADS). However, excessive reduction in radiation dose can compromise the detection accuracy of GGNs and Lung-RADS classification due to the subtle density differences between GGNs and lung tissue. Iterative reconstruction has proven effective in minimizing image noise for lung LDCT and ultra-low-dose CT (ULDCT) [ 5 – 15 ], in the model-based iterative reconstruction (MBIR) algorithm, the regularization component used can significantly reduce noise in ULDCT images, but it may also lead to image distortion and show notable differences compared to filtered back projection reconstructed images. Consequently, diagnosing GGNs on lung ULDCT images remains a challenge. In recent years, artificial intelligence (AI)-based image reconstruction techniques have further minimized noise in ULDCT images while improving the identification accuracy of pulmonary nodule, particularly solid nodules [ 16 – 25 ]. However, clinical studies focusing on the detection of GGNs using AI-based deep learning reconstruction for ULDCT remain limited. Moreover, in most prior lung ULDCT studies, the employed deep learning reconstruction algorithms were trained solely in the image domain to reduce noise, fundamentally incapable of eliminating streak and cone-beam artifacts. Notably, an AI MBIR (AIIR) algorithm has been introduced for dramatic decreasing images noise [ 23 ]. This innovative algorithm replaces the regularization component of MBIR with an AI-driven convolutional neural network. Through iterative cycles of forward and backward projections, AIIR integrates comprehensive modeling of optical properties, noise characteristics, anatomical structures, and physical statistics. Simultaneously, it performs deep learning-driven CT image reconstruction in both projection and image domains, achieving three significant advancements: (1) a remarkable reduction in CT image noise, (2) substantial elimination of streak and cone-beam artifacts [ 23 – 25 ], and (3) resolution of the inherent issue of slow reconstruction speed associated with MBIR. The aim of our study was to assess the impact of AIIR on image quality of lung ULDCT, as well as its influence on the detection and diagnostic classification of GGNs. Materials and Methods Study population This prospective, observational study was granted approval by the medical ethics committee (No.202200601) of our hospital, and written informed consent was s obtained from all patients at our hospital. The inclusion criteria for the study were as follows: (1) patients undergoing lung CT follow-up for GGN, with a previous CT examination confirming the presence of at least one GGN; (2) patients without any metal implants or internal fixators in the chest; (3) patients who were not pregnant. The exclusion criteria were as follow: (1) presence of respiratory motion artifacts; (2) raw data collapse in ULDCT. CT examination and image reconstruction. Lung CT examinations were performed using a 320-row CT scanner (uCT960+, United Image Healthcare, Shanghai, China). Initially, a standard-dose CT (SDCT) scan was executed, followed by an ULDCT scan. The SDCT scan parameters included a tube voltage of 120kV and automatic tube current adjustment with a reference tube current time product of 120mAs. For the ULDCT scan, the parameters were set to a tube voltage of 100kV and automatic tube current adjustment with a reference tube current time product of 10mAs. Both scans shared consistent parameters, specifically a collimation width of 80mm, a pitch of 1.0, and a gantry rotation speed of 0.5s/rot. A hybrid iterative reconstruction (HIR) and a lung sharpness algorithm was employed to produce SDCT-HIR and ULDCT-HIR images, while a lung sharpness algorithm was employed to produce ULDCT-AIIR images. All images were obtained using an image thickness and spacing of 1mm, a matrix of 512×512, a window width of 1400Hu, and a window level of -500Hu. Radiation dose assessment The volume CT dose index and radiation dose-length product was recorded. The effective dose was calculated by multiplying the dose-length product with a chest conversion coefficient (k = 0.014 mSv/mGy cm) [ 26 ]. Lung CT image quality assessment The regions of interest were positioned on the axial images within both pectoralis major muscles at the level of the second thoracic vertebra, as well as within the thoracic aorta at the tracheal bifurcation level and the diaphragm's apex level. The regions of interest were maximized while excluding the aortic wall and calcification, ensuring consistency in size and location across SDCT-HIR, ULDCT-HIR, and ULDCT-AIIR images. The noise at the lung apex was calculated based on the mean of the standard deviation of CT values for both psoas majors. Similarly, the noise at the middle and bottom parts of the lung was calculated based on the standard deviation of CT values for the thoracic aorta at the tracheal bifurcation level and the diaphragm's apex level, respectively. The image graininess, striped and cone-beam artifacts, clarity of small blood vessels (< 3mm in diameter), homogeneity of lung tissue density, and overall image quality were scored using a double-blind method and a five-point scale by radiologists 1 and 2, one with 23 years of experience and the other with 10 years of experience in chest CT (Table 1 ). All GGNs larger than 3 mm in diameter on SDCT-HIR images were identified through manual interpretation by these two radiologists combined with a deep learning-based computer-aided diagnosis system (United Intelligent Healthcare, uAI-ChestCare). Subsequently, both radiologists thoroughly reviewed the images and confirmed the authenticity of all identified GGNs. The diameter of each GGN was determined by averaging its long and short diameters. After a 2-month interval, the clarity of GGNs was scored by radiologist 1 and radiologist 2, who were only informed of the locations of GGNs but uninformed of patient information and image types, independently evaluated the GGNs with a diameter > 3mm on ULDCT-HIR and ULDCT-AIIR images in a random order. The scoring criteria were elaborated in Table 1 . Table 1 Image Quality Scoring Criteria evaluating indicator Scoring criteria Image graininess 1 = severe graininess; 2 = significant graininess; 3 = moderate graininess; 4 = slight graininess; 5 = no graininess Striped and cone-beam artifacts 1 = severe striped and cone-beam artifacts; 2 = significant striped and cone-beam artifacts; 3 = moderate striped and cone-beam artifacts; 4 = slight striped and cone-beam artifacts; 5 = no striped or cone-beam artifacts Clarity of small blood vessels (< 3mm in diameter) 1 = very poor clarity; 2 = poor clarity; 3 = average clarity; 4 = good clarity; 5 = excellent clarity Homogeneity of lung tissue density 1 = severe unevenness; 2 = significant unevenness; 3 = moderate unevenness; 4 = slight unevenness; 5 = uniform lung tissue density Overall image quality 1 = very poor; 2 = poor; 3 = average; 4 = good; 5 = excellent Clarity of GGNs 1 = completely invisible to the naked eye; 2 = very blurry yet observable; ; 3 = slightly blurry; 4 = clear; 5 = very clear Lung GGN detection Two additional radiologists (radiologist 3 and radiologist 4, with 20 and 8 years of chest CT experience, respectively) independently identified GGNs > 3 mm in diameter through manual interpretation on both ULDCT-HIR and ULDCT-AIIR images. The two readers were unaware of both image types and GGN locations. Using SDCT-HIR images as the reference standard, we assessed the detection of GGNs > 3 mm on ULDCT-HIR and ULDCT-AIIR images. Lung-RADS classification for GGNs Two months after completing the GGNs detection assessment, radiologists 3 and radiologist 4 only informed of the locations of GGNs, but uninformed of patient information and image types, independently classified the GGNs with a diameter > 3mm on SDCT-HIR, ULDCT-HIR and ULDCT-AIIR images according to Lung-RADS2022 in a random order [ 27 ]. The consistency in Lung-RADS classification between the two sets of ULDCT images and SDCT-HIR images was assessed. Statistical analysis Statistical analysis was performed using SPSS 22.0 software (version 22.0; IBM SPSS Statistics). The Friedman test and subsequent pairwise tests were used for multiple group comparisons of quantitative data. The Wilcoxon test was used for comparison of two paired quantitative data. For comparisons of count data, the McNamar test was employed. Consistency evaluation was conducted using the Kappa test (k ≥ 0.80 indicates excellent consistency, 0.60 < k < 0.80 indicates good consistency, 0.40 < k ≤ 0.60 indicates moderate consistency, 0.20 < k ≤ 0.40 indicates fair consistency, and k ≤ 0.20 indicates poor consistency). A P-value < 0.05 was considered statistically significant. Results Demographic characteristics, CT radiation dose, and GGNs on SDCT-HIR images From July 1, 2022 to August 1, 2023, a total of 53 patients were recruited at our hospital. Among the 53 patients initially considered, 2 patients were excluded because of respiratory motion artifacts, while another patient was excluded due to the collapse of raw data related to ULDCT. Consequently, a total of 50 patients were included in final. Table 2 provides an overview of their demographic characteristics as well as the CT radiation dose. When compared to SDCT, ULDCT demonstrated a remarkable 93.7% reduction in radiation dose (P<0.001). A total of 81 GGNs were included in this study, comprising 68 pGGNs and 13 mGGNs, as identified on SDCT-HIR images. The average diameter was 7.6 ± 3.4mm, ranging from 4 to 22mm. The average CT value was − 593.5 ± 128.9 Hu, with a range from − 845.0 to -289.6 Hu. Compared to previous CT images, 12 GGNs showed increase in size or solid component volume, while none showed any reduction in size or solid component volume. Table 2 Demographic Characteristics of Patients and CT Radiation Dose Parameters (n = 50) Characteristic Number (%), mean ± SD (range) Age (years) 51.9 ± 12.8 (26–73) Male gender (%) 20 (40.0%) Weight (kg) 64.3 ± 10.4 (41.0–90.0) Body mass index (kg/m 2 ) 23.2 ± 2.4 (17.0-29.4) SDCT radiation dose parameters Radiation dose-length product (mGy·cm) 448.0 ± 70.3 (300.0-607.3) Volume CT dose index (mGy) 11.2 ± 1.4 (8.4–15.7 ) Effective radiation dose (mSv) 6.27 ± 0.98 (4.19–8.50) ULDCT radiation dose parameters Volume CT dose index (mGy·cm) 0.7 ± 0.2 (0.5–1.3) Radiation dose and length product (mGy·cm) 28.7 ± 9.4 (18.1–53.4) Effective radiation dose (mSv) 0.40 ± 0.13 (0.25–0.75 ) Lung CT Image Quality Upon conducting an overall comparison of the noise among SDCT-HIR, ULDCT-HIR, and ULDCT-AIIR images, significant differences were observed. Furthermore, pairwise comparisons between any two sets of CT images also revealed notable disparities (P < 0.001). Specifically, ULDCT-AIIR images exhibited the lowest noise (Table 3 ). The interobserver agreement for image quality assessment demonstrated good consistency, with κ values as follows: 0.770 (95% CI: 0.684–0.856) for graininess, 0.754 (95% CI: 0.662–0.856) for striped and cone-beam artifacts, 0.742 (95% CI: 0.652–0.832) for small vessel clarity, 0.746 (95% CI: 0.658–0.834) for lung tissue density homogeneity, and 0.715 (95% CI: 0.623–0.807) for overall image quality. Notably, the scores of these indicators for ULDCT-HIR images were significantly lower than those for SDCT-HIR