Identification of the optimal threshold for predicting the infiltration degree of T1-stage lung adenocarcinoma using solid component volume and three-dimensional consolidation-to-tumor ratio in threshold segmentation

Introduction

Lung cancer has the highest incidence and mortality rates globally. In the tumor-node-metastasis (TNM) classification, the T staging of lung cancer primarily evaluates its solid components, directly impacting subsequent treatment strategies and prognosis assessment. Therefore, research on the solid components of non-pure solid nodules is particularly important. Pathologically, the presence of infiltrative cancer lesions larger than 5 mm is diagnostic for invasive adenocarcinoma (IAC), which serves as the threshold for distinguishing between minimally IAC (MIA) and IAC (1). Currently, measuring the solid components of lung tumors involves measuring the size of the solid part on lung window settings, and the average of the long and short axes is determined as its size.

Nonetheless, predicting the degree of invasion in T1-stage lung adenocarcinoma is clinically challenging, particularly in patients with heterogeneous ground-glass opacity. The size of the lung nodule, volume of the solid component, consolidation-to-tumor ratio (CTR), and the average computed tomography (CT) value are commonly used indicators for predicting the invasiveness of nodules at this stage. CTR is currently measured as the ratio of the diameter of the solid component on lung window settings to the diameter of the nodule. Numerous studies (2-5) have reported that CTR has certain value in malignancy grading, pathological subtypes, guidance for surgical planning, lymph node metastasis, and prognostic prediction in lung adenocarcinoma. However, the solid components within lung nodules may not exist as a single entity and have an irregular shape and uneven intensity; in addition, manual measurement errors and differences in window width and level settings introduce significant challenges. Therefore, artificial intelligence (AI)-based automatic measurement of volume ratios of CTR based on threshold segmentation may provide a more accurate assessment compared to diameter ratio-based CTR (6). However, no uniform standard of CT threshold settings has been established for volume measurement of solid components and 3D CTR, and the settings of different AI software vary. This study measured the solid component volume and 3D CTR of T1-stage lung adenocarcinoma under seven different thresholds [−550 to 0 Hounsfield unit (HU)] using Shukun Technology, an intelligent healthcare technology platform, which was intended to explore and define the optimal thresholds for effectively predicting the invasiveness of T1-stage lung adenocarcinoma. We present this article in accordance with the STARD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1160/rc).

Methods

Collection of clinical data and participants

A retrospective analysis was conducted on 820 cases (936 nodules) of T1-stage lung adenocarcinoma with complete postoperative pathological confirmation and imaging data at Zhengzhou People’s Hospital. Additionally, data from 236 cases (243 nodules) of T1-stage lung adenocarcinoma with complete postoperative pathological confirmation and imaging data were collected from Huadong Hospital Affiliated to Fudan University. Approval was obtained from the Institutional Review Boards of Zhengzhou People’s Hospital (No. 2024011139) and Huadong Hospital Affiliated to Fudan University (No. 2019K134), and the requirement for written informed consent for this retrospective study was waived. The study was conducted in accordance with the ethical standards of the responsible institution regarding human subjects, as well as in compliance with the Declaration of Helsinki and its subsequent amendments.

Grouping strategy

Patients with postoperative pathology confirming MIA and IAC were classified into the invasive lesion group, while patients with postoperative pathology confirming atypical adenomatous hyperplasia (AAH) and adenocarcinoma in situ (AIS) were classified into the pre-invasive lesion group.

Inclusion and exclusion criteria

Inclusion criteria

The maximum diameter of the pulmonary nodules was less than 30 mm. Patients did not receive neoadjuvant therapy such as radiotherapy or chemotherapy. The T1-stage lung adenocarcinoma (including AAH, AIS, MIA, and IAC) was confirmed by routine pathological examination after surgery. Thin-slice chest CT images were obtained within 1 month before surgery with a layer thickness of less than 1.5 mm. The clinical data of each patient were complete. All imaging data were processed using whole chest diagnostic module of Shukun Technology, which accurately identified and quantified detailed imaging phenotypic features of resected lung nodules.

