Prediction of non-small cell lung cancer subtypes is possible through restricted spectrum imaging

Background: To evaluate the utility of restricted spectrum imaging (RSI) for predicting subtypes of non-small cell lung cancer (NSCLC).

Methods: A total of 97 patients with NSCLC (30 with squamous cell carcinoma (SCC) and 67 with adenocarcinoma (AC)) were included. The parameters f₁, f₂, f₃, apparent diffusion coefficient (ADC), and maximum standardized uptake value (SUV_max) were measured and compared between the two subtypes. Logistic regression analysis was used to identify independent predictors, and a combined diagnostic model was developed. The performance of the model was assessed using receiver operating characteristic (ROC) curve analysis, calibration curves, and decision curve analysis (DCA).

Results: Compared with the AC group, the SCC group exhibited significantly higher SUV_max, f₂, and f₃ values, and lower ADC and f₁ values (all P < 0.05). Smoking status, f₁, SUV_max, and ADC were independent predictors of NSCLC subtypes. The combined model demonstrated superior diagnostic accuracy (AUC = 0.909; sensitivity = 73.33%; specificity = 89.55%) compared with individual predictors (AUC = 0.693, 0.819, 0.767, and 0.742 for smoking status, f₁, SUV_max, and ADC, respectively; all P < 0.01). Bootstrap resampling (1000 samples) validated the robustness of the model (AUC = 0.895). Calibration curves and DCA confirmed the model’s stability and clinical utility.

Conclusion: RSI can effectively differentiate NSCLC subtypes.

Introduction

Non-small cell lung cancer (NSCLC) is one of the most fatal malignancies globally (1). Squamous cell carcinoma (SCC) and adenocarcinoma (AC) are both prevalent histological subtypes of NSCLC, but their diagnosis and prognosis differ considerably (2). For example, in terms of surgical procedures, SCC may require more extensive airway reconstruction, whereas AC surgery is less invasive; in terms of drug therapy, patients with AC are more suitable for targeted therapy, while those with SCC are more suitable for immunotherapy. Additionally, SCC and AC differ in the assessment of recurrence risk and drug resistance (3–5). Therefore, accurate assessment of NSCLC subtypes before treatment is of great significance for the development of personalized treatment plans in clinical practice.

Although image-guided biopsy and bronchoscopy remain the gold standards for NSCLC subtype identification, they are invasive and pose risks such as bleeding and pneumothorax (6). Advances in quantitative imaging techniques have provided noninvasive alternatives for tumor characterization. Diffusion-weighted imaging (DWI), which assesses the diffusion of water molecules within tissues, and ¹¹⁸F-fluorodeoxyglucose positron emission tomography (¹⁸F-FDG PET), which evaluates tumor metabolism, are widely used for lung cancer assessment (7, 8). Restricted spectrum imaging (RSI) is an advanced diffusion MRI model that improves on conventional DWI by distinguishing water diffusion into restricted, hindered, and free compartments through a linear combination of diffusion-weighted models (9, 10). This enables RSI to quantitatively characterize the movement of water molecules in biological tissues with greater precision. RSI has been preliminarily applied in tumor evaluation. For instance, a study by Krishnan et al. showed that RSI helped improve the risk stratification of patients with glioblastoma (11); a study by Zhang et al. found that RSI-derived metrics could be used to noninvasively and effectively identify microvascular invasion in hepatocellular carcinoma (12); and a breast-related study conducted by He et al. concluded that RSI was able to quantitatively characterize breast lesions and normal fibroglandular tissue (13). However, in lung cancer research, to the best of our knowledge, only a few studies have explored the value of RSI in identifying benign and malignant lesions (14).

This study aims to evaluate the diagnostic value of RSI-derived quantitative parameters in differentiating SCC from AC, compare these parameters with classical ¹⁸F-FDG PET and DWI metrics, and combine them to develop a diagnostic tool. The ultimate goal is to provide a novel reference for the noninvasive assessment of NSCLC subtypes.