images. However, the scores for ULDCT-AIIR images were significantly improved compared to those for ULDCT-HIR images and were on par with those for SDCT-HIR images (Table 3 and Fig. 1 – 3 ). Table 3 Image Quality Assessment (50 patients with 81 GGNs) Evaluating indicator SDCT-HIR ULDCT-HIR ULDCT-AIIR P value Pulmonary apex noise (Hu) 64.7 ± 11.8 120.4 ± 30.0 51.7 ± 7.1 Δ < 0.001 Noise at the middle of the lung (Hu) 70.0 ± 12.1 107.4 ± 18.5 52.8 ± 4.4 Δ < 0.001 Noise at the bottom of the lung 64.8 ± 11.7 121.3 ± 22.6 48.3 ± 6.0 Δ < 0.001 Score of image quality Radiologist 1 Image graininess 4.98 ± 0.14 2.64 ± 0.53 Δ 4.76 ± 0.52 ★☆ < 0.001 Striped and cone-beam artifacts 4.96 ± 0.20 2.34 ± 0.48 Δ 4.86 ± 0.35 ★☆ < 0.001 Small vessel clarity 4.96 ± 0.20 2.82 ± 0.60 Δ 4.64 ± 0.56 ★☆ < 0.001 Lung tissue density homogeneity 4.88 ± 0.33 2.94 ± 0.59 Δ 4.58 ± 0.61 ★☆ < 0.001 Overall image quality 4.96 ± 0.20 2.68 ± 0.55 Δ 4.72 ± 0.54 ★☆ < 0.001 Clarity of GGNs 3.70 ± 0.98 2.64 ± 0.81 Δ 3.65 ± 0.91 ★☆ < 0.001 Radiologist 2 Image graininess 4.98 ± 0.14 2.64 ± 0.53 Δ 4.80 ± 0.45 ★☆ < 0.001 Striped and cone-beam artifacts 4.90 ± 0.30 2.40 ± 0.49 Δ 4.92 ± 0.27 ★☆ < 0.001 Small vessel clarity 4.90 ± 0.30 2.88 ± 0.56 Δ 4.74 ± 0.49 ★☆ < 0.001 Lung tissue density homogeneity 4.84 ± 0.37 2.90 ± 0.54 Δ 4.58 ± 0.54 ★☆ < 0.001 Overall image quality 4.90 ± 0.30 2.80 ± 0.45 Δ 4.74 ± 0.49 ★☆ < 0.001 Clarity of GGNs 3.70 ± 0.95 2.58 ± 0.89 Δ 3.70 ± 0.95 ★☆ < 0.001 Note: Δ significantly lower than SDCT-HIR image; ★ no statistical difference compared to SDCT-HIR images; ☆ significantly higher than ULDCT-HIR images; SDCT-HIR: standard-dose CT hybrid iterative reconstruction; ULDCT-HIR: ultra-low-dose CT hybrid iterative reconstruction; ULDCT-AIIR: ultra-low-dose CT artificial intelligence model-based iterative reconstruction The interobserver consistency of GGN clarity score was good ( κ = 0.666, 95% CI:0.593–0.739). The score for GGN clarity demonstrated comparable results on ULDCT-AIIR and SDCT-HIR images, while the score for GGN clarity on ULDCT-AIIR images was better than that on ULDCT-HIR images. GGNs detection on ULDCT images The detection results for the GGNs with a diameter > 3mm are presented in Table 4 . It was observed that the detection rate was higher on ULDCT-AIIR images compared to ULDCT-HIR images (all P-values = 0.000). Table 4 Test Results of Detection of GGNs (n = 81) GGNs Radiologist 3 Radiologist 4 ULDCT-HIR ULDCT-AIIR ULDCT-HIR ULDCT-AIIR Detected (%) 52 (64.2%) 77 (95.1%) ☆ 55 (67.9%) 78 (96.3%) ☆ Undetected (%) 29 (35.8%) 4 (4.9%) Δ 26 (32.1%) 3 (3.7%) Δ pGGN (%) 27 (33.3%) 4 (4.9%) Δ 26 (32.1%) 3 (3.7%) Δ mGGN (%) 2 (2.5%) 0 (0.0%) 0 (0.0%) 0 (0.0%) Diameter (mm) 4.0-14.5 4.0-4.5 4.0–7.0 4.0–7.0 False positive 1 2 2 1 Note: ULDCT-HIR: Ultra-low-dose CT hybrid iterative reconstruction; ULDCT-AIIR: Ultra-low-dose CT artificial intelligence model-based iterative reconstruction; pGGN: pure ground glass nodule; mGGN: mixed ground glass nodule; ☆ significantly higher than ULDCT-HIR images; Δ significantly lower than ULDCT-HIR images Lung-RADS classification The Lung-RADS classification results are presented in Table 5 . The interobserver consistency proved to be good ( κ = 0.707, 95% CI:0.609–0.805). Inter-modality agreement analysis revealed moderate consistency between ULDCT-HIR and SDCT-HIR images, with κ values of 0.343 (95% CI: 0.153–0.533) for Radiologist 3 and 0.411 (95% CI: 0.227–0.595) for Radiologist 4. In contrast, ULDCT-AIIR demonstrated significantly better agreement with SDCT-HIR, showing good consistency for both radiologists (Radiologist 3: κ = 0.772, 95% CI: 0.600-0.944; Radioloist: κ = 0.743, 95% CI: 0.563–0.923). Table 5 Lung-RADS Category of GGNs (n = 81) Quantity Lung-RADS Radiologist 3 Radiologist 4 SDCT-HIR ULDCT-HIR ULDCT-AIIR SDCT-HIR ULDCT-HIR ULDCT-AIIR 0 0 7 0 1 6 1 1 0 10 4 3 14 2 2 68 54 64 65 53 66 3 6 4 6 5 2 5 4A 0 0 0 0 0 0 4B 1 1 1 1 1 1 4X 6 5 6 6 5 6 Note: SDCT-HIR: standard-dose CT hybrid iterative reconstruction; ULDCT-HIR: ultra-low-dose CT hybrid iterative reconstruction; ULDCT-AIIR: ultra-low-dose CT artificial intelligence model-based iterative reconstruction Discussion Recent studies have demonstrated that AI-based denoising techniques significantly improve image quality and enhance the detection rate of pulmonary nodules on ULDCT [ 17 – 19 , 21 , 28 – 29 ]. However, most existing research has focused on solid pulmonary nodules, with limited investigation into their impact on GGNs, particularly pGGNs, in ULDCT imaging. Furthermore, while the majority of prior studies employed AI denoising methods trained exclusively in the image domain, research on dual-domain deep learning approaches—combining both image and projection domain training (e.g., AIIR technology)—remains scarce in the context of lung ULDCT. Yang et al. [ 29 ] reported that AIIR significantly enhanced the detection rate for pulmonary nodules on lung ULDCT, but they did not assess the impact of AIIR on image quality and Lung-RADS classification. In our study, we found that AIIR significantly improved the image quality of ULDCT, achieving a remarkable 93.7% reduction in radiation dose compare to SDCT. Notably, ULDCT-AIIR images demonstrated exceptional detection rates for GGNs while maintaining excellent concordance with SDCT-HIR images in Lung-RADS classification. This advancement enables clear visualization and accurate diagnostic classification of GGNs while achieving a dramatic reduction in radiation dose from pulmonary CT scans. It holds particular significance for patients requiring long-term GGN surveillance, as it substantially mitigates the risks associated with cumulative ionizing radiation exposure—especially critical given recent studies highlighting potential carcinogenic effects from CT scans[ 30 , 31 ]. During the quantitative assessment of image noise, we discovered that ULDCT-HIR images exhibited notable noise levels. However, the noise of ULDCT-AIIR images was dramatically reduced, even lower than the low noise of SDCT-HIR images. Furthermore, the image graininess score of ULDCT-AIIR images underwent a significant enhancement when compared to ULDCT-HIR images. Additionally, AIIR technology excelled in minimizing striped artifacts. These impressive results are attributed to the synergistic benefits of AI and MBIR inherent in AIIR, which endow it with exceptional image denoising proficiency, outperforming HIR [ 23 ]. This study revealed that, despite no significant disparity in the score of lung tissue density homogeneity between ULDCT-AIIR and SDCT-HIR images, the former exhibited slightly lower homogeneity. This observation can be attributed to the utilization of a sharpening algorithm, rather than a standard one, during the reconstruction of ULDCT-AIIR images to achieve clearer lung visualizations. Prior research has indicated that model-based iterative reconstruction, when employed in lung sharpening algorithms, tends to result in the emergence of uneven normal lung tissue [ 32 ]. Given that AIIR incorporates model-based iterative reconstruction data information [ 23 ], it follows that lung tissue unevenness may manifest in AIIR images generated using the lung sharpening algorithm. Our study demonstrated that the detection rate of GGNs on ULDCT-AIIR images reached remarkably high values of 95.1% and 96.3%. Only a minimal number of pGGNs were missed, all of which measured ≤ 7 mm in diameter. Nevertheless, these undetected GGNs on ULDCT-AIIR images were clinically insignificant. Although false-positive GGNs were observed on ULDCT-AIIR images, their occurrence was notably infrequent. Retrospective analysis suggested this phenomenon might be attributed to suboptimal visualization of localized image details. The consistency of Lung-RADS classification between ULDCT-HIR and SDCT-HIR images was moderate, primarily due to the high noise of ULDCT-HIR images. This high noise of ULDCT-HIR images obscured some GGNs, leading to their classification as category 1. Additionally, even when a portion of the GGNs was not fully obscured by the high noise of ULDCT-HIR images, the diagnosis was still affected, resulting in some GGNs being classified as category 0. However, it's worth noting that the inconsistent was rare between ULDCT-AIIR and SDCT-HIR images. Our study also had several limitations. First, since most GGNs in this study lacked pathological confirmation, we were unable to determine the definitive diagnostic accuracy for lung cancer on ULDCT-AIIR images. However, we maintain that despite not evaluating the concordance between GGN diagnosis and pathological findings, assessing the consistency of Lung-RADS classifications across different imaging modalities remains significant clinical value. Our findings demonstrate excellent agreement in Lung-RADS categorization of GGNs between ULDCT-AIIR and SDCT-HIR images, suggesting that GGN classification remains largely unaffected when using AIIR at ultra-low radiation doses. Second, when the GGNs was classified according to Lung-RADS2022, radiologists were informed of the locations of GGNs to simplify the analysis process. Thirdly, the enrolled patients exhibited a relatively low number of GGNs in category 3 and 4; however, these GGNs possess higher CT values due to solid components, making them less susceptible to concealment by noise compared to pGGNs. Lastly, our study did not include patients with a BMI exceeding 30 kg/m 2 , necessitating further verification of feasibility in obese individuals. In summary, AIIR significantly can enhance image quality in ULDCT while maintaining excellent GGNs detection capability. Moreover, there is good consistency in Lung-RADS classification between ULDCT-AIIR and SDCT-HIR images. Abbreviations CT computed tomography LDCT low-dose CT GGN pulmonary ground-glass nodule pGGN pure GGN mGGN mixed GGN Lung-RADS Lung Imaging-Report and Data System ULDCT ultra-low-dose CT SDCT standard-dose CT MBIR model-based iterative reconstruction AI artificial intelligence AIIR artificial intelligence model-based iterative reconstruction HIR hybrid iterative reconstruction Declarations Acknowledgements Not applicable. Author contributions YC: Collection of cases, Conceptualization, Writing—Original Draft, Investigation, Data curation. JD: Collection of cases, Investigation, Formal analysis, Data curation ZZ: Image Postprocessing, Investigation, Data curation XC: Collection of cases, Investigation, Data curation JL: Software, Postprocessing, Data curation WX: Conceptualization, Investigation CH: Collection of cases, Investigation QZ: Investigation, Review JC: Investigation, Writing, Review & Editing, Methodology, Resources, Supervision. Funding This study was supported by Hubei Province Key Laboratory of Occupational Hazard Identification and Control, Wuhan University of Science and Technology（No：OHIC2021G04). Data availability The datasets used and analyzed during the current study are available from the corresponding author on reasonable request. Ethical approval and consent to participate This study has been reviewed and approved by the Ethics Committee of Puren Hospital Affiliated with Wuhan University of Science and Technology (Approval No. 202200601), and strictly adheres to the ethical principles outlined in the Declaration of Helsinki. In compliance with the guidelines established by the Council for International Organizations of Medical Sciences (CIOMS), written informed consent was obtained from all participants. To ensure privacy and confidentiality, all participant data were thoroughly anonymized and de-identified. All procedures were conducted in accordance with relevant guidelines and regulatory requirements. Consent for publication Not applicable. Competing interests All authors have no competing interests. References SEER cancer statistics review. 