Exclusion criteria

The maximum diameter of the pulmonary shadow exceeded 30 mm. Other types of tumors or benign lesions were confirmed by routine pathological examination after surgery. CT image artifacts impairing CT data analysis, or a lack of CT imaging within 1 month before surgery, or a layer thickness exceeding 1.5 mm. Incomplete patient clinical data. Inability to accurately identify imaging data and quantify detailed imaging phenotypic features using the module of Shukun Technology.

Imaging data collection and the selection of CT threshold

Chest CT examinations were mainly performed using the following four machines: Somatom Definition, NeuViz 16Classic, Somatom Definition Flash, and GE Discovery CT750 HD. The scanning parameters were as follows: 120 kVp; 100–200 mAs; pitch 0.75–1.5; and collimation 1–1.5 mm. All imaging data were reconstructed using a medium sharpness reconstruction algorithm with a thickness of 0.625–1.5 mm. The thin-layer DICOM format images that were collected were first compressed and subsequently imported into the AI workstation developed by Shukun Technology. This workstation was then utilized for a comprehensive diagnostic analysis of the entire chest. The AI software was capable of automatically parsing the metadata (such as device model and scanning parameters) within DICOM files and automatically identifying and correcting imaging differences among various devices. Furthermore, the software could automatically segment, identify, and quantify the solid component volume (mm³) and three-dimensional (3D) CTR (%) at different thresholds. A total of seven CT thresholds were included in the analysis: −550, −450, −350, −250, −150, −50, and 0 HU. The detailed process is illustrated in Figure 1. The comparison of changes in the solid components of ground-glass nodules at different thresholds is depicted in Figure 2A-2G.

Figure 1 Flowchart with the objective of uncovering the optimal thresholds for predicting the infiltration degree of T1-stage lung adenocarcinoma by means of the solid component volume and the 3D CTR. Hospital 1, Zhengzhou People’s Hospital; Hospital 2, Huadong Hospital Affiliated to Fudan University. 3D CTR, three-dimensional consolidation-to-tumor ratio; AAH, atypical adenomatous hyperplasia; AI, artificial intelligence; AIS, adenocarcinoma in situ; CT, computed tomography; HU, Hounsfield unit; IAC, invasive adenocarcinoma; MIA, minimally invasive adenocarcinoma; ROC, receiver operating characteristic.

Figure 2 Comparison of the solid components of a ground-glass nodule using the threshold segmentation method with seven different threshold settings [(A) with a threshold of −550 HU, (B) with a threshold of −450 HU, (C) with a threshold of −350 HU, (D) with a threshold of −250 HU, (E) with a threshold of −150 HU, (F) with a threshold of −50 HU, and (G) with a threshold of 0 HU]. The green areas represent the solid parts within the lung nodules, indicating regions of high density within the nodules without air or fluid-filled spaces. The red circles denote the boundaries of the lung nodules, which are the edges that distinguish the nodules from the surrounding lung tissue. HU, Hounsfield unit.

Statistical analysis

The results were analyzed and curves were plotted using SPSS 23.0 (IBM, Armonk, NY, USA) statistical analysis software. The count data were expressed as the number of cases and were analyzed by the Chi-squared test. The Kolmogorov-Smirnov test was used to analyze the quantitative data, which conformed to a normal distribution and was expressed as mean ± standard deviation. A two-sample t-test was performed to assess the differences in quantitative data. In this study, P<0.05 was considered statistically significant. Receiver operating characteristic (ROC) curves were plotted for solid component volume and 3D CTR at different thresholds, and the area under the ROC curve (AUC) was compared to evaluate the predictive performance of different threshold models. Additionally, the AUC, sensitivity, specificity, and diagnostic threshold or optimal cutoff value of all the independent predictors were analyzed and calculated by a ROC curve.

Results

Patient characteristics

Through a systematic data collection process across two specialized centers, a lot of 1,179 pulmonary nodules were successfully collected, including 366 nodules in the pre-invasive lesion group (38 AAH, 328 AIS) and 813 nodules in the invasive lesion group (371 MIA, 442 IAC). The patients had an average age of 56.67±11.51 years, with a statistically significant difference between the two groups. The pre-invasive lesion group had an average age of 52.55±11.63 years, while the invasive lesion group had an average age of 58.53±10.97 years. This indicates a higher risk of occurrence in younger individuals in the pre-invasive lesion group compared to the invasive lesion group. However, no statistically significant difference in gender was observed between the pre-invasive and invasive lesion groups (P=0.275). Solid component volume and 3D CTR calculated with the seven different threshold settings could be used to differentiate between pre-invasive and invasive lesions in T1-stage lung adenocarcinoma (P<0.05). A detailed comparison of characteristics is presented in Table 1.

