Introduction
Lung cancer is the second most common type of cancer worldwide, causing 1.8 million deaths annually [1]. The most frequently occurring cell type is a non-small cell lung cancer (NSCLC) [2]. Early detection through the use of low-dose CT scans and accurate lung cancer staging are important tools for appropriate treatment planning and improving patient survival [3,4]. Mediastinal staging is a crucial factor in designing treatment plans for NSCLC, and contralateral mediastinal lymph node involvement (N3) is a criterion for identification of an inoperable case [5]. Positron emission tomography integrated computed tomography (PET/CT) is a useful non-invasive mediastinal staging modality in NSCLC but it has a high false positive rate, which varies from 6.5% to 47.5% [4,6,7], especially in the regions with a high prevalence of inflammatory or infectious lung diseases such as pulmonary tuberculosis [3,4]. Additionally, PET/CT is not widely available in some countries, including Thailand. Due to the high false-positive rate of PET/CT, patients may undergo invasive procedures such as endobronchial ultrasound-guided fine needle aspiration (EBUS) and mediastinoscopy which carry a risk of complications more frequently [8]. Furthermore, transbronchial biopsy may not be able to reach all areas of mediastinal lymph nodes, as recommended by the International Association for the Study of Lung Cancer (IASLC) lymph node stations [9], such as prevascular, subaortic, paraaortic, paraesophageal, pulmonary ligament, and some of left paratracheal stations [10].
Recently, texture analysis, a key component of Radiomics, has been developed to assess and quantify tumor heterogeneity which cannot be detected by the human eye. Texture analysis can be applied to a variety of imaging modalities, including CT, Magnetic resonance imaging (MRI), and PET/CT. The first-order CT texture analysis (CTTA) involves creating a pixel intensity histogram in the area of interest, which provides quantitative measures of tissue heterogeneity. This approach is easily accessible and does not require expensive post-processing software. [11]. The concept behind CTTA is that the more heterogeneous a tumor is, the more biologically aggressive its behavior and the more resistance it will have to treatment [11].
The second-order and higher-order CTTA are more advanced techniques that provide more delicate details of the tumor, but they require specific software. A few studies have shown that the CTTA of mediastinal lymph nodes can differentiate between malignant and benign lymph nodes based on the hypothesis that metastatic lymph nodes are heterogeneous and of high density due to the loss of normal hilar fat. [10,12]. However, this is based on advanced CTTA techniques such as Run-length non-uniformity (RLNU), Gray-level non-uniformity (GLNU), and advanced metrics, while data about CT histogram is limited [7,13,14].
The CT histogram of mediastinal lymph nodes could be a significant method for mediastinal lymph nodal staging in NSCLC; however, there is limited data available, particularly in Thailand. This insufficiency motivated our search for an objective of this study, which was to evaluate the diagnostic value of the CT features and histogram analysis in differentiating between malignant and benign mediastinal lymph nodes in patients with NSCLC.
Materials and methods
Patient population
This study is a retrospective review of non-contrast chest CT scans of patients diagnosed with NSCLC. The population was searched using the keyword ‘primary lung cancer’ in the Envision program from January 2019 to June 2021, revealing 328 cases of primary lung cancer. Only 74 patients met inclusion criteria, which included the availability of pretreatment CT images in the CMU-PACS system, pathological confirmation of the lymph nodes through fine needle aspiration (FNA) or histology within an interval of less than three months and presence of ≥ 5 mm lymph nodes. Exclusion criteria included the absence of plain CT scans (30 patients) and poor image quality (4 patients), leaving 40 patients with 80 lymph nodes in this study (Fig 1). The study was approved by the institutional review board (RAD-2565-09120) and waived the requirement for informed consent due to its retrospective nature. The review of medical records was performed in accordance with institutional ethics review board guidelines.
[Figure omitted. See PDF.]
NSCLC; non-small cell lung cancer, CMU-PACS; Chiang Mai University Picture archiving and communication system.
Data collection
The demographic data of patients, including gender, age, and pathologic reports, was collected from medical records. All protected health information (PHI) was securely stored in a password-protected computer in the Radiology Department. No traceable data that could identify the patients was collected.