1975–2015, national cancer institute. https://seer.cancer.gov/csr/1975_2015/ National Lung Screening Trial Research Team, Aberle DR, Adams AM, Berg CD, Black WC, Clapp JD, et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med. 2011;365(5):395–409. Quaia E, Baratella E, Cernic S, Lorusso A, Casagrande F, Cioffi V, et al. Analysis of the impact of digital tomosynthesis on the radiological investigation of patients with suspected pulmonary lesions on chest radiography. Eur Radiol. 2012;22(9):1912–22. Bach PB, Mirkin JN, Oliver TK, Azzoli CG, Berry DA, Brawley OW, et al. Benefits and harms of CT screening for lung cancer: A systematic review. JAMA. 2012;307(22):2418–29. Ye K, Zhu Q, Li M, Lu Y, Yuan H. A feasibility study of pulmonary nodule detection by ultralow-dose CT with adaptive statistical iterative reconstruction-V technique. Eur J Radiol. 2019;119:108652. Ju YH, Lee G, Lee JW, Hong SB, Suh YJ, Jeong YJ. Ultra-low-dose lung screening CT with model-based iterative reconstruction: An assessment of image quality and lesion conspicuity. Acta Radiol. 2018;59(5):553–59. Takahashi EA, Koo CW, White DB, Lindell RM, Sykes AG, Levin DL, et al. Prospective pilot evaluation of radiologists and computer-aided pulmonary nodule detection on ultra-low-dose CT with tin filtration. J Thorac Imaging. 2018;33(6):396–401. Zhang M, Qi W, Sun Y, Jiang Y, Liu X, Hong N. Screening for lung cancer using sub-millisievert chest CT with iterative reconstruction algorithm: Image quality and nodule detectability. Br J Radiol. 2018;91(1090):20170658. Ye K, Chen M, Li J, Zhu Q, Lu Y, Yuan H. Ultra-low-dose CT reconstructed with ASiR-V using SmartmA for pulmonary nodule detection and Lung-RADS classifications compared with low-dose CT. Clin Radiol. 2021;76(2):156.e1-156.e8. Fletcher JG, Levin DL, Sykes AG, Lindell RM, White DB, Kuzo RS, et al. Observer performance for detection of pulmonary nodules at chest CT over a large range of radiation dose levels. Radiology. 2020;297(3):699–707. Thapaliya S, Gilligan LA, Brady SL, Anton CG, Crotty EJ, Nasser MP, et al. Comparison of 0.3-mSv CT to standard-dose CT for detection of lung nodules in children and young adults with cancer. AJR Am J Roentgenol. 2021;217(6):1444–51. Han D, Cai J, Heus A, Heuvelmans M, Imkamp K, Dorrius M, et al. Detection and size quantification of pulmonary nodules in ultralow-dose versus regular-dose CT: A comparative study in COPD patients. Br J Radiol. 2023;96(1144):20220709. Kim H, Park CM, Kim SH, Lee SM, Park SJ, Lee KH, et al. Persistent pulmonary subsolid nodules: Model-based iterative reconstruction for nodule classification and measurement variability on low-dose CT. Eur Radiol. 2014;24(11):2700–8. Hu QJ, Liu YW, Chen C, Kang SC, Sun ZY, Wang YJ, et al. Prospective study of low- and standard-dose chest CT for pulmonary nodule detection: A comparison of image quality, size measurements and radiation exposure. Curr Med Sci. 2021;41(5):966–73. Kim SK, Kim C, Lee KY, Cha J, Lim HJ, Kang EY, et al. Accuracy of model-based iterative reconstruction for CT volumetry of part-solid nodules and solid nodules in comparison with filtered back projection and hybrid iterative reconstruction at various dose settings: An anthropomorphic chest phantom study. Korean J Radiol. 2019;20(7):1195–206. Mikayama R, Shirasaka T, Kojima T, Sakai Y, Yabuuchi H, Kondo M, et al. Deep-learning reconstruction for ultra-low-dose lung CT: Volumetric measurement accuracy and reproducibility of artificial ground-glass nodules in a phantom study. Br J Radiol. 2022;95(1130):20210915. Gorenstein L, Onn A, Green M, Mayer A, Segev S, Marom EM. A novel artificial intelligence based denoising method for ultra-low dose CT used for lung cancer screening. Acad Radiol. 2023;30(11):2588–97. Goto M, Nagayama Y, Sakabe D, Emoto T, Kidoh M, Oda S, et al. Lung-optimized deep-learning-based reconstruction for ultralow-dose CT. Acad Radiol. 2023;30(3):431–40. Kim C, Kwack T, Kim W, Cha J, Yang Z, Yong HS. Accuracy of two deep learning-based reconstruction methods compared with an adaptive statistical iterative reconstruction method for solid and ground-glass nodule volumetry on low-dose and ultra-low-dose chest computed tomography: A phantom study. PLoS ONE. 2022;17(6):e0270122. Cao Q, Mao Y, Qin L, Quan G, Yan F, Yang W. Improving image quality and lung nodule detection for low-dose chest CT by using generative adversarial network reconstruction. Br J Radiol. 2022;95(1138):20210125. Singh R, Digumarthy SR, Muse VV, Kambadakone AR, Blake MA, Tabari A, et al. Image quality and lesion detection on deep learning reconstruction and iterative reconstruction of submillisievert chest and abdominal CT. AJR Am J Roentgenol. 2020;214(3):566–73. Koetzier LR, Mastrodicasa D, Szczykutowicz TP, van der Werf NR, Wang AS, Sandfort V, et al. Deep learning image reconstruction for CT: Technical principles and clinical prospects. Radiology. 2023;306(3):e221257. Ji MT, Wang RR, Wang Q, Li HS, Zhao YX. Feasibility study of double-low scanning protocol combined with artificial intelligence iterative reconstruction algorithm for abdominal computed tomography enhancement in patients with obesity. BMC Med Imaging. 2025;25(1):276. Gong H, Peng L, Du X, An J, Peng R, Guo R, et al. Artificial intelligence iterative reconstruction in computed tomography angiography: an evaluation on pulmonary arteries and aorta with routine dose settings. J Comput Assist Tomogr. 2024;48(2):244–50. Li J, Zhu J, Zou Y, Zhang G, Zhu P, Wang N, et al. Diagnostic CT of colorectal cancer with artificial intelligence iterative reconstruction: a clinical evaluation. Eur J Radiol. 2024;171:111301. Deak PD, Smal Y, Kalender WA. Multisection CT protocols: Sex- and age-specific conversion factors used to determine effective dose from dose-length product. Radiology. 2010;257(1):158–66. Christensen J, Prosper AE, Wu CC, Chung J, Lee E, Elicker B, Hunsaker AR, et al. ACR Lung-RADS v2022: Assessment categories and management recommendations. J Am Coll Radiol. 2024;21(3):473–88. Hata A, Yanagawa M, Yoshida Y, Miyata T, Tsubamoto M, Honda O, et al. Combination of deep learning-based denoising and iterative reconstruction for ultra-low-dose CT of the chest: Image quality and Lung-RADS evaluation. AJR Am J Roentgenol. 2020;215(6):1321–8. Yang L, Liu H, Han J, Xu S, Zhang G, Wang Q, et al. Ultra-low-dose CT lung screening with artificial intelligence iterative reconstruction: evaluation via automatic nodule-detection software. Clin Radiol. 2023;78(7):525–31. Bosch de Basea M, Thierry-Chef I, Harbron R, Hauptmann M, Byrnes G, Bernier MO et al. Risk of hematological malignancies from CT radiation exposure in children, adolescents and young adults. Nat Med. 2023;29(12):3111-9.Erratum in: Nat Med. 2025 Apr 11. 10.1038/s41591-025-03689-5 Smith-Bindman R, Chu PW, Azman Firdaus H, Stewart C, Malekhedayat M, Alber S et al. Projected lifetime cancer risks from current computed tomography imaging. JAMA Intern Med. 2025;185(6):710–719.Erratum in: JAMA Intern Med. 2025;185(6):747. Hata A, Yanagawa M, Honda O, Miyata T, Tomiyama N. Ultra-low-dose chest computed tomography for interstitial lung disease using model-based iterative reconstruction with or without the lung setting. Med (Baltim). 2019;98(22):e15936. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 28 Aug, 2025 Reviewers invited by journal 25 Aug, 2025 Editor assigned by journal 19 Aug, 2025 Editor invited by journal 30 Jul, 2025 Submission checks completed at journal 29 Jul, 2025 First submitted to journal 29 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7206057","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":508677720,"identity":"65a11f61-58c3-4f14-b9b8-9f6b3162dcca","order_by":0,"name":"Ying Chen","email":"","orcid":"","institution":"Puren Hospital Affiliated to Wuhan University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Ying","middleName":"","lastName":"Chen","suffix":""},{"id":508677721,"identity":"3e0bb1bf-8d27-4846-bb33-9768cddaf50e","order_by":1,"name":"Jing Deng","email":"","orcid":"","institution":"Puren Hospital Affiliated to Wuhan University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Deng","suffix":""},{"id":508677722,"identity":"a71f2796-3b88-4bd9-b715-a071fde202de","order_by":2,"name":"Zhuo Zhu","email":"","orcid":"","institution":"Puren Hospital Affiliated to Wuhan University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Zhuo","middleName":"","lastName":"Zhu","suffix":""},{"id":508677723,"identity":"888358dc-a163-44ab-a616-0a0ac291bd5f","order_by":3,"name":"Xi Cai","email":"","orcid":"","institution":"Puren Hospital Affiliated to Wuhan University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Xi","middleName":"","lastName":"Cai","suffix":""},{"id":508677724,"identity":"74644f27-875d-493d-b7ca-ab4f5b9de6a3","order_by":4,"name":"Jiashuai Li","email":"","orcid":"","institution":"Puren Hospital Affiliated to Wuhan University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Jiashuai","middleName":"","lastName":"Li","suffix":""},{"id":508677725,"identity":"bcc7bce0-715d-42b2-a87e-ca7efa1cec82","order_by":5,"name":"Wei Xu","email":"","orcid":"","institution":"Puren Hospital Affiliated to Wuhan University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Wei","middleName":"","lastName":"Xu","suffix":""},{"id":508677726,"identity":"a87c087d-8898-4fec-830d-15af23cf8710","order_by":6,"name":"Chenglin He","email":"","orcid":"","institution":"Puren Hospital Affiliated to Wuhan University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Chenglin","middleName":"","lastName":"He","suffix":""},{"id":508677727,"identity":"f217eb2c-81a6-49ec-b92b-175dcb5074ae","order_by":7,"name":"Qing Zhang","email":"","orcid":"","institution":"Puren Hospital Affiliated to Wuhan University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Qing","middleName":"","lastName":"Zhang","suffix":""},{"id":508677728,"identity":"b64cbf5a-74a6-4f12-ac23-cba02968186f","order_by":8,"name":"Jianxin Cao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA60lEQVRIiWNgGAWjYJACgwQgwcbA2PggoUJCTp4ELczNBg/OWBgbNhBvGXub4MO2ikSGAwTU8bcfPlDwoOZOYp90YxtD4jyJBMYG5oePbuDRInEmLcEg4dgzYzaZg20PErdJ5LEzsBkb5+DzB0OOgUEC22E5NonEdgOglmLGBh42abxa+N8Atfw7zAPU0iaROEciseEAIS0SQFsS28C2ALU0EKFF4sazBIPEvsPGQC3NQE9JGBs2E/ALf3/yMcMf3w4nzp+R/vDhj5o6OXn25oeP8WkBAjYDVD4zfuVgJQ8IqxkFo2AUjIIRDQA2kEuMAs2HWwAAAABJRU5ErkJggg==","orcid":"","institution":"Puren Hospital Affiliated to Wuhan University of Science and Technology","correspondingAuthor":true,"prefix":"","firstName":"Jianxin","middleName":"","lastName":"Cao","suffix":""}],"badges":[],"createdAt":"2025-07-24 13:23:26","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7206057/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7206057/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":90472410,"identity":"831b2f44-b6bb-49aa-a141-fefbb4f409af","added_by":"auto","created_at":"2025-09-03 06:34:40","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":3815653,"visible":true,"origin":"","legend":"\u003cp\u003ea patient with a body mass index of 27.5 kg/m\u003csup\u003e2\u003c/sup\u003e scanned using SDCT with an effective dose of 7.7mSv and ULDCT with an effective dose of 0.4mSv. SDCT-HIR image (A) achieved scores of 5 in image graininess, striped artifacts, small vessel clarity, lung tissue density homogeneity, and overall image quality. In contrast, ULDCT-HIR image (B) had scores of 3 in these evaluating indicators. However, ULDCT- AIIR image (C) achieved scores of 5 in these evaluating indicators.