ROC curve analysis for accurately predicting the infiltration degree of T1-stage lung adenocarcinoma based on solid component volume

The solid component volume was analyzed to predict the differences between the two groups. When the CT threshold was set to −550 HU, the AUC for predicting the invasiveness of stage T1 lung adenocarcinoma based on the solid component volume was 0.848, with a 95% confidence interval (CI) of 0.826 to 0.869. At −450 HU, the AUC was 0.854, with a 95% CI of 0.833 to 0.875. At −350 HU, the AUC was 0.855, with a 95% CI of 0.834 to 0.876. At -250 HU, the AUC was 0.835, with a 95% CI of 0.813 to 0.858. At −150 HU, the AUC was 0.823, with a 95% CI of 0.800 to 0.847. At −50 HU, the AUC was 0.802, with a 95% CI of 0.778 to 0.827. At 0 HU, the AUC was 0.775, with a 95% CI of 0.749 to 0.800. The highest predictive performance for the invasiveness of stage T1 lung adenocarcinoma based on the 3D solid component volume was achieved at the threshold of −350 HU, as shown in Figure 3 of the ROC curves. These results indicate that the solid component volume can reliably be used as a predictive factor for the invasiveness of stage T1 lung adenocarcinoma, especially at the CT threshold of −350 HU. The optimal threshold of −350 HU was the most discriminative for distinguishing between invasive and non-invasive cases of T1 lung adenocarcinoma.

Figure 3 ROC curves for predicting infiltration degree of T1-stage lung adenocarcinoma based on solid component volume at different thresholds. CI, confidence interval; HU, Hounsfield unit; ROC, receiver operating characteristic; V −550, solid component volume with a threshold of −550 HU; V −450, solid component volume with a threshold of −450 HU; V −350, solid component volume with a threshold of −350 HU; V −250, solid component volume with a threshold of −250 HU; V −150, solid component volume with a threshold of −150 HU; V −50, solid component volume with a threshold of −50 HU; V 0, solid component volume with a threshold of 0 HU.

ROC curves derived from 3D CTR for gauging infiltration degree of T1-stage lung adenocarcinoma

The differences in infiltration degree between the two groups were predicted based on 3D CTR. The predictive performance of 3D CTR for the invasiveness of stage T1 lung adenocarcinoma was highest at the threshold of −350 HU with an AUC 0.823 (95% CI: 0.799–0.847), as shown by the ROC curves in Figure 4. These results suggest that both the 3D CTR and the solid component volume are reliable predictive factors for the invasiveness of T1-stage lung adenocarcinoma. The optimal threshold of −350 HU indicated that this value provides the best discrimination between invasive and non-invasive cases of T1 lung adenocarcinoma, similar to the solid component volume.

Figure 4 ROC curves analysis: assessing the invasiveness of T1-stage lung adenocarcinoma with 3D CTR at diverse threshold values. 3D CTR −550, three-dimensional consolidation-to-tumor ratio with a threshold of −550 HU; 3D CTR −450, three-dimensional consolidation-to-tumor ratio with a threshold of −450 HU; 3D CTR −350, three-dimensional consolidation-to-tumor ratio with a threshold of −350 HU; 3D CTR −250, three-dimensional consolidation-to-tumor ratio with a threshold of −250 HU; 3D CTR −150, three-dimensional consolidation-to-tumor ratio with a threshold of −150 HU; 3D CTR −50, three-dimensional consolidation-to-tumor ratio with a threshold of −50 HU; 3D CTR 0, three-dimensional consolidation-to-tumor ratio with a threshold of 0 HU; CI, confidence interval; HU, Hounsfield unit; ROC, receiver operating characteristic.