The non-contrast CT images were obtained from machines from three manufacturers, including SOMATOM Definition (Siemens Healthcare, Forchheim, Germany), SOMATOM Force (Siemens Healthcare, Forchheim, Germany), and Philips IQon (Philips Healthcare, Best, The Netherlands). CT studies were obtained on multidetector CT systems with a breath-held helical acquisition of the entire thorax, 120 kV and modified mAs. A collimation of 128 × 0.6 mm with z-flying focal spot technique was used. The images were reconstructed with 1-mm slice thickness and the field of view (FOV) that covered the entire thorax. Axial images at interested mediastinal lymph nodes were used for evaluation.
During surgery, the surgeons resected lymph nodes with some adipose connective tissue from the corresponding anatomic regions. The mediastinal lymph nodes were identified on the CT scan and correlated with the pathological reports by using size, location, and endobronchial ultrasound (EBUS) findings of the lymph nodes. The pathology of all the selected lymph nodes was verified by surgical pathology or fine needle aspiration from EBUS. The preoperative CT images of mediastinal lymph nodes were reviewed independently by two board-certified diagnostic radiologists; Y.W. and P.P. who had eleven years and five years of experience in chest imaging interpretation, respectively. Both were blinded to the pathological reports and clinical data. They focused on the following characteristics: size (short axis diameter), shape, margins, and internal composition, including the presence of calcification, necrosis, or fat components. Necrosis was defined as a focal area of low attenuation, appearing less dense than the surrounding tissues on CT scans [15].
The data from the CT scanner was sent to the workstation for CT histogram and the three-dimensional software automatically generated histograms of each node. The histogram features were computed by segmenting the whole volume of each lymph node using a three-dimensional volume of interest (VOI) approach. This method ensured comprehensive volumetric analysis, capturing spatial heterogeneity throughout the lymph node. The images were viewed using a standard mediastinal window setting with a level of 30 Hounsfield Units (HU) and a width of 400 HU (Fig 2).
[Figure omitted. See PDF.]
A board-certified diagnostic radiologist (P.P.) segmented the mediastinal lymph nodes using a semi-automated method and analyzed them with six histogram measurements: mean attenuation, mean positive pixel (MPP), standard deviation (SD), skewness, kurtosis, and entropy using the Synapse Vincent system (Fujifilm Corp., Tokyo, Japan) (Fig 3). The histogram parameters used in this study are defined as follows [11]: Mean attenuation refers to the average density of the region of interest in Hounsfield units (HU); Mean positive pixel indicates the average density of the region of interest where the attenuation is greater than 0 HU; Skewness measures the asymmetry of the histogram, indicating whether the data distribution is skewed to the left or right; Kurtosis describes the peakedness or flatness of the histogram, reflecting the distribution of pixel intensities; and Entropy quantifies the irregularity or complexity of the histogram, with higher entropy values indicating greater heterogeneity.
[Figure omitted. See PDF.]
Method of data analysis
Descriptive analysis involved the calculation of frequencies and proportions for categorical variables. The interobserver agreement was reported using the Intraclass Correlation Coefficient (ICC) for continuous quantitative variables, and the kappa test was used for categorical variables. The ICC values below 0.5 suggest poor reliability, while values ranging from 0.5 to 0.75 indicate moderate reliability. Values between 0.75 and 0.9 indicate good reliability, and those exceeding 0.90 indicate excellent reliability [16]. The kappa values were evaluated based on the Viera classification, with the following interpretations: 0.01–0.20 indicating slight agreement, 0.21–0.40 indicating fair agreement, 0.41–0.60 indicating moderate agreement, 0.61–0.80 indicating substantial agreement, and 0.81–0.99 indicating almost perfect agreement [17]. The means with standard deviations (SDs) or medians with interquartile ranges (IQRs) of continuous variables were determined based on the distribution of the data. The collected data were evaluated using descriptive statistics, including absolute and relative frequencies. Group differences were analyzed using the Mann-Whitney test for continuous data and the Fisher’s exact test for categorical data. Receiver-operating characteristic (ROC) curve analysis was used to assess diagnostic accuracy. Statistical significance was determined at a p-value of less than 0.05. The analyses were performed using IBM SPSS software version 25.