Materials and methods

Population

This research was approved by the local ethics review board, and all participants provided written informed consent. From June 2021 to October 2025, a total of 142 patients suspected of having lung cancer based on clinical evaluation or CT imaging underwent chest multiparametric scanning. The exclusion criteria were as follows: 1) Patients with claustrophobia or other conditions that prevented the completion of all imaging sequences (n = 8); 2) Scans with poor image quality (e.g., significant artifacts) that made them unsuitable for analysis (n = 14); 3) Cases where the interval between scanning and biopsy exceeded two weeks (n = 10); 4) Histological findings that did not indicate SCC or AC (n = 7); and 5) Patients who had received relevant treatment before scanning (n = 6). After applying these criteria, 97 patients were included in the study. Baseline characteristics such as age, sex, smoking status, and tumor size were recorded.

Scanning protocols

The MRI sequences (3.0 T system, uPMR790, United Imaging, Shanghai, China) included axial T2-weighted imaging (T2WI) and DWI with multiple b-values. The ¹⁸F-FDG used in this study was sourced from FracerLab FX-FDG (GE Minitrac), with a purity > 95% and a pH range of 4.5–8.5. Patients fasted for at least 6 hours to ensure their serum glucose levels remained ≤ 6.5 mmol/L before receiving an ¹⁸F-FDG injection (0.11 mCi/kg). The PET scan began 60 minutes after injection, covering the upper thoracic inlet to the lower lung margin with the patient in the supine position (15, 16). PET image reconstruction was performed using the ordered subset expectation maximization (OSEM) method (2 iterations, 20 subsets, voxel size 2.6 × 2.6 × 2.0 mm³). A detailed summary of the protocol specifications is provided in Table 1.

Table 1

Table 1. Scanning protocol.

Parameter generation

The acquired images were transferred to a post-processing workstation (uWS-MR005, United Imaging, Shanghai, China) for registration, motion correction, and in-depth analysis. DWI and RSI data were processed using diffusion analysis software from the Advanced Analysis Toolkit. The DWI parametric pseudo-color map was generated using Equation 1:

$\begin{array}{ll}{\text{S}}_{\text{b}}/{\text{S}}_0=\exp (-\text{ b}\times \mathrm{ADC})& (1)\end{array}$

where ADC represents the apparent diffusion coefficient, b is the diffusion sensitizing factor, and S₀ and S_b denote the signal intensities (SIs) at b = 0 s/mm² and b = [specified value] s/mm², respectively (8). The RSI parametric pseudo-color map was constructed using Equation 2:

$\begin{array}{ll}S(b)=f_1{e}^{-bD1}+f_2{e}^{-bD2}+f_3{e}^{-bD3},D1<D2<D3& (2)\end{array}$

where f₁, f₂, and f₃ represent the volume fractions of the restricted, hindered, and free water diffusion compartments, respectively, and D1, D2, and D3 denote the ADCs of these compartments. To prevent overfitting, ensure model linearization, and maintain comparability across compartments, D1, D2, and D3 were globally assigned values of 1.0 × 10^-3 mm²/s, 2.0 × 10^-3 mm²/s, and 3.0 × 10^-3 mm²/s, respectively, based on established theoretical values and experimental data (17, 18).

Tumor margins on axial T2-weighted images were manually delineated slice by slice to define regions of interest (ROIs). Lesions with cystic degeneration, necrosis, hemorrhagic artifacts, or blood vessels were excluded. The finalized ROIs were then mapped onto pseudo-color DWI and RSI parameter maps, and the mean values were extracted. The volume of interest (VOI) was automatically delineated, and the maximum standardized uptake value (SUV_max) was calculated using PET fusion software. Two radiologists performed these procedures independently: an attending radiologist with 8 years of experience and an associate chief radiologist with 15 years of experience. Both were blinded to each other’s results and the patients’ clinicopathological details.

Histopathologic assessment

Tumor specimens were obtained through surgical resection or biopsy. Histological subtype classification was performed in accordance with the guidelines of the International Association for the Study of Lung Cancer (IASLC) (19).

Data analysis

We employed R (version 3.5.3, R Foundation, Auckland, New Zealand) and SPSS (version 27.0, MedCalc Software, Ostend, Belgium) to conduct data analysis. To assess the interobserver agreement for the parameters, we utilized the interclass correlation coefficient (ICC). An ICC > 0.75 was considered indicative of satisfactory reliability (20). Based on the characteristics of the variables, different statistical tests were applied to compare data between the SCC and AC groups. These tests included the Mann–Whitney U test, independent samples t - test and chi - square test.