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7206057/v1/f973f7b4750dffe47671a10a.png"},{"id":90472403,"identity":"ac228486-2c48-415c-a24a-ff0c08dcea39","added_by":"auto","created_at":"2025-09-03 06:34:39","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":8811644,"visible":true,"origin":"","legend":"\u003cp\u003ea patient with a body mass index of 22.8 kg/m\u003csup\u003e2\u003c/sup\u003e scanned using SDCT with an effective dose of 6.2 mSv and ULDCT with an effective dose of 0.3 mSv. SDCT-HIR images (A, D) exhibited multiple pGGNs. Compared to SDCT-HIR images, ULDCT-HIR images (B, E) exhibited increased graininess and striped artifacts, resulting in reduced clarity of certain pGGNs. However, the image quality of ULDCT-AIIR images (C, F) aligned with that of SDCT-HIR images.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7206057/v1/56e240479f2b97d2dfe5b896.png"},{"id":90474130,"identity":"bcc48310-440c-4c8f-843c-1c13c8d43992","added_by":"auto","created_at":"2025-09-03 06:42:40","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":4604725,"visible":true,"origin":"","legend":"\u003cp\u003ea patient with a body mass index of 27.8 kg/m\u003csup\u003e2\u003c/sup\u003e scanned using SDCT with an effective dose of 6.5 mSv and ULDCT with an effective dose of 0.3 mSv. SDCT-HIR image (A) exhibited no graininess or striped artifacts, presenting a clear pGGN in the right upper lobe. Conversely, ULDCT-HIR image (B) demonstrated significant graininess and striped artifacts, obscuring the pGGN in the right upper lobe, which was very blurry. In contrast, ULDCT-AIIR image (C) exhibited no graininess or striped artifacts, revealing a clear pGGN in the right upper lobe.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7206057/v1/0803c4a94b8f6f47c7b91df4.png"},{"id":90475572,"identity":"5d5c923b-03f4-4a28-9bcf-8b0bd3e1eb89","added_by":"auto","created_at":"2025-09-03 06:58:49","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":18681587,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7206057/v1/0fac9877-f28c-4183-ba32-e1c78df2a3f5.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Artificial intelligence model-based iterative reconstruction for lung ultra-low-dose CT: image quality, ground-glass nodules detectability, and Lung-RADS evaluation","fulltext":[{"header":"Introduction","content":"\u003cp\u003eLung cancer remains the most prevalent malignancy, and early detection and intervention can significantly enhance the 5-year survival rate of patients with non-small cell lung cancer, raising it to 80% [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. The National Lung Screening Trial in the US revealed that lung low-dose CT(LDCT) scans decrease the mortality risk of lung cancer patients by 20% compared to chest X-rays [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Current guidelines suggest a maximum radiation dose of 3.0 mSv for LDCT in non-obese individuals, significantly exceeding the radiation exposure from chest X-rays, which falls between 0.03 to 0.1 mSv [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Hence, there is a need to further minimize the radiation dose associated with lung CT scans.\u003c/p\u003e\u003cp\u003ePulmonary ground-glass nodule (GGN), encompassed pure GGN (pGGN) and mixed GGN (mGGN), serve as the primary CT manifestation of early lung cancer, making it crucial to classify them based on Lung Imaging-Report and Data System (Lung-RADS). However, excessive reduction in radiation dose can compromise the detection accuracy of GGNs and Lung-RADS classification due to the subtle density differences between GGNs and lung tissue. Iterative reconstruction has proven effective in minimizing image noise for lung LDCT and ultra-low-dose CT (ULDCT) [\u003cspan additionalcitationids=\"CR6 CR7 CR8 CR9 CR10 CR11 CR12 CR13 CR14\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], in the model-based iterative reconstruction (MBIR) algorithm, the regularization component used can significantly reduce noise in ULDCT images, but it may also lead to image distortion and show notable differences compared to filtered back projection reconstructed images. Consequently, diagnosing GGNs on lung ULDCT images remains a challenge.\u003c/p\u003e\u003cp\u003eIn recent years, artificial intelligence (AI)-based image reconstruction techniques have further minimized noise in ULDCT images while improving the identification accuracy of pulmonary nodule, particularly solid nodules [\u003cspan additionalcitationids=\"CR17 CR18 CR19 CR20 CR21 CR22 CR23 CR24\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. However, clinical studies focusing on the detection of GGNs using AI-based deep learning reconstruction for ULDCT remain limited. Moreover, in most prior lung ULDCT studies, the employed deep learning reconstruction algorithms were trained solely in the image domain to reduce noise, fundamentally incapable of eliminating streak and cone-beam artifacts. Notably, an AI MBIR (AIIR) algorithm has been introduced for dramatic decreasing images noise [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. This innovative algorithm replaces the regularization component of MBIR with an AI-driven convolutional neural network. Through iterative cycles of forward and backward projections, AIIR integrates comprehensive modeling of optical properties, noise characteristics, anatomical structures, and physical statistics. Simultaneously, it performs deep learning-driven CT image reconstruction in both projection and image domains, achieving three significant advancements: (1) a remarkable reduction in CT image noise, (2) substantial elimination of streak and cone-beam artifacts [\u003cspan additionalcitationids=\"CR24\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e], and (3) resolution of the inherent issue of slow reconstruction speed associated with MBIR.\u003c/p\u003e\u003cp\u003eThe aim of our study was to assess the impact of AIIR on image quality of lung ULDCT, as well as its influence on the detection and diagnostic classification of GGNs.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003e\u003cb\u003eStudy population\u003c/b\u003e\u003c/p\u003e\u003cp\u003e This prospective, observational study was granted approval by the medical ethics committee (No.202200601) of our hospital, and written informed consent was s obtained from all patients at our hospital. The inclusion criteria for the study were as follows: (1) patients undergoing lung CT follow-up for GGN, with a previous CT examination confirming the presence of at least one GGN; (2) patients without any metal implants or internal fixators in the chest; (3) patients who were not pregnant. The exclusion criteria were as follow: (1) presence of respiratory motion artifacts; (2) raw data collapse in ULDCT.\u003c/p\u003e\u003cp\u003e\u003cb\u003eCT examination and image reconstruction.\u003c/b\u003e\u003c/p\u003e\u003cp\u003eLung CT examinations were performed using a 320-row CT scanner (uCT960+, United Image Healthcare, Shanghai, China). Initially, a standard-dose CT (SDCT) scan was executed, followed by an ULDCT scan. The SDCT scan parameters included a tube voltage of 120kV and automatic tube current adjustment with a reference tube current time product of 120mAs. For the ULDCT scan, the parameters were set to a tube voltage of 100kV and automatic tube current adjustment with a reference tube current time product of 10mAs. Both scans shared consistent parameters, specifically a collimation width of 80mm, a pitch of 1.0, and a gantry rotation speed of 0.5s/rot. A hybrid iterative reconstruction (HIR) and a lung sharpness algorithm was employed to produce SDCT-HIR and ULDCT-HIR images, while a lung sharpness algorithm was employed to produce ULDCT-AIIR images. All images were obtained using an image thickness and spacing of 1mm, a matrix of 512\u0026times;512, a window width of 1400Hu, and a window level of -500Hu.\u003c/p\u003e\u003cp\u003e\u003cb\u003eRadiation dose assessment\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe volume CT dose index and radiation dose-length product was recorded. The effective dose was calculated by multiplying the dose-length product with a chest conversion coefficient (k\u0026thinsp;=\u0026thinsp;0.014 mSv/mGy cm) [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e\u003cp\u003e\u003cb\u003eLung CT image quality assessment\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe regions of interest were positioned on the axial images within both pectoralis major muscles at the level of the second thoracic vertebra, as well as within the thoracic aorta at the tracheal bifurcation level and the diaphragm's apex level. The regions of interest were maximized while excluding the aortic wall and calcification, ensuring consistency in size and location across SDCT-HIR, ULDCT-HIR, and ULDCT-AIIR images. The noise at the lung apex was calculated based on the mean of the standard deviation of CT values for both psoas majors. Similarly, the noise at the middle and bottom parts of the lung was calculated based on the standard deviation of CT values for the thoracic aorta at the tracheal bifurcation level and the diaphragm's apex level, respectively.\u003c/p\u003e\u003cp\u003eThe image graininess, striped and cone-beam artifacts, clarity of small blood vessels (\u0026lt;\u0026thinsp;3mm in diameter), homogeneity of lung tissue density, and overall image quality were scored using a double-blind method and a five-point scale by radiologists 1 and 2, one with 23 years of experience and the other with 10 years of experience in chest CT (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). All GGNs larger than 3 mm in diameter on SDCT-HIR images were identified through manual interpretation by these two radiologists combined with a deep learning-based computer-aided diagnosis system (United Intelligent Healthcare, uAI-ChestCare). Subsequently, both radiologists thoroughly reviewed the images and confirmed the authenticity of all identified GGNs. The diameter of each GGN was determined by averaging its long and short diameters.