Results
General data
This study included 40 patients and 80 lymph nodes. The mean age of the patients was 66.9 ± 7.89 years, with a slightly higher proportion of male patients compared to female patients. Out of the 80 mediastinal lymph nodes, 46 were benign and 34 were metastatic nodes, as proven by pathology. Of the metastatic lymph nodes, 29 (36.3%) were adenocarcinoma and 5 (6.3%) were squamous cell carcinoma. The demographic data are demonstrated in Table 1.
[Figure omitted. See PDF.]
CT findings and histogram
The level of agreement between the two radiologists demonstrated excellent correlation in quantitative nodal size measurements (Intraclass Correlation Coefficient (ICC) = 0.94, p-value = 0.001). However, the agreement in qualitative nodal morphology evaluation showed fair to moderate agreement between the two radiologists (Kappa ranging from 0.412 to 0.756; p-value < 0.05) (Table 2).
[Figure omitted. See PDF.]
Based on the analysis of the median values of nodal size and volume, it was found that the malignant lymph nodes were larger than the benign nodes. The benign lymph nodes had a median short axis of 8.0 mm (ranging from 6 to 10 mm), while the malignant lymph nodes had a median short axis of 14.0 mm (ranging from 12 to 20 mm). The difference between the two groups was statistically significant (p-value < 0.001). The median volume of the malignant group was 2.49 ml, which was significantly different from that of the benign group, which was 0.87 ml (p-value < 0.001).
When considering the morphologic appearances of the lymph nodes from the benign and malignant groups, analysis showed that the majority of the malignant nodes had an irregular shape (76.5%), ill-defined margins (73.5%), and contained necrotic components (50.0%), as compared to the benign nodes, with differences being statistically significant (p-value < 0.05). It was observed that the malignant lymph nodes did not have any internal fat component. However, the benign lymph nodes generally had an oval shape (65.2%), well-defined margins (93.5%), and a presence of internal hilar fat (17.4%). Table 3 shows the comparison between the CT findings from the malignant and benign mediastinal lymph nodes in patients with non-small cell lung cancer.
[Figure omitted. See PDF.]
A comparison of CT histogram parameters between benign and malignant nodal groups revealed significant differences in the mean attenuation, skewness, kurtosis, and entropy, with p-values of 0.004, 0.041, 0.005, and <0.001, respectively. However, there were no significant differences in the Maximum Pixel Percentage (MPP) and Standard Deviation (SD), with p-values of 1.758, and 0.083, respectively. Table 3 demonstrates the CT findings and histogram parameters, and the differences in histogram features between benign and malignant lymph nodes are illustrated using box plots (Fig 4). Additionally, the receiver-operating characteristics (ROC) curve analysis, incorporating all CT histogram parameters, yielded an area under the curve (AUC) of 0.870 for differentiating between benign and malignant lymph nodes (Fig 5).
[Figure omitted. See PDF.]
[Figure omitted. See PDF.]
The ROC curve analysis of qualitative features (shape, margin, and internal components) evaluated by radiologists for distinguishing benign from malignant lymph nodes showed an AUC of 0.827 (95% CI: 0.728–0.926). In comparison, quantitative histogram analysis achieved a higher AUC of 0.870 (95% CI: 0.781–0.959). However, the difference between the two methods was not statistically significant (p = 0.336).
The ROC curve analysis for the discrimination of benign versus malignant lymph nodes showed an AUC of 0.919 for nodal size and 0.849 for nodal volume (Fig 6). Further analysis revealed that a cut-off nodal size of 13.0 mm tended to indicate malignancy, with a sensitivity of 0.706, a specificity of 0.957, a Youden’s Index of 0.662, an accuracy of 0.850, and a positive likelihood ratio (LR+) of 16.23. Similarly, a cut-off nodal volume of 1.632 ml indicated a tendency toward malignancy, with a sensitivity of 0.735, a specificity of 0.870, a Youden’s Index of 0.605, an accuracy of 0.813, and an LR+ of 5.64. The results are summarized in Table 4.