Cut-off points for solid component volume and CTR in discriminating between pre-invasive and invasive lesions at the optimal threshold

Based on the ROC curve results for predicting pre-invasive and invasive lesions of T1-stage lung adenocarcinoma using solid component volume and 3D CTR at different thresholds, the threshold of −350 HU demonstrated the highest predictive performance. Similarly, the predictive performance of solid component volume was superior to that of 3D CTR at the same threshold. The ROC curves were redrawn based on solid component volume and 3D CTR at the threshold of −350 HU (Figure 5). The optimal cut-off point for solid component volume prediction was determined to be 45.5 mm³. For 3D CTR, the optimal cut-off point was 10.85%. Details are provided in Table 2.

Figure 5 Cut-off values for solid component volume and CTR in discriminating between pre-invasive and invasive lesions at the optimal threshold (−350 HU). 3D CTR −350, three-dimensional consolidation-to-tumor ratio with a threshold of −350 HU; CI, confidence interval; HU, Hounsfield unit; ROC, receiver operating characteristic; V −350, solid component volume with a threshold of −350 HU.

Discussion

T1-stage lung adenocarcinoma can be categorized into AAH, AIS, MIA, and IAC. The first two types are pre-invasive lesions and are considered to be in an “indolent” growth phase, often allowing for extended follow-up intervals; hence, their changes over time can be observed without delaying the diagnosis and treatment (7,8). In contrast, the latter two types are invasive lesions with a high risk of progression. In cases where clinical differentiation is difficult or the patient is unwilling to undergo surgery, shorter clinical follow-up intervals are required to reduce the risk of progression. Differentiating between pre-invasive and invasive lesions, and formulating corresponding clinical treatment plans are often based on past experience. Therefore, differences in regional practices, physician awareness, and diagnostic levels lead to divergent treatment opinions for the same lung nodule, often causing anxiety in patients (9,10). Lung nodule biopsy provides a definitive diagnosis, but thoracic surgeons often do not recommend preoperative biopsies for small nodules with significant ground-glass opacity and a small amount of solid component due to the small size of the nodules, the inability to accurately diagnose from small biopsy samples, and the low accuracy rate of the biopsy (11). Predicting the invasiveness of lung nodules through non-invasive methods remains a key area of research for the assessment of follow-up intervals and the formulation of treatment plans. The advent of AI in the field of lung nodule diagnosis and treatment has enabled the accurate quantification of the 3D characteristics of lung nodules.

This paper aimed to explore the predictive ability of the solid component volume and 3D CTR of lung nodules at different threshold settings (−550, −450, −350, −250, −150, −50, 0 HU) for the invasiveness of stage T1 lung adenocarcinoma. Univariate analysis of variance revealed that both the solid component volume and 3D CTR can distinguish between pre-invasive and invasive lesions, regardless of the CT threshold setting. Research has shown that different reconstruction kernels would have a certain impact on the density measurement of CT images (12). For example, when using different reconstruction kernels to quantitatively measure the lung density, it was found that the difference in HU values between different reconstruction kernels could reach 45 HU. This deviation was mainly caused by the balance among the bandwidth of the reconstruction kernel, noise amplification, and noise suppression of the iterative reconstruction algorithm. This may have a certain impact on the accuracy of CT measurement values. Therefore, in this paper, the CT difference grouping was set at 50 HU. These findings indicated that both are stable predictive factors for stage T1 lung adenocarcinoma, which may promote the development of more personalized and effective treatment strategies.