\u003c/p\u003e\u003cp\u003eAfter a 2-month interval, the clarity of GGNs was scored by radiologist 1 and radiologist 2, who were only informed of the locations of GGNs but uninformed of patient information and image types, independently evaluated the GGNs with a diameter\u0026thinsp;\u0026gt;\u0026thinsp;3mm on ULDCT-HIR and ULDCT-AIIR images in a random order. The scoring criteria were elaborated in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eImage Quality Scoring Criteria\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eevaluating indicator\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eScoring criteria\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eImage graininess\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1\u0026thinsp;=\u0026thinsp;severe graininess; 2\u0026thinsp;=\u0026thinsp;significant graininess; 3\u0026thinsp;=\u0026thinsp;moderate graininess; 4\u0026thinsp;=\u0026thinsp;slight graininess; 5\u0026thinsp;=\u0026thinsp;no graininess\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStriped and cone-beam artifacts\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1\u0026thinsp;=\u0026thinsp;severe striped and cone-beam artifacts; 2\u0026thinsp;=\u0026thinsp;significant striped and cone-beam artifacts; 3\u0026thinsp;=\u0026thinsp;moderate striped and cone-beam artifacts; 4\u0026thinsp;=\u0026thinsp;slight striped and cone-beam artifacts; 5\u0026thinsp;=\u0026thinsp;no striped or cone-beam artifacts\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClarity of small blood vessels (\u0026lt;\u0026thinsp;3mm in diameter)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1\u0026thinsp;=\u0026thinsp;very poor clarity; 2\u0026thinsp;=\u0026thinsp;poor clarity; 3\u0026thinsp;=\u0026thinsp;average clarity; 4\u0026thinsp;=\u0026thinsp;good clarity; 5\u0026thinsp;=\u0026thinsp;excellent clarity\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHomogeneity of lung tissue density\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1\u0026thinsp;=\u0026thinsp;severe unevenness; 2\u0026thinsp;=\u0026thinsp;significant unevenness; 3\u0026thinsp;=\u0026thinsp;moderate unevenness; 4\u0026thinsp;=\u0026thinsp;slight unevenness; 5\u0026thinsp;=\u0026thinsp;uniform lung tissue density\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOverall image quality\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1\u0026thinsp;=\u0026thinsp;very poor; 2\u0026thinsp;=\u0026thinsp;poor; 3\u0026thinsp;=\u0026thinsp;average; 4\u0026thinsp;=\u0026thinsp;good; 5\u0026thinsp;=\u0026thinsp;excellent\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClarity of GGNs\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1\u0026thinsp;=\u0026thinsp;completely invisible to the naked eye; 2\u0026thinsp;=\u0026thinsp;very blurry yet observable; ; 3\u0026thinsp;=\u0026thinsp;slightly blurry; 4\u0026thinsp;=\u0026thinsp;clear; 5\u0026thinsp;=\u0026thinsp;very clear\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eLung GGN detection\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTwo additional radiologists (radiologist 3 and radiologist 4, with 20 and 8 years of chest CT experience, respectively) independently identified GGNs\u0026thinsp;\u0026gt;\u0026thinsp;3 mm in diameter through manual interpretation on both ULDCT-HIR and ULDCT-AIIR images. The two readers were unaware of both image types and GGN locations. Using SDCT-HIR images as the reference standard, we assessed the detection of GGNs\u0026thinsp;\u0026gt;\u0026thinsp;3 mm on ULDCT-HIR and ULDCT-AIIR images.\u003c/p\u003e\u003cp\u003e\u003cb\u003eLung-RADS classification for GGNs\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTwo months after completing the GGNs detection assessment, radiologists 3 and radiologist 4 only informed of the locations of GGNs, but uninformed of patient information and image types, independently classified the GGNs with a diameter\u0026thinsp;\u0026gt;\u0026thinsp;3mm on SDCT-HIR, ULDCT-HIR and ULDCT-AIIR images according to Lung-RADS2022 in a random order [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. The consistency in Lung-RADS classification between the two sets of ULDCT images and SDCT-HIR images was assessed.\u003c/p\u003e\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eStatistical analysis\u003c/h2\u003e\u003cp\u003eStatistical analysis was performed using SPSS 22.0 software (version 22.0; IBM SPSS Statistics). The Friedman test and subsequent pairwise tests were used for multiple group comparisons of quantitative data. The Wilcoxon test was used for comparison of two paired quantitative data. For comparisons of count data, the McNamar test was employed. Consistency evaluation was conducted using the Kappa test (k\u0026thinsp;\u0026ge;\u0026thinsp;0.80 indicates excellent consistency, 0.60\u0026thinsp;\u0026lt;\u0026thinsp;k\u0026thinsp;\u0026lt;\u0026thinsp;0.80 indicates good consistency, 0.40\u0026thinsp;\u0026lt;\u0026thinsp;k\u0026thinsp;\u0026le;\u0026thinsp;0.60 indicates moderate consistency, 0.20\u0026thinsp;\u0026lt;\u0026thinsp;k\u0026thinsp;\u0026le;\u0026thinsp;0.40 indicates fair consistency, and k\u0026thinsp;\u0026le;\u0026thinsp;0.20 indicates poor consistency). A P-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cb\u003eDemographic characteristics, CT radiation dose, and GGNs on SDCT-HIR images\u003c/b\u003e\u003c/p\u003e\u003cp\u003eFrom July 1, 2022 to August 1, 2023, a total of 53 patients were recruited at our hospital. Among the 53 patients initially considered, 2 patients were excluded because of respiratory motion artifacts, while another patient was excluded due to the collapse of raw data related to ULDCT. Consequently, a total of 50 patients were included in final. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e provides an overview of their demographic characteristics as well as the CT radiation dose. When compared to SDCT, ULDCT demonstrated a remarkable 93.7% reduction in radiation dose (P\u0026lt;0.001).\u003c/p\u003e\u003cp\u003eA total of 81 GGNs were included in this study, comprising 68 pGGNs and 13 mGGNs, as identified on SDCT-HIR images. The average diameter was 7.6\u0026thinsp;\u0026plusmn;\u0026thinsp;3.4mm, ranging from 4 to 22mm. The average CT value was \u0026minus;\u0026thinsp;593.5\u0026thinsp;\u0026plusmn;\u0026thinsp;128.9 Hu, with a range from \u0026minus;\u0026thinsp;845.0 to -289.6 Hu. Compared to previous CT images, 12 GGNs showed increase in size or solid component volume, while none showed any reduction in size or solid component volume.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eDemographic Characteristics of Patients and CT Radiation Dose Parameters (n\u0026thinsp;=\u0026thinsp;50)\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCharacteristic\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNumber (%), mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD (range)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge (years)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e51.9\u0026thinsp;\u0026plusmn;\u0026thinsp;12.8 (26\u0026ndash;73)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMale gender (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e20 (40.0%)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWeight (kg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e64.3\u0026thinsp;\u0026plusmn;\u0026thinsp;10.4 (41.0\u0026ndash;90.0)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBody mass index (kg/m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e23.2\u0026thinsp;\u0026plusmn;\u0026thinsp;2.4 (17.0-29.4)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSDCT radiation dose parameters\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRadiation dose-length product (mGy\u0026middot;cm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e448.0\u0026thinsp;\u0026plusmn;\u0026thinsp;70.3 (300.0-607.3)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVolume CT dose index (mGy)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e11.2\u0026thinsp;\u0026plusmn;\u0026thinsp;1.4 (8.4\u0026ndash;15.7 )\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEffective radiation dose (mSv)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e6.27\u0026thinsp;\u0026plusmn;\u0026thinsp;0.98 (4.19\u0026ndash;8.50)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eULDCT radiation dose parameters\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVolume CT dose index (mGy\u0026middot;cm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.7\u0026thinsp;\u0026plusmn;\u0026thinsp;0.2 (0.5\u0026ndash;1.3)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRadiation dose and length product (mGy\u0026middot;cm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e28.7\u0026thinsp;\u0026plusmn;\u0026thinsp;9.4 (18.1\u0026ndash;53.4)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEffective radiation dose (mSv)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.40\u0026thinsp;\u0026plusmn;\u0026thinsp;0.13 (0.25\u0026ndash;0.75 )\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eLung CT Image Quality\u003c/b\u003e\u003c/p\u003e\u003cp\u003eUpon conducting an overall comparison of the noise among SDCT-HIR, ULDCT-HIR, and ULDCT-AIIR images, significant differences were observed. Furthermore, pairwise comparisons between any two sets of CT images also revealed notable disparities (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Specifically, ULDCT-AIIR images exhibited the lowest noise (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe interobserver agreement for image quality assessment demonstrated good consistency, with κ values as follows: 0.770 (95% CI: 0.684\u0026ndash;0.856) for graininess, 0.754 (95% CI: 0.662\u0026ndash;0.856) for striped and cone-beam artifacts, 0.742 (95% CI: 0.652\u0026ndash;0.832) for small vessel clarity, 0.746 (95% CI: 0.658\u0026ndash;0.834) for lung tissue density homogeneity, and 0.715 (95% CI: 0.623\u0026ndash;0.807) for overall image quality. Notably, the scores of these indicators for ULDCT-HIR images were significantly lower than those for SDCT-HIR images. However, the scores for ULDCT-AIIR images were significantly improved compared to those for ULDCT-HIR images and were on par with those for SDCT-HIR images (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eImage Quality Assessment (50 patients with 81 GGNs)\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEvaluating indicator\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSDCT-HIR\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eULDCT-HIR\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eULDCT-AIIR\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eP value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePulmonary apex noise (Hu)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e64.7\u0026thinsp;\u0026plusmn;\u0026thinsp;11.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e120.4\u0026thinsp;\u0026plusmn;\u0026thinsp;30.0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e51.7\u0026thinsp;\u0026plusmn;\u0026thinsp;7.1\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNoise at the middle of the lung (Hu)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e70.0\u0026thinsp;\u0026plusmn;\u0026thinsp;12.1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e107.4\u0026thinsp;\u0026plusmn;\u0026thinsp;18.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e52.8\u0026thinsp;\u0026plusmn;\u0026thinsp;4.4\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNoise at the bottom of the lung\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e64.8\u0026thinsp;\u0026plusmn;\u0026thinsp;11.