[Figure omitted. See PDF.]
[Figure omitted. See PDF.]
Discussion
Our study showed that malignant lymph nodes were statistically significantly larger than benign ones. We used a cut-off nodal short axis of 13.0 mm and nodal volume of 1.632 ml to indicate malignant lymph nodes with precision, with an accuracy of 0.850 and 0.813, respectively. This supports most previously published studies that often use a short axis greater than 1 cm as the cutoff for a metastatic lymph node in NSCLC [18]. However, the sensitivity of using a short axis less than 1 cm to identify a normal lymph node is only between 46% and 57% [19,20], indicating that non-enlarged nodes can still be malignant. Meanwhile, emerging studies on normal lymph nodes suggest that relying solely on size to differentiate malignant from benign mediastinal nodes is not reliable, as these studies indicate a relatively high frequency of malignancy [21,22].
In our study, the majority of malignant nodes had an irregular shape (76.5%), ill-defined margins (73.5%), and contained necrotic components (50.0%). This could be due to the fact that benign nodes become more rounded or take on an irregular shape as a result of malignant infiltration. However, the contour of the node as seen on CT and MR imaging can have greater diagnostic value, as malignant nodes tend to show irregular or ill-defined borders as a result of extracapsular disease extension or uneven growth [13,23]. On CT images, large metastatic nodes often display a heterogeneous appearance. Heterogeneity is a well-known characteristic of cancer, reflecting alterations in tissue architecture that are probably the result of various factors such as cell infiltration, abnormal angiogenesis, myxoid changes, and tumor necrosis [24,25]. A lower density nodal center seen on CT can be a result of necrosis. Even normal-sized nodes with necrosis should be considered as potentially malignant in these patients. Although low attenuation nodes can be a feature of malignant infiltration, they are also found in tuberculosis and fungal infections, which were not included in our study. Therefore, low attenuation nodes alone are not diagnostic of malignancy [13,23].
In our study, the presence of internal hilar fat was seen exclusively in benign lymph nodes. This is consistent with published studies which suggest that the presence of a fatty hilum is a commonly used parameter for differentiating between benign and malignant lymph nodes [26,27]. A typical normal lymph node has a uniform appearance, and the presence of hilar fat can be indicative of benignity, although this is not always the case [23].
Our study found that interobserver agreement for qualitative nodal morphology evaluation was fair to moderate (Kappa: 0.412–0.756; p < 0.05), suggesting limitations in subjective assessment of shape, margin, and internal components using visual inspection alone. Although the difference was not statistically significant (p = 0.336), quantitative histogram analysis achieved a higher AUC (0.870) than qualitative feature assessment by radiologists (AUC 0.827), indicating its potential for better distinguishing between benign and malignant lymph nodes. A possible explanation is that subjective image assessment relies on a radiologist’s perception, making it prone to inconsistency and variability. In contrast, objective assessment uses quantitative metrics for a standardized approach [28]. Radiomics and quantitative imaging extract measurable data to characterize tissue properties. The histogram analysis provides additional quantitative insights into the heterogeneity of mediastinal lymph nodes in NSCLC patients, which are not achievable through conventional morphologic evaluation alone.
Some prior studies postulated that CT texture analysis is able to inform the histopathological tumor characteristics, and that the more heterogeneous a tumor is, the more biologically aggressive its behavior tends to be [11,29]. In this study, we utilized a first-order CT texture analysis and found significant differences in mean attenuation, skewness, kurtosis, and entropy between benign and malignant lymph nodes. In malignant lymph nodes, a higher density is expected due to the loss of fatty tissue in the hilar region, in contrast to benign lymph nodes. As expected, our results showed that malignant lymph nodes tend to have high mean attenuation and more negative skewness. However, the MPP in this study showed similarities between the two groups and was not statistically significant which may be attributed to several factors. One potential explanation is the heterogeneity within the malignant lymph nodes themselves. While malignant nodes are generally expected to have higher attenuation due to the loss of fatty tissue, the presence of necrotic areas, which were found in about 50% of the malignant nodes, could lower the overall attenuation [24], balancing out the mean values when compared to benign nodes. Additionally, benign lymph nodes can sometimes show increased attenuation due to reactive changes, such as inflammation or fibrosis, which can occur in response to infection or other non-malignant processes [27]. These changes can elevate the attenuation values of benign nodes, making the MPP values between benign and malignant nodes more comparable.