R (version 3.5.3, R Foundation, Auckland, New Zealand) and SPSS (version 27.0, IBM Corp., Armonk, NY, USA) were used for data analysis. The interclass correlation coefficient (ICC) was employed to assess the interobserver agreement for the parameters. An ICC > 0.75 was considered indicative of satisfactory reliability (20). Based on the characteristics of the variables, different statistical tests were applied to compare data between the SCC and AC groups, including the Mann–Whitney U test, independent samples t-test, and chi-square test. The diagnostic performance of the parameters was assessed using the area under the receiver operating characteristic curve (AUC). The DeLong test was used to compare differences in AUC values. Logistic regression (LR) analysis was performed to identify independent predictors and develop a multiparameter composite diagnostic tool. Bootstrap resampling (1000 samples), calibration curves, and decision curve analysis (DCA) were used for internal validation and evaluation of the diagnostic tool. Statistical significance was set at P < 0.05.

Results

Baseline characteristics

A total of 30 patients with SCC and 67 patients with AC were included. Significant differences were observed between the two groups in maximum lesion diameter (P < 0.001), smoking status (P < 0.001), and sex distribution (P = 0.022). However, there was no significant difference in age between the two groups (P = 0.614). The clinical characteristics are summarized in Figure 1, Table 2.

Figure 1

Figure 1. A male patient with adenocarcinoma of the upper lobe of the right lung (arrowheads, maximum diameter = 5.5cm, no smoking). (a) Map of T2-weighted imaging; (b) Map of DWI (b = 600 s/mm²); (c) Map of ¹⁸F-FDG PET; (d) Pseudo colored map of ADC; (e) Pseudo colored map of f₁; (f) Pseudo colored map of f₂; (g) Pseudo colored map of f₃; (h) Pathological images (original magnification, ×100).

Table 2

Table 2. Comparison of various variables between SCC and AC groups.

ICC test

Measurements of f₁, f₂, f₃, ADC, and SUV_max showed excellent interobserver agreement, with all ICC values > 0.80. Therefore, the average values from both readers were used for subsequent analysis.

Parameter comparison

The SCC group exhibited significantly higher SUV_max, f₂ and f₃ values compared to the AC group, while ADC and f₁ values were lower (all P < 0.05, Figure 2, Table 2).

Figure 2

Figure 2. Comparison of (a) maximum lesion diameter, (b) ADC, (c) SUV_max, (d) f₁, (e) f₂ and (g) f₃ between squamous cell carcinoma (SCC) and adenocarcinoma (AC) groups.

LR analysis

Univariate analysis identified sex distribution, smoking status, maximum lesion diameter, f₁, f₂, f₃, ADC, and SUV_max as significant predictors for differentiating SCC from AC (all P < 0.05). Multivariate analysis identified smoking status, f₁, ADC, and SUV_max as independent predictors of differentiation, with corresponding P-values of 0.024, 0.001, 0.033, and 0.018, respectively (Table 3).

Table 3

Table 3. Univariate and multivariate analyses.

Diagnostic performance

The composite model of independent predictors achieved the best diagnostic performance, with an AUC of 0.909, a sensitivity of 73.33%, and a specificity of 89.55%. This performance was significantly superior to that of individual parameters (sex, smoking status, maximum lesion diameter, f₁, f₂, f₃, ADC, and SUV_max; AUCs = 0.715, 0.693, 0.715, 0.819, 0.769, 0.780, 0.742, and 0.767; Z = 4.175, 4.252, 3.525, 2.252, 2.994, 2.758, 3.194, and 2.960; P < 0.001, < 0.001, < 0.001, = 0.024, = 0.003, = 0.006, = 0.001, and = 0.003, respectively) (Figure 3, Table 4).

Figure 3

Figure 3. The areas under receiver-operator characteristic (ROC) curves: (a) includes lines for f₁, f₂, f₃, ADC, and Combined Diagnosis (smoking + f₁ + SUV_max + ADC); (b) includes lines for SUV_max, Maximum Diameter, Female, Smoking, and Combined Diagnosis (smoking + f₁ + SUV_max + ADC).

Table 4

Table 4. Predictive performance of different variables.

Validation

Internal validation using bootstrap resampling confirmed the robustness of the composite model, yielding an AUC of 0.895 (95% CI: 0.875–0.906). The calibration curve and DCA plots demonstrated good calibration and clinical utility of the model for patients with NSCLC (Figure 4).