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e121.3\u0026thinsp;\u0026plusmn;\u0026thinsp;22.6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e48.3\u0026thinsp;\u0026plusmn;\u0026thinsp;6.0\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eScore of image quality\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRadiologist 1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eImage graininess\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e4.98\u0026thinsp;\u0026plusmn;\u0026thinsp;0.14\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.64\u0026thinsp;\u0026plusmn;\u0026thinsp;0.53\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e4.76\u0026thinsp;\u0026plusmn;\u0026thinsp;0.52\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStriped and cone-beam artifacts\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e4.96\u0026thinsp;\u0026plusmn;\u0026thinsp;0.20\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.34\u0026thinsp;\u0026plusmn;\u0026thinsp;0.48\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e4.86\u0026thinsp;\u0026plusmn;\u0026thinsp;0.35\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSmall vessel clarity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e4.96\u0026thinsp;\u0026plusmn;\u0026thinsp;0.20\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.82\u0026thinsp;\u0026plusmn;\u0026thinsp;0.60\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e4.64\u0026thinsp;\u0026plusmn;\u0026thinsp;0.56\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLung tissue density homogeneity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e4.88\u0026thinsp;\u0026plusmn;\u0026thinsp;0.33\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.94\u0026thinsp;\u0026plusmn;\u0026thinsp;0.59\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e4.58\u0026thinsp;\u0026plusmn;\u0026thinsp;0.61\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOverall image quality\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e4.96\u0026thinsp;\u0026plusmn;\u0026thinsp;0.20\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.68\u0026thinsp;\u0026plusmn;\u0026thinsp;0.55\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e4.72\u0026thinsp;\u0026plusmn;\u0026thinsp;0.54\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClarity of GGNs\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e3.70\u0026thinsp;\u0026plusmn;\u0026thinsp;0.98\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.64\u0026thinsp;\u0026plusmn;\u0026thinsp;0.81\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e3.65\u0026thinsp;\u0026plusmn;\u0026thinsp;0.91\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRadiologist 2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eImage graininess\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e4.98\u0026thinsp;\u0026plusmn;\u0026thinsp;0.14\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.64\u0026thinsp;\u0026plusmn;\u0026thinsp;0.53\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e4.80\u0026thinsp;\u0026plusmn;\u0026thinsp;0.45\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStriped and cone-beam artifacts\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e4.90\u0026thinsp;\u0026plusmn;\u0026thinsp;0.30\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.40\u0026thinsp;\u0026plusmn;\u0026thinsp;0.49\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e4.92\u0026thinsp;\u0026plusmn;\u0026thinsp;0.27\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSmall vessel clarity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e4.90\u0026thinsp;\u0026plusmn;\u0026thinsp;0.30\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.88\u0026thinsp;\u0026plusmn;\u0026thinsp;0.56\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e4.74\u0026thinsp;\u0026plusmn;\u0026thinsp;0.49\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLung tissue density homogeneity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e4.84\u0026thinsp;\u0026plusmn;\u0026thinsp;0.37\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.90\u0026thinsp;\u0026plusmn;\u0026thinsp;0.54\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e4.58\u0026thinsp;\u0026plusmn;\u0026thinsp;0.54\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOverall image quality\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e4.90\u0026thinsp;\u0026plusmn;\u0026thinsp;0.30\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.80\u0026thinsp;\u0026plusmn;\u0026thinsp;0.45\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e4.74\u0026thinsp;\u0026plusmn;\u0026thinsp;0.49\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClarity of GGNs\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e3.70\u0026thinsp;\u0026plusmn;\u0026thinsp;0.95\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.58\u0026thinsp;\u0026plusmn;\u0026thinsp;0.89\u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e\u003cp\u003e3.70\u0026thinsp;\u0026plusmn;\u0026thinsp;0.95\u003csup\u003e★☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eNote: \u003csup\u003eΔ\u003c/sup\u003e significantly lower than SDCT-HIR image; \u003csup\u003e★\u003c/sup\u003e no statistical difference compared to SDCT-HIR images; \u003csup\u003e☆\u003c/sup\u003e significantly higher than ULDCT-HIR images; SDCT-HIR: standard-dose CT hybrid iterative reconstruction; ULDCT-HIR: ultra-low-dose CT hybrid iterative reconstruction; ULDCT-AIIR: ultra-low-dose CT artificial intelligence model-based iterative reconstruction\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe interobserver consistency of GGN clarity score was good ( κ\u0026thinsp;=\u0026thinsp;0.666, 95% CI:0.593\u0026ndash;0.739). The score for GGN clarity demonstrated comparable results on ULDCT-AIIR and SDCT-HIR images, while the score for GGN clarity on ULDCT-AIIR images was better than that on ULDCT-HIR images.\u003c/p\u003e\u003cp\u003e\u003cb\u003eGGNs detection on ULDCT images\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe detection results for the GGNs with a diameter\u0026thinsp;\u0026gt;\u0026thinsp;3mm are presented in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. It was observed that the detection rate was higher on ULDCT-AIIR images compared to ULDCT-HIR images (all P-values\u0026thinsp;=\u0026thinsp;0.000).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eTest Results of Detection of GGNs (n\u0026thinsp;=\u0026thinsp;81)\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eGGNs\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eRadiologist 3\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u003cp\u003eRadiologist 4\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eULDCT-HIR\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eULDCT-AIIR\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eULDCT-HIR\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eULDCT-AIIR\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDetected (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e52 (64.2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e77 (95.1%) \u003csup\u003e☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e55 (67.9%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e78 (96.3%) \u003csup\u003e☆\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eUndetected (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e29 (35.8%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4 (4.9%) \u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e26 (32.1%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e3 (3.7%) \u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003epGGN (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e27 (33.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4 (4.9%) \u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e26 (32.1%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e3 (3.7%) \u003csup\u003eΔ\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003emGGN (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2 (2.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0 (0.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0 (0.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0 (0.0%)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDiameter (mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e4.0-14.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4.0-4.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e4.0\u0026ndash;7.0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e4.0\u0026ndash;7.0\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFalse positive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"6\"\u003eNote: ULDCT-HIR: Ultra-low-dose CT hybrid iterative reconstruction; ULDCT-AIIR: Ultra-low-dose CT artificial intelligence model-based iterative reconstruction; pGGN: pure ground glass nodule; mGGN: mixed ground glass nodule; \u003csup\u003e☆\u003c/sup\u003esignificantly higher than ULDCT-HIR images; \u003csup\u003eΔ\u003c/sup\u003esignificantly lower than ULDCT-HIR images\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eLung-RADS classification\u003c/strong\u003e\u003c/p\u003e\u003cp\u003eThe Lung-RADS classification results are presented in Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e. The interobserver consistency proved to be good ( κ\u0026thinsp;=\u0026thinsp;0.707, 95% CI:0.609\u0026ndash;0.805). Inter-modality agreement analysis revealed moderate consistency between ULDCT-HIR and SDCT-HIR images, with κ values of 0.343 (95% CI: 0.153\u0026ndash;0.533) for Radiologist 3 and 0.411 (95% CI: 0.227\u0026ndash;0.595) for Radiologist 4. In contrast, ULDCT-AIIR demonstrated significantly better agreement with SDCT-HIR, showing good consistency for both radiologists (Radiologist 3: κ\u0026thinsp;=\u0026thinsp;0.772, 95% CI: 0.600-0.944; Radioloist: κ\u0026thinsp;=\u0026thinsp;0.743, 95% CI: 0.563\u0026ndash;0.923).