Several features associated with texture derived from CT and MRI images are able to reflect unique histopathological tumor characteristics at a microstructural level. These features have proven useful in discriminating benign from malignant lymph nodes through the use of CT texture features [29]. The kurtosis of first-grade radiomics features reflects the peak apex degree of the gray distribution within the pixel. High kurtosis implies that the gray area is further away from the mean distribution, while low kurtosis indicates the opposite trend. Meanwhile, skewness reflects the offset characteristics and symmetry of the gray distribution relative to the mean. Higher skewness suggests higher tumor heterogeneity and a greater possibility of progression [30,31]. Presumably, higher heterogeneity in malignant lymph nodes is caused by the presence of tumor deposits, which results in increased CT heterogeneity that can be quantified using entropy-related second-order statistics [29].
According to hypotheses pertinent to heterogeneity in malignancy, it can be expected that malignant tumors would have a less peaked histogram, a low kurtosis value, and a high entropy value. However, our study found the opposite results. The CT histogram texture features were analyzed to identify factors associated with lymph node differentiation. The possible explanation is that the density of lymph nodes can vary significantly based on their anatomical location and the underlying pathology [6]. Benign nodes, especially in regions with a high prevalence of infectious agents, particularly in Thailand, might develop calcifications or other changes that increase their heterogeneity on CT images. On the other hand, malignant nodes often exhibit rapid growth and central necrosis, leading to more uniform low-density areas [24], which could explain the lower entropy observed in malignant nodes.
Another possible explanation for these results is the presence of reactive changes in benign lymph nodes, which can occur due to inflammatory processes or infections. These reactive changes can introduce variability in the texture of benign nodes [27], leading to higher kurtosis and lower entropy values. However, our study suggests that integrating all CT histogram parameters is valuable for differentiating between benign and malignant lymph nodes, with an AUC of 0.870. This finding supports the idea that evaluating all CT histogram parameters collectively, rather than individually, can enhance the accuracy of nodal evaluation. Additionally, the use of first-order CT texture analysis, which focuses on basic histogram features, may not capture the complex microstructural differences between malignant and benign nodes as effectively as higher-order texture analysis methods.
This study has some limitations that should be considered. Firstly, the sample size is small and the study is retrospective in nature, which may limit the generalizability of the findings. Additionally, during post-processing, imaging filters such as Laplacian or Gaussian bandpass filters were not applied, potentially affecting some histogram parameters. To address these limitations and enhance the robustness of the findings, further evaluation using second-order statistical CT texture analysis is recommended, as it may provide more detailed and comprehensive information.
Conclusion
In summary, our study shows that morphologic CT features have the potential in distinguishing between malignant and benign mediastinal lymph nodes in patients with NSCLC. The histogram parameters obtained through CTTA can provide additional information that could assist in the differentiation between metastatic and benign lymph nodes. It is important to note that CTTA is not intended to replace tissue diagnosis, but it can be a valuable addition to the radiologist’s confidence for the characterization of lymph nodes and can have a significant impact on patient management. Further studies are warranted to compare CTTA with other diagnostic modalities, such as EBUS/EUS and F18-FDG PET/CT, on a patient-by-patient basis using a larger sample size, in order to determine its full potential in clinical practice.
Supporting information
S1 table. The Cut-off Values and Diagnostic Performance of Histogram Parameters for Differentiating Benign and Malignant Lymph Nodes.
https://doi.org/10.1371/journal.pone.0321921.s001
(DOCX)
S1 file. Raw data.
https://doi.org/10.1371/journal.pone.0321921.s002
(XLSX)
S2 file. Ethics approval letter (RAD-2565–09120).
https://doi.org/10.1371/journal.pone.0321921.s003
Acknowledgments
The authors would like to express our sincere gratitude to Mr. Yanakawee Khatsitalee for his assistance with statistical analysis and consultation for this study