Recently, the Japan Clinical Oncology Group (JCOG) has sparked widespread discussion among peers with its multiple studies investigating the most suitable surgical approaches for early-stage non-small cell lung cancer (NSCLC)—such as JCOG0802 (13,14), JCOG0804 (15), and JCOG1211 (16). In these studies, tumor size and CTR played a significant role. Multiple studies (17-19) have shown that the CTR of lung nodules is an independent risk factor predicting the invasiveness of lung nodules; therefore, CTR has been used to guide thoracic surgery planning. However, CTR is based on manual measurements. These measurements often depend on the proportion of the diameter of the discernible solid component in lung window views to the entire length of the nodule. Such measurements have poor repeatability due to manual measurement errors, non-uniform of solid components, and different window width and level settings. However, the concept of solid component volume and 3D CTR based on the threshold segmentation method has been proposed in recent years. AI-based on deep learning technology can accurately segment CT images, distinguish between ground-glass opacity and solid components, and quantify them precisely. Tsuchida et al. (20) have revealed that its predictive efficacy regarding the invasiveness of early-stage lung adenocarcinoma surpasses the previously put-forward CTR. Sun et al. (6) reported the poor consistency of manual CTR measurements. Specifically, AI-quantified 3D values (3D CTR) (AUC =0.811) showed superior diagnostic performance compared to manual measurements (AUC =0.697), also slightly outperforming AI diameter ratio [one-dimensional (1D) CTR] (AUC =0.806) and area ratio [two-dimensional (2D) CTR] (AUC =0.796). Therefore, adopting 3D CTR in such research seemed to yield more precise conclusions.

However, no reference standards have been established for the threshold values used in measuring the solid components. Sun et al. proposed that the AI-based 1D CTR, 2D CTR, and 3D CTR can all be used to predict invasive and non-IAC. The team evaluated 278 lung adenocarcinoma patients with postoperative pathological confirmation. They found that 3D CTR exhibited superior discriminative performance compared to the first two methods, with the solid component threshold set at −450 HU. Gao et al. (21) employed AI automatic measurement values and analyzed 299 non-IACs and 889 IACs, revealing that their model had a good predictive performance (AUC =0.892), with the solid component threshold set at −300 HU. Gao et al. (22) analyzed 228 nodules and found that their AI model also had good predictive performance (AUC =0.849), which was superior to their morphological feature model, with the solid component threshold set at −350 HU. Moreover, Zhu et al.’s (23) research also showed that deep learning automatic measurement values were superior to manual measurements, with the optimal predictive threshold for prognosis determined at 0 HU. Different research teams employed various threshold values for the solid component measurements, indicating the need for standardization in this area.

This study investigated seven threshold values, and −350 HU was identified as optimal threshold for distinguishing between pre-invasive and invasive lesions. The best cutoff point for 3D CTR was 10.85%, while the optimal cutoff point for the solid component volume was 45.5 mm³. These optimal cutoff points could assist in clinical decision-making, particularly for heterogeneous ground-glass nodules, by indicating when a more aggressive approach to monitoring and intervention might be warranted. Furthermore, in patients with a CTR greater than 10.85% who exhibit heterogeneous ground-glass nodules, clinicians should be vigilant about the possibility of invasive lesions. They should also consider providing shorter follow-up intervals. If signs of progression such as enlargement, increased density, or enhancement are detected, active intervention should be considered. The study also showed that regardless of the threshold setting, the solid component volume demonstrated a higher predictive performance than CTR. However, whether this is attributable to the differences in measurement methods, the presence of non-uniform internal solid components, or uneven density within the lung nodule requires further exploration.

Study limitations

Nevertheless, the limitations of the study should be acknowledged. The retrospective design, differences in imaging acquisition conditions and slice thickness between the two centers, and exclusion of lung adenocarcinoma patients with diameters larger than 3 cm may introduce bias. Although using AI software can automatically parse the metadata in DICOM files and automatically identify and adjust the imaging differences of different devices to obtain quantitative parameters, biases may still arise when using multi-center data. In addition, sub-lobar resection is currently advocated for AIS and MIA; other classification systems have also been described, such as IAC patients being classified into the invasive group and AAH, AIS, and MIA patients into the non-invasive group. This grouping method may be more meaningful in the preoperative surgical plan for lung nodules, which will be discussed further later.

Conclusions

In conclusion, our study investigated the predictive ability of solid component volume and 3D CTR based on the threshold segmentation method for distinguishing between pre-invasive and invasive lesions of T1-stage lung adenocarcinoma. Solid component volume and 3D CTR represent stable predictive factors at seven different thresholds from −550 to 0 HU. This study identified −350 HU as the optimal threshold, with the corresponding best cut-off points of 10.85% for 3D CTR and 45.5 mm³ for solid component volume.