Figure 4

Figure 4. Calibration curve (a) and decision curve analysis (b) of the combination of independent predictors (smoking + f₁ + SUV_max + ADC).

Discussion

Currently, ¹⁸F-FDG PET and DWI are widely used noninvasive imaging modalities for assessing glucose metabolism and water molecule diffusion, respectively. SUV_max, derived from ¹⁸F-FDG PET, reflects the peak glucose metabolism of tumors, while ADC, derived from DWI, quantifies the diffusion rate of water within tissues (21). Previous studies have reported that SCC has greater proliferative and invasive potential than AC, leading to distinct ¹⁸F-FDG metabolism levels and water diffusion rates. This makes SUV_max and ADC effective for distinguishing between the two subtypes (22, 23). In this study, SUV_max was significantly higher and ADC was lower in the SCC group than in the AC group, which is consistent with previous findings. Moreover, multivariate analysis confirmed that SUV_max and ADC are independent predictors, further supporting the utility of ¹⁸F-FDG PET and DWI in the assessment of NSCLC subtypes.

Building on the aforementioned research, this study innovatively introduced RSI into the differentiation of SCC from AC. RSI is an advanced diffusion imaging technique that assesses the movement of water molecules within human tissues (24). Unlike DWI, RSI does not assume a Gaussian distribution of water diffusion. Instead, it distinguishes and quantifies diffusion across multiple microstructural compartments (restricted, hindered, and free diffusion), allowing for a more precise evaluation of water movement (25). However, clinical studies on RSI have yielded inconsistent findings. For example, in breast cancer, both f₁ and f₃ can differentiate benign from malignant lesions, with malignant lesions exhibiting increased f₁ and decreased f₃ values (26). In contrast, in rectal cancer, only f₁ can effectively distinguish high-grade from low-grade tumors, with higher f₁ values observed in high-grade cases (27). In the present study, compared with the AC group, the SCC group exhibited lower f₁ values and higher f₂ and f₃ values. Among these parameters, f₁ was identified as an independent predictor for distinguishing SCC from AC. The parameters f₁, f₂, and f₃ represent the proportions of restricted, hindered, and free diffusion, respectively, and their sum equals 1 (10). A possible explanation for the observed differences lies in the biological characteristics of the two tumor types. Although SCC is typically more invasive and has a higher cell density than AC (which would suggest an increase in the restricted diffusion compartment) (28, 29), the associated tissue microischemia and micronecrosis may expand the extracellular space and increase the free water content. This shift promotes greater hindered and free water diffusion. When the increase in restricted diffusion is outweighed by the increase in hindered and free diffusion, the f₁ fraction decreases, leading to elevated f₂ and f₃ values (30). Additionally, differences in cell arrangement between SCC and AC may also contribute to this outcome.

Clinical factors, including age, sex distribution, smoking status, and maximum lesion diameter, were incorporated into the analysis. The results indicated that while sex distribution, smoking status, and maximum lesion diameter contributed to the differentiation of SCC from AC, only smoking status emerged as an independent predictor. This finding aligns with previous studies, reinforcing the role of smoking status as a simple and effective indicator for NSCLC subtyping (31). The underlying mechanism may be attributed to smoking-induced bronchial squamous epithelial carcinogenesis (32).

Despite these promising results, several limitations must be acknowledged. First, this was a single-center study with a relatively small sample size and an uneven distribution of tumor subtypes (30 cases of SCC vs. 67 cases of AC), which may affect the stability and generalizability of the predictive model. Second, research on RSI sequences—particularly in lung imaging—is still limited, and the optimal b-value for lesion evaluation has not yet been established. Third, MRI has limitations in detecting microscopic lesions. Fourth, despite the use of various techniques to mitigate respiratory and cardiovascular pulsation artifacts, their impact remains significant. Fifth, previous studies have suggested an association between tumour location and subtypes of NSCLC. Among the clinical factors considered in this paper, lesion location was not included as a potential influencing factor, which could have an adverse impact on the research results. Future research will focus on expanding sample sizes, conducting multicenter studies, reducing distribution disparities among different lesion subtypes, and incorporating additional clinical factors such as lesion location. Additionally, efforts will be made to optimize scanning protocols and improve image quality to ensure more stable and reliable results.