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eLung-RADS Category of GGNs (n\u0026thinsp;=\u0026thinsp;81)\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"7\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colspan=\"6\" nameend=\"c7\" namest=\"c2\"\u003e\u003cp\u003eQuantity\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eLung-RADS\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e\u003cp\u003eRadiologist 3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c7\" namest=\"c5\"\u003e\u003cp\u003eRadiologist 4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSDCT-HIR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eULDCT-HIR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eULDCT-AIIR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eSDCT-HIR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eULDCT-HIR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003eULDCT-AIIR\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e14\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e68\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e54\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e64\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e65\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e53\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e66\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e4A\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e4B\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e4X\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e6\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"7\"\u003eNote: SDCT-HIR: standard-dose CT hybrid iterative reconstruction; ULDCT-HIR: ultra-low-dose CT hybrid iterative reconstruction; ULDCT-AIIR: ultra-low-dose CT artificial intelligence model-based iterative reconstruction\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eRecent studies have demonstrated that AI-based denoising techniques significantly improve image quality and enhance the detection rate of pulmonary nodules on ULDCT [\u003cspan additionalcitationids=\"CR18\" citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. However, most existing research has focused on solid pulmonary nodules, with limited investigation into their impact on GGNs, particularly pGGNs, in ULDCT imaging. Furthermore, while the majority of prior studies employed AI denoising methods trained exclusively in the image domain, research on dual-domain deep learning approaches\u0026mdash;combining both image and projection domain training (e.g., AIIR technology)\u0026mdash;remains scarce in the context of lung ULDCT. Yang et al. [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e] reported that AIIR significantly enhanced the detection rate for pulmonary nodules on lung ULDCT, but they did not assess the impact of AIIR on image quality and Lung-RADS classification. In our study, we found that AIIR significantly improved the image quality of ULDCT, achieving a remarkable 93.7% reduction in radiation dose compare to SDCT. Notably, ULDCT-AIIR images demonstrated exceptional detection rates for GGNs while maintaining excellent concordance with SDCT-HIR images in Lung-RADS classification. This advancement enables clear visualization and accurate diagnostic classification of GGNs while achieving a dramatic reduction in radiation dose from pulmonary CT scans. It holds particular significance for patients requiring long-term GGN surveillance, as it substantially mitigates the risks associated with cumulative ionizing radiation exposure\u0026mdash;especially critical given recent studies highlighting potential carcinogenic effects from CT scans[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eDuring the quantitative assessment of image noise, we discovered that ULDCT-HIR images exhibited notable noise levels. However, the noise of ULDCT-AIIR images was dramatically reduced, even lower than the low noise of SDCT-HIR images. Furthermore, the image graininess score of ULDCT-AIIR images underwent a significant enhancement when compared to ULDCT-HIR images. Additionally, AIIR technology excelled in minimizing striped artifacts. These impressive results are attributed to the synergistic benefits of AI and MBIR inherent in AIIR, which endow it with exceptional image denoising proficiency, outperforming HIR [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThis study revealed that, despite no significant disparity in the score of lung tissue density homogeneity between ULDCT-AIIR and SDCT-HIR images, the former exhibited slightly lower homogeneity. This observation can be attributed to the utilization of a sharpening algorithm, rather than a standard one, during the reconstruction of ULDCT-AIIR images to achieve clearer lung visualizations. Prior research has indicated that model-based iterative reconstruction, when employed in lung sharpening algorithms, tends to result in the emergence of uneven normal lung tissue [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Given that AIIR incorporates model-based iterative reconstruction data information [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e], it follows that lung tissue unevenness may manifest in AIIR images generated using the lung sharpening algorithm.\u003c/p\u003e\u003cp\u003eOur study demonstrated that the detection rate of GGNs on ULDCT-AIIR images reached remarkably high values of 95.1% and 96.3%. Only a minimal number of pGGNs were missed, all of which measured\u0026thinsp;\u0026le;\u0026thinsp;7 mm in diameter. Nevertheless, these undetected GGNs on ULDCT-AIIR images were clinically insignificant. Although false-positive GGNs were observed on ULDCT-AIIR images, their occurrence was notably infrequent. Retrospective analysis suggested this phenomenon might be attributed to suboptimal visualization of localized image details.\u003c/p\u003e\u003cp\u003eThe consistency of Lung-RADS classification between ULDCT-HIR and SDCT-HIR images was moderate, primarily due to the high noise of ULDCT-HIR images. This high noise of ULDCT-HIR images obscured some GGNs, leading to their classification as category 1. Additionally, even when a portion of the GGNs was not fully obscured by the high noise of ULDCT-HIR images, the diagnosis was still affected, resulting in some GGNs being classified as category 0. However, it's worth noting that the inconsistent was rare between ULDCT-AIIR and SDCT-HIR images.\u003c/p\u003e\u003cp\u003eOur study also had several limitations. First, since most GGNs in this study lacked pathological confirmation, we were unable to determine the definitive diagnostic accuracy for lung cancer on ULDCT-AIIR images. However, we maintain that despite not evaluating the concordance between GGN diagnosis and pathological findings, assessing the consistency of Lung-RADS classifications across different imaging modalities remains significant clinical value. Our findings demonstrate excellent agreement in Lung-RADS categorization of GGNs between ULDCT-AIIR and SDCT-HIR images, suggesting that GGN classification remains largely unaffected when using AIIR at ultra-low radiation doses. Second, when the GGNs was classified according to Lung-RADS2022, radiologists were informed of the locations of GGNs to simplify the analysis process. Thirdly, the enrolled patients exhibited a relatively low number of GGNs in category 3 and 4; however, these GGNs possess higher CT values due to solid components, making them less susceptible to concealment by noise compared to pGGNs. Lastly, our study did not include patients with a BMI exceeding 30 kg/m\u003csup\u003e2\u003c/sup\u003e, necessitating further verification of feasibility in obese individuals.\u003c/p\u003e\u003cp\u003eIn summary, AIIR significantly can enhance image quality in ULDCT while maintaining excellent GGNs detection capability. Moreover, there is good consistency in Lung-RADS classification between ULDCT-AIIR and SDCT-HIR images.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eCT \u0026nbsp;\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;computed tomography\u003c/p\u003e\n\u003cp\u003eLDCT\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;low-dose CT\u003c/p\u003e\n\u003cp\u003eGGN\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;pulmonary ground-glass nodule\u003c/p\u003e\n\u003cp\u003epGGN\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;pure GGN\u003c/p\u003e\n\u003cp\u003emGGN\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;mixed GGN\u003c/p\u003e\n\u003cp\u003eLung-RADS\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Lung Imaging-Report and Data System\u003c/p\u003e\n\u003cp\u003eULDCT\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;ultra-low-dose CT\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSDCT\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;standard-dose CT\u003c/p\u003e\n\u003cp\u003eMBIR\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;model-based iterative reconstruction\u003c/p\u003e\n\u003cp\u003eAI\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;artificial intelligence\u003c/p\u003e\n\u003cp\u003eAIIR\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;artificial intelligence model-based iterative reconstruction\u003c/p\u003e\n\u003cp\u003eHIR \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;hybrid iterative reconstruction\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eYC: Collection of cases, Conceptualization, Writing\u0026mdash;Original Draft, Investigation, Data curation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eJD: Collection of cases, Investigation, Formal analysis, Data curation\u003c/p\u003e\n\u003cp\u003eZZ: Image Postprocessing, Investigation, Data curation\u003c/p\u003e\n\u003cp\u003eXC: Collection of cases, Investigation, Data curation\u003c/p\u003e\n\u003cp\u003eJL: Software, Postprocessing, Data curation\u003c/p\u003e\n\u003cp\u003eWX: Conceptualization, Investigation\u003c/p\u003e\n\u003cp\u003eCH: Collection of cases, Investigation\u003c/p\u003e\n\u003cp\u003eQZ: Investigation, Review\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eJC: Investigation, Writing, Review \u0026amp; Editing, Methodology, Resources, Supervision.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was supported by Hubei Province Key Laboratory of Occupational Hazard Identification and Control, Wuhan University of Science and Technology（No：OHIC2021G04).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used and analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study has been reviewed and approved by the Ethics Committee of Puren Hospital Affiliated with Wuhan University of Science and Technology (Approval No. 202200601), and strictly adheres to the ethical principles outlined in the Declaration of Helsinki. In compliance with the guidelines established by the Council for International Organizations of Medical Sciences (CIOMS), written informed consent was obtained from all participants. To ensure privacy and confidentiality, all participant data were thoroughly anonymized and de-identified. All procedures were conducted in accordance with relevant guidelines and regulatory requirements.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors have no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSEER cancer statistics review. 1975\u0026ndash;2015, national cancer institute. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://seer.cancer.gov/csr/1975_2015/\u003c/span\u003e\u003cspan address=\"https://seer.cancer.gov/csr/1975_2015/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNational Lung Screening Trial Research Team, Aberle DR, Adams AM, Berg CD, Black WC, Clapp JD, et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med. 2011;365(5):395\u0026ndash;409.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eQuaia E, Baratella E, Cernic S, Lorusso A, Casagrande F, Cioffi V, et al. Analysis of the impact of digital tomosynthesis on the radiological investigation of patients with suspected pulmonary lesions on chest radiography. Eur Radiol. 2012;22(9):1912\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBach PB, Mirkin JN, Oliver TK, Azzoli CG, Berry DA, Brawley OW, et al. Benefits and harms of CT screening for lung cancer: A systematic review. JAMA. 2012;307(22):2418\u0026ndash;29.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYe K, Zhu Q, Li M, Lu Y, Yuan H. A feasibility study of pulmonary nodule detection by ultralow-dose CT with adaptive statistical iterative reconstruction-V technique. Eur J Radiol. 2019;119:108652.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJu YH, Lee G, Lee JW, Hong SB, Suh YJ, Jeong YJ. Ultra-low-dose lung screening CT with model-based iterative reconstruction: An assessment of image quality and lesion conspicuity. Acta Radiol. 2018;59(5):553\u0026ndash;59.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTakahashi EA, Koo CW, White DB, Lindell RM, Sykes AG, Levin DL, et al. Prospective pilot evaluation of radiologists and computer-aided pulmonary nodule detection on ultra-low-dose CT with tin filtration. J Thorac Imaging. 