Acknowledgments

We would like to express our sincere appreciation and gratitude to those who helped to coordinate the study.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-24-1160/rc

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1160/coif). L.Z. is an employee of Shukun (Beijing) Technology Co., which provided data analysis and interpretation for this research. However, the company had no role in the study design, data collection, analysis, or manuscript preparation. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Boards of Zhengzhou People’s Hospital (No. 2024011139) and Huadong Hospital Affiliated to Fudan University (No. 2019K134). The requirement for individual consent was waived due to the retrospective nature of the study, and there is no infringement of patients’ privacy.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Travis WD, Brambilla E, Noguchi M, Nicholson AG, Geisinger KR, Yatabe Y, et al. International association for the study of lung cancer/american thoracic society/european respiratory society international multidisciplinary classification of lung adenocarcinoma. J Thorac Oncol 2011;6:244-85. [Crossref] [PubMed]
2. Zhang C, Pan Y, Li H, Zhang Y, Li B, Zhang Y, et al. Extent of surgical resection for radiologically subsolid T1N0 invasive lung adenocarcinoma: When is a wedge resection acceptable? J Thorac Cardiovasc Surg 2024;167:797-809.e2. [Crossref] [PubMed]
3. Jing W, Liu M, Li W, Li D, Wu Y, Lv F. Prognostic implication of consolidation-to-tumor ratio in early lung adenocarcinoma: a retrospective cross-sectional study. Quant Imaging Med Surg 2024;14:3366-80. [Crossref] [PubMed]
4. Zheng Y, Ju S, Huang R, Zhao J. Lymph node metastasis risk factors in clinical stage IA3 lung adenocarcinoma. J Cancer Res Ther 2023;19:34-8. [Crossref] [PubMed]
5. Wu Y, Song W, Wang D, Chang J, Wang Y, Tian J, Zhou S, Dong Y, Zhou J, Li J, Zhao Z, Che G. Prognostic value of consolidation-to-tumor ratio on computed tomography in NSCLC: a meta-analysis. World J Surg Oncol 2023;21:190. [Crossref] [PubMed]
6. Sun J, Zhang L, Hu B, Du Z, Cho WC, Witharana P, Sun H, Ma D, Ye M, Chen J, Wang X, Yang J, Zhu C, Shen J. Deep learning-based solid component measuring enabled interpretable prediction of tumor invasiveness for lung adenocarcinoma. Lung Cancer 2023;186:107392. [Crossref] [PubMed]
7. Chen ML, Liu YL, Zhu HB, Li XT, Qi LP, Sun YS. The differential diagnosis of lung precursor glandular lesions, micro-invasive adenocarcinoma, and invasive adenocarcinoma using low dose spectral computed tomography perfusion imaging. Quant Imaging Med Surg 2024;14:814-23. [Crossref] [PubMed]
8. Shi W, Hu Y, Sun Y, Chang G, Yang Y, Qian H, Wei Z, Zhao L, Li M, Zheng H, Li X. Exploring the optimal threshold of 3D consolidation tumor ratio value segmentation based on artificial intelligence for predicting the invasive degree of T1 lung adenocarcinoma. Quant Imaging Med Surg 2024;14:8988-98. [Crossref] [PubMed]
9. Lu L, Zhang B, Li W, Li J, Li L. Prevalence and Risk Factors of Psychological Distress in Patients With Early-Stage Lung Cancer During Preoperative Period: A Cross-Sectional Study. J Clin Nurs 2024; Epub ahead of print. [Crossref]
10. Hillyer GC, Milano N, Bulman WA. Pulmonary nodules and the psychological harm they can cause: A scoping review. Respir Med Res 2024;86:101121. [Crossref] [PubMed]
11. Wang B, Zhong F, An W, Liao M. The diagnostic value of CT-guided percutaneous puncture biopsy of pulmonary ground-glass nodules: a meta-analysis. Acta Radiol 2023;64:1431-8. [Crossref] [PubMed]