Conclusion

Smoking status, f₁, SUV_max, and ADC are independent predictors for the differentiation of AC from SCC. The combination of these parameters shows potential as a biomarker for the classification of NSCLC subtypes.

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement

The studies involving humans were approved by Henan Provincial People’s Hospital. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

Author contributions

ZH: Software, Writing – original draft. YZ: Writing – original draft, Methodology. LS: Writing – original draft, Methodology. BD: Software, Writing – review & editing. YY: Writing – review & editing, Investigation. ZW: Writing – review & editing, Software, Validation. XY: Writing – review & editing, Methodology. NM: Writing – review & editing, Formal analysis. FF: Writing – review & editing.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This work was supported by the Natural Science Foundation of Henan Province (252300420565), the Science and Technology Project of Henan Province (252102311105), and the Key Project of Henan Province Medical Science and Technology Project (LHGJ20240036, LHGJ20240053).

Acknowledgments

We acknowledge the support received from Natural Science Foundation of Henan.

Conflict of interest

Authors YY and ZW were employed by the company United Imaging Healthcare.

The remaining author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

1. Siegel RL, Giaquinto AN, and Jemal A. Cancer statistics, 2024. CA Cancer J Clin. (2024) 74:12–49. doi: 10.3322/caac.21820

PubMed Abstract | Crossref Full Text | Google Scholar

2. Hendriks LEL, Remon J, Faivre-Finn C, Garassino MC, Heymach JV, Kerr KM, et al. Non-small-cell lung cancer. Nat Rev Dis Primers. (2024) 10:71. doi: 10.1038/s41572-024-00551-9

PubMed Abstract | Crossref Full Text | Google Scholar

3. Riely GJ, Wood DE, Ettinger DS, Aisner DL, Akerley W, Bauman JR, et al. Non-small cell lung cancer, version 4.2024, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw. (2024) 22:249–74. doi: 10.6004/jnccn.2204.0023

PubMed Abstract | Crossref Full Text | Google Scholar

4. Liu L, Soler J, Reckamp KL, and Sankar K. Emerging targets in non-small cell lung cancer. Int J Mol Sci. (2024) 25:10046. doi: 10.3390/ijms251810046

PubMed Abstract | Crossref Full Text | Google Scholar

5. Miao D, Zhao J, Han Y, Zhou J, Li X, Zhang T, et al. Management of locally advanced non-small cell lung cancer: State of the art and future directions. Cancer Commun (Lond). (2024) 44:23–46. doi: 10.1002/cac2.12505

PubMed Abstract | Crossref Full Text | Google Scholar

6. Stamatis G. Staging of lung cancer: the role of noninvasive, minimally invasive and invasive techniques. Eur Respir J. (2015) 46:521–31. doi: 10.1183/09031936.00126714

PubMed Abstract | Crossref Full Text | Google Scholar

7. Hinzpeter R, Kulanthaivelu R, Kohan A, Murad V, Mirshahvalad SA, Avery L, et al. Predictive [18F]-FDG PET/CT-based radiogenomics modelling of driver gene mutations in non-small cell lung cancer. Acad Radiol. (2024) 31:5314–23. doi: 10.1016/j.acra.2024.06.038

PubMed Abstract | Crossref Full Text | Google Scholar

8. Ohno Y, Yui M, Takenaka D, Yoshikawa T, Koyama H, Kassai Y, et al. Computed DWI MRI results in superior capability for N-stage assessment of non-small cell lung cancer than that of actual DWI, STIR imaging, and FDG-PET/CT. J Magn Reson Imaging. (2023) 57:259–72. doi: 10.1002/jmri.28288

PubMed Abstract | Crossref Full Text | Google Scholar

9. White NS, Leergaard TB, D’Arceuil H, Bjaalie JG, and Dale AM. Probing tissue microstructure with restriction spectrum imaging: Histological and theoretical validation. Hum Brain Mapp. (2013) 34:327–46. doi: 10.1002/hbm.21454

PubMed Abstract | Crossref Full Text | Google Scholar

10. White NS, McDonald C, Farid N, Kuperman J, Karow D, Schenker-Ahmed NM, et al. Diffusion-weighted imaging in cancer: physical foundations and applications of restriction spectrum imaging. Cancer Res. (2014) 74:4638–52. doi: 10.1158/0008-5472.CAN-13-3534