2018;33(6):396\u0026ndash;401.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang M, Qi W, Sun Y, Jiang Y, Liu X, Hong N. Screening for lung cancer using sub-millisievert chest CT with iterative reconstruction algorithm: Image quality and nodule detectability. Br J Radiol. 2018;91(1090):20170658.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYe K, Chen M, Li J, Zhu Q, Lu Y, Yuan H. Ultra-low-dose CT reconstructed with ASiR-V using SmartmA for pulmonary nodule detection and Lung-RADS classifications compared with low-dose CT. Clin Radiol. 2021;76(2):156.e1-156.e8.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFletcher JG, Levin DL, Sykes AG, Lindell RM, White DB, Kuzo RS, et al. Observer performance for detection of pulmonary nodules at chest CT over a large range of radiation dose levels. Radiology. 2020;297(3):699\u0026ndash;707.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eThapaliya S, Gilligan LA, Brady SL, Anton CG, Crotty EJ, Nasser MP, et al. Comparison of 0.3-mSv CT to standard-dose CT for detection of lung nodules in children and young adults with cancer. AJR Am J Roentgenol. 2021;217(6):1444\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHan D, Cai J, Heus A, Heuvelmans M, Imkamp K, Dorrius M, et al. Detection and size quantification of pulmonary nodules in ultralow-dose versus regular-dose CT: A comparative study in COPD patients. Br J Radiol. 2023;96(1144):20220709.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKim H, Park CM, Kim SH, Lee SM, Park SJ, Lee KH, et al. Persistent pulmonary subsolid nodules: Model-based iterative reconstruction for nodule classification and measurement variability on low-dose CT. Eur Radiol. 2014;24(11):2700\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHu QJ, Liu YW, Chen C, Kang SC, Sun ZY, Wang YJ, et al. Prospective study of low- and standard-dose chest CT for pulmonary nodule detection: A comparison of image quality, size measurements and radiation exposure. Curr Med Sci. 2021;41(5):966\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKim SK, Kim C, Lee KY, Cha J, Lim HJ, Kang EY, et al. Accuracy of model-based iterative reconstruction for CT volumetry of part-solid nodules and solid nodules in comparison with filtered back projection and hybrid iterative reconstruction at various dose settings: An anthropomorphic chest phantom study. Korean J Radiol. 2019;20(7):1195\u0026ndash;206.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMikayama R, Shirasaka T, Kojima T, Sakai Y, Yabuuchi H, Kondo M, et al. Deep-learning reconstruction for ultra-low-dose lung CT: Volumetric measurement accuracy and reproducibility of artificial ground-glass nodules in a phantom study. Br J Radiol. 2022;95(1130):20210915.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGorenstein L, Onn A, Green M, Mayer A, Segev S, Marom EM. A novel artificial intelligence based denoising method for ultra-low dose CT used for lung cancer screening. Acad Radiol. 2023;30(11):2588\u0026ndash;97.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGoto M, Nagayama Y, Sakabe D, Emoto T, Kidoh M, Oda S, et al. Lung-optimized deep-learning-based reconstruction for ultralow-dose CT. Acad Radiol. 2023;30(3):431\u0026ndash;40.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKim C, Kwack T, Kim W, Cha J, Yang Z, Yong HS. Accuracy of two deep learning-based reconstruction methods compared with an adaptive statistical iterative reconstruction method for solid and ground-glass nodule volumetry on low-dose and ultra-low-dose chest computed tomography: A phantom study. PLoS ONE. 2022;17(6):e0270122.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCao Q, Mao Y, Qin L, Quan G, Yan F, Yang W. Improving image quality and lung nodule detection for low-dose chest CT by using generative adversarial network reconstruction. Br J Radiol. 2022;95(1138):20210125.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSingh R, Digumarthy SR, Muse VV, Kambadakone AR, Blake MA, Tabari A, et al. Image quality and lesion detection on deep learning reconstruction and iterative reconstruction of submillisievert chest and abdominal CT. AJR Am J Roentgenol. 2020;214(3):566\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKoetzier LR, Mastrodicasa D, Szczykutowicz TP, van der Werf NR, Wang AS, Sandfort V, et al. Deep learning image reconstruction for CT: Technical principles and clinical prospects. Radiology. 2023;306(3):e221257.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJi MT, Wang RR, Wang Q, Li HS, Zhao YX. Feasibility study of double-low scanning protocol combined with artificial intelligence iterative reconstruction algorithm for abdominal computed tomography enhancement in patients with obesity. BMC Med Imaging. 2025;25(1):276.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGong H, Peng L, Du X, An J, Peng R, Guo R, et al. Artificial intelligence iterative reconstruction in computed tomography angiography: an evaluation on pulmonary arteries and aorta with routine dose settings. J Comput Assist Tomogr. 2024;48(2):244\u0026ndash;50.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLi J, Zhu J, Zou Y, Zhang G, Zhu P, Wang N, et al. Diagnostic CT of colorectal cancer with artificial intelligence iterative reconstruction: a clinical evaluation. Eur J Radiol. 2024;171:111301.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDeak PD, Smal Y, Kalender WA. Multisection CT protocols: Sex- and age-specific conversion factors used to determine effective dose from dose-length product. Radiology. 2010;257(1):158\u0026ndash;66.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChristensen J, Prosper AE, Wu CC, Chung J, Lee E, Elicker B, Hunsaker AR, et al. ACR Lung-RADS v2022: Assessment categories and management recommendations. J Am Coll Radiol. 2024;21(3):473\u0026ndash;88.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHata A, Yanagawa M, Yoshida Y, Miyata T, Tsubamoto M, Honda O, et al. Combination of deep learning-based denoising and iterative reconstruction for ultra-low-dose CT of the chest: Image quality and Lung-RADS evaluation. AJR Am J Roentgenol. 2020;215(6):1321\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang L, Liu H, Han J, Xu S, Zhang G, Wang Q, et al. Ultra-low-dose CT lung screening with artificial intelligence iterative reconstruction: evaluation via automatic nodule-detection software. Clin Radiol. 2023;78(7):525\u0026ndash;31.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBosch de Basea M, Thierry-Chef I, Harbron R, Hauptmann M, Byrnes G, Bernier MO et al. Risk of hematological malignancies from CT radiation exposure in children, adolescents and young adults. Nat Med. 2023;29(12):3111-9.Erratum in: Nat Med. 2025 Apr 11. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41591-025-03689-5\u003c/span\u003e\u003cspan address=\"10.1038/s41591-025-03689-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSmith-Bindman R, Chu PW, Azman Firdaus H, Stewart C, Malekhedayat M, Alber S et al. Projected lifetime cancer risks from current computed tomography imaging. JAMA Intern Med. 2025;185(6):710\u0026ndash;719.Erratum in: JAMA Intern Med. 2025;185(6):747.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHata A, Yanagawa M, Honda O, Miyata T, Tomiyama N. Ultra-low-dose chest computed tomography for interstitial lung disease using model-based iterative reconstruction with or without the lung setting. Med (Baltim). 2019;98(22):e15936.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Artificial Intelligence, Model-based Iterative Reconstruction, Ultra-low Dose, CT, Lung, Ground-glass Nodule","lastPublishedDoi":"10.21203/rs.3.rs-7206057/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7206057/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eObjective: \u003c/strong\u003eTo assess the effect of artificial intelligence model-based iterative reconstruction (AIIR) on image quality of lung ultra-low-dose CT (ULDCT), as well as its influence on the detection and diagnostic classification of GGNs.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eFifty-three patients diagnosed with GGNs underwent both lung standard-dose CT (SDCT) and ultra-low-dose CT (ULDCT) scans. SDCT hybrid iterative reconstruction (SDCT-HIR) images, ULDCT hybrid iterative reconstruction (ULDCT-HIR) images, and ULDCT-AIIR images were generated using a lung sharpness algorithm. Image noise measurements were performed. Image quality was independently scored by Radiologists 1 and 2. Separately, GGNsdetectability and Lung-RADS classification were independently evaluated by Radiologists 3 and 4.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003e​Two patients were excluded due to respiratory motion artifacts and one due to raw data errors, resulting in 50 patients with 81 GGNs for final analysis. Compared to SDCT, ULDCT achieved a remarkable 93.7% reduction in radiation dose​(SDCT: 6.27 ± 0.98 mSv vs. ULDCT: 0.40 ± 0.13 mSv, P \u0026lt; 0.001). Moreover, ULDCT-AIIR images exhibited the lowest noise levels (P \u0026lt; 0.001). The image quality scores of ULDCT-AIIR were significantly superior to those of ULDCT-HIR (P \u0026lt; 0.001) and comparable to SDCT-HIR (P \u0026gt; 0.05). For GGNs detection, Radiologist 3 reported rates of 64.2% (ULDCT-HIR) vs. 95.1% (ULDCT-AIIR), while Radiologist 4 reported 67.9% (ULDCT-HIR) vs. 96.3% (ULDCT-AIIR). ULDCT-AIIR images demonstrated significantly higher detection rates than ULDCT-HIR images (P \u0026lt; 0.001). The consistency in the Lung-RADS classification was moderate between ULDCT-HIR and SDCT-HIR images (κ=0.343 and 0.411 for radiologist 3 and 4, respectively), but good between ULDCT-AIIR and SDCT-HIR images (κ=0.772 and 0.743 for radiologist 3 and 4, respectively).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion:\u003c/strong\u003e AIIR can significantly enhance image quality in ULDCT while maintaining excellent GGNs detection capability. Moreover, there is good consistency in Lung-RADS classification between ULDCT-AIIR and SDCT-HIR images.\u003c/p\u003e","manuscriptTitle":"Artificial intelligence model-based iterative reconstruction for lung ultra-low-dose CT: image quality, ground-glass nodules detectability, and Lung-RADS evaluation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-03 06:34:35","doi":"10.21203/rs.3.rs-7206057/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"64800070012459749350280489613410354189","date":"2025-08-28T04:49:37+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-25T17:46:12+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-19T06:17:34+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-07-30T08:18:40+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-30T03:40:33+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Imaging","date":"2025-07-30T03:37:38+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"18626a6c-18e5-4283-9dfc-16ec273d5d7d","owner":[],"postedDate":"September 3rd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-09-03T06:34:35+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-03 06:34:35","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7206057","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7206057","identity":"rs-7206057","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}