12. Rodriguez A, Ranallo FN, Judy PF, Fain SB. The effects of iterative reconstruction and kernel selection on quantitative computed tomography measures of lung density. Med Phys 2017;44:2267-80. [Crossref] [PubMed]
13. Hattori A, Suzuki K, Takamochi K, Wakabayashi M, Sekino Y, Tsutani Y, Nakajima R, Aokage K, Saji H, Tsuboi M, Okada M, Asamura H, Nakamura K, Fukuda H, Watanabe SIJapan Clinical Oncology Group. West Japan Oncology Group. Segmentectomy versus lobectomy in small-sized peripheral non-small-cell lung cancer with radiologically pure-solid appearance in Japan (JCOG0802/WJOG4607L): a post-hoc supplemental analysis of a multicentre, open-label, phase 3 trial. Lancet Respir Med 2024;12:105-16. [Crossref] [PubMed]
14. Saji H, Okada M, Tsuboi M, Nakajima R, Suzuki K, Aokage K, et al. Segmentectomy versus lobectomy in small-sized peripheral non-small-cell lung cancer (JCOG0802/WJOG4607L): a multicentre, open-label, phase 3, randomised, controlled, non-inferiority trial. Lancet 2022;399:1607-17. [Crossref] [PubMed]
15. Miyoshi T, Ito H, Wakabayashi M, Hashimoto T, Sekino Y, Suzuki K, et al. Risk factors for loss of pulmonary function after wedge resection for peripheral ground-glass opacity dominant lung cancer. Eur J Cardiothorac Surg 2023;64:ezad365. [Crossref] [PubMed]
16. Aokage K, Suzuki K, Saji H, Wakabayashi M, Kataoka T, Sekino Y, et al. Segmentectomy for ground-glass-dominant lung cancer with a tumour diameter of 3 cm or less including ground-glass opacity (JCOG1211): a multicentre, single-arm, confirmatory, phase 3 trial. Lancet Respir Med 2023;11:540-9. [Crossref] [PubMed]
17. Yang Y, Xu J, Wang W, Ma M, Huang Q, Zhou C, Zhao J, Duan Y, Luo J, Jiang J, Ye L. A nomogram based on the quantitative and qualitative features of CT imaging for the prediction of the invasiveness of ground glass nodules in lung adenocarcinoma. BMC Cancer 2024;24:438. [Crossref] [PubMed]
18. Zhang P, Li T, Tao X, Jin X, Zhao S. HRCT features between lepidic-predominant type and other pathological subtypes in early-stage invasive pulmonary adenocarcinoma appearing as a ground-glass nodule. BMC Cancer 2021;21:1124. [Crossref] [PubMed]
19. Su H, Dai C, Xie H, Ren Y, She Y, Kadeer X, Xie D, Zheng H, Jiang G, Chen C. Risk Factors of Recurrence in Patients With Clinical Stage IA Adenocarcinoma Presented as Ground-Glass Nodule. Clin Lung Cancer 2018;19:e609-17. [Crossref] [PubMed]
20. Tsuchida H, Tanahashi M, Suzuki E, Yoshii N, Watanabe T, Yobita S, Uchiyama S, Iguchi K, Nakamura M, Endo T. Impact of ground glass opacity in a three-dimensional analysis for pathological findings and prognosis in stage IA pure solid lung cancer. J Thorac Dis 2023;15:3829-39. [Crossref] [PubMed]
21. Gao J, Qi Q, Li H, Wang Z, Sun Z, Cheng S, Yu J, Zeng Y, Hong N, Wang D, Wang H, Yang F, Li X, Li Y. Artificial-intelligence-based computed tomography histogram analysis predicting tumor invasiveness of lung adenocarcinomas manifesting as radiological part-solid nodules. Front Oncol 2023;13:1096453. [Crossref] [PubMed]
22. Gao R, Gao Y, Zhang J, Zhu C, Zhang Y, Yan C. A nomogram for predicting invasiveness of lung adenocarcinoma manifesting as pure ground-glass nodules: incorporating subjective CT signs and histogram parameters based on artificial intelligence. J Cancer Res Clin Oncol 2023;149:15323-33. [Crossref] [PubMed]
23. Zhu Y, Chen LL, Luo YW, Zhang L, Ma HY, Yang HS, Liu BC, Li LJ, Zhang WB, Li XM, Xie CM, Yang JC, Wang DL, Li Q. Prognostic impact of deep learning-based quantification in clinical stage 0-I lung adenocarcinoma. Eur Radiol 2023;33:8542-53. [Crossref] [PubMed]