PubMed Abstract | Crossref Full Text | Google Scholar

11. Krishnan AP, Karunamuni R, Leyden KM, Seibert TM, Delfanti RL, Kuperman JM, et al. Restriction spectrum imaging improves risk stratification in patients with glioblastoma. AJNR Am J Neuroradiol. (2017) 38:882–9. doi: 10.3174/ajnr.A5099

PubMed Abstract | Crossref Full Text | Google Scholar

12. Zhang Y, Yang C, Sheng R, Dai Y, and Zeng M. Preoperatively identify the microvascular invasion of hepatocellular carcinoma with the restricted spectrum imaging. Acad Radiol. (2023) 30:S30–9. doi: 10.1016/j.acra.2023.06.010

PubMed Abstract | Crossref Full Text | Google Scholar

13. He L, Qin Y, Hu Q, Liu Z, Zhang Y, and Ai T. Quantitative characterization of breast lesions and normal fibroglandular tissue using compartmentalized diffusion-weighted model: comparison of intravoxel incoherent motion and restriction spectrum imaging. Breast Cancer Res. (2024) 26:71. doi: 10.1186/s13058-024-01828-3

PubMed Abstract | Crossref Full Text | Google Scholar

14. Liu X, Meng N, Zhou Y, Fu F, Yuan J, Wang Z, et al. Tri-compartmental restriction spectrum imaging based on ¹⁸F-FDG PET/MR for identification of primary benign and Malignant lung lesions. J Magn Reson Imaging. (2025) 61:830–40. doi: 10.1002/jmri.29438

PubMed Abstract | Crossref Full Text | Google Scholar

15. Chen S, Gu Y, Yu H, Chen X, Cao T, Hu L, et al. NEMA NU2-2012 performance measurements of the United Imaging uPMR790: an integrated PET/MR system. Eur J Nucl Med Mol Imaging. (2021) 48:1726–35. doi: 10.1007/s00259-020-05135-9

PubMed Abstract | Crossref Full Text | Google Scholar

16. Liu G, Cao T, Hu L, Zheng J, Pang L, Hu P, et al. Validation of MR-based attenuation correction of a newly released whole-body simultaneous PET/MR system. BioMed Res Int. (2019) 2019:8213215. doi: 10.1155/2019/8213215

PubMed Abstract | Crossref Full Text | Google Scholar

17. Conlin CC, Feng CH, Rodriguez-Soto AE, Karunamuni RA, Kuperman JM, et al. Improved characterization of diffusion in normal and cancerous prostate tissue through optimization of multicompartmental signal models. J Magn Reson Imaging. (2021) 53:628–39. doi: 10.1002/jmri.27393

PubMed Abstract | Crossref Full Text | Google Scholar

18. Loubrie S, Zou J, Rodriguez-Soto AE, Lim J, Andreassen MMS, Cheng Y, et al. Discrimination between benign and Malignant lesions with restriction spectrum imaging MRI in an enriched breast cancer screening cohort. J Magn Reson Imaging. (2025) 61:1876–87. doi: 10.1002/jmri.29599

PubMed Abstract | Crossref Full Text | Google Scholar

19. Rami-Porta R, Nishimura KK, Giroux DJ, Detterbeck F, Cardillo G, Edwards JG, et al. The international association for the study of lung cancer lung cancer staging project: proposals for revision of the TNM stage groups in the forthcoming (Ninth) edition of the TNM classification for lung cancer. J Thorac Oncol. (2024) 19:1007–27. doi: 10.1016/j.jtho.2024.02.011

PubMed Abstract | Crossref Full Text | Google Scholar

20. Shieh G. Choosing the best index for the average score intraclass correlation coefficient. Behav Res Methods. (2016) 48:994–1003. doi: 10.3758/s13428-015-0623-y

PubMed Abstract | Crossref Full Text | Google Scholar

21. Deininger K, Korf P, Lauber L, Grimm R, Strecker R, Steinacker J, et al. From phantoms to patients: improved fusion and voxel-wise analysis of diffusion-weighted imaging and FDG-positron emission tomography in positron emission tomography/magnetic resonance imaging for combined metabolic-diffusivity index (cDMI). Diagn (Basel). (2024) 14:1787. doi: 10.3390/diagnostics14161787

PubMed Abstract | Crossref Full Text | Google Scholar

22. Huang Z, Wang H, Ting F, Chen Y, Fan H, Li X, et al. Metabolic and multi-model intravoxel incoherent motion parameters based 18F-FDG PET/MRI for predicting subtypes of inoperable non-small cell lung cancer. BMC Cancer. (2025) 25:322. doi: 10.1186/s12885-025-13543-z

PubMed Abstract | Crossref Full Text | Google Scholar

23. Hu WD, Wang HC, Wang YB, Cui LL, and Chen XH. Correlation study on 18F-FDG PET/CT metabolic characteristics of primary lesion with clinical stage in lung cancer. Q J Nucl Med Mol Imaging. (2021) 65:172–7. doi: 10.23736/S1824-4785.19.03146-7

PubMed Abstract | Crossref Full Text | Google Scholar

24. Yin H, Liu W, Xue Q, Song C, Ren J, Li Z, et al. The value of restriction spectrum imaging in predicting lymph node metastases in rectal cancer: a comparative study with diffusion-weighted imaging and diffusion kurtosis imaging. Insights Imaging. (2024) 15:302. doi: 10.1186/s13244-024-01852-z

PubMed Abstract | Crossref Full Text | Google Scholar

25. Besasie BD, Sunnapwar AG, Gao F, Troyer D, Clarke GD, White H, et al. Restriction spectrum imaging-magnetic resonance imaging to improve prostate cancer imaging in men on active surveillance. J Urol. (2021) 206:44–51. doi: 10.1097/JU.0000000000001692

PubMed Abstract | Crossref Full Text | Google Scholar

26. Qin Y, Tang C, Hu Q, Zhang Y, Yi J, Dai Y, et al. Quantitative assessment of restriction spectrum MR imaging for the diagnosis of breast cancer and association with prognostic factors. J Magn Reson Imaging. (2023) 57:1832–41. doi: 10.1002/jmri.28468

PubMed Abstract | Crossref Full Text | Google Scholar

27. Xiong Z, Geng Z, Lian S, Yin S, Xu G, Zhang Y, et al. Discriminating rectal cancer grades using restriction spectrum imaging. Abdom Radiol (NY). (2022) 47:2014–22. doi: 10.1007/s00261-022-03500-w

PubMed Abstract | Crossref Full Text | Google Scholar

28. Li G, Huang R, Zhu M, Du M, Zhu J, Sun Z, et al. Native T1-mapping and diffusion-weighted imaging (DWI) can be used to identify lung cancer pathological types and their correlation with Ki-67 expression. J Thorac Dis. (2022) 14:443–54. doi: 10.21037/jtd-22-77

PubMed Abstract | Crossref Full Text | Google Scholar

29. Li S, Liu J, Zhang W, Lu H, Wang W, Lin L, et al. T1 mapping and multimodel diffusion-weighted imaging in the assessment of cervical cancer: a preliminary study. Br J Radiol. (2023) 96:20220952. doi: 10.1259/bjr.20220952

PubMed Abstract | Crossref Full Text | Google Scholar

30. Meng N, Liu X, Zhou Y, Yu X, Wu Y, Fu F, et al. Multiparametric 18F-FDG PET/MRI based on restrictive spectrum imaging and amide proton transfer-weighted imaging facilitates the assessment of lymph node metastases in non-small cell lung cancer. Radiol Med. (2025) 130:1003–12. doi: 10.1007/s11547-025-01992-2

PubMed Abstract | Crossref Full Text | Google Scholar

31. Sabbula BR, Gasalberti DP, Mukkamalla SKR, and Anjum F. Squamous cell lung cancer. In: StatPearls. StatPearls Publishing, Treasure Island (FL (2024).

PubMed Abstract | Google Scholar

32. Liu B, Liu Y, Zou J, Zou M, and Cheng Z. Smoking is Associated with Lung Adenocarcinoma and Lung Squamous Cell Carcinoma Progression through Inducing Distinguishing lncRNA Alterations in Different Genders. Anticancer Agents Med Chem. (2022) 22:1541–50. doi: 10.2174/1871520621666210727115147

PubMed Abstract | Crossref Full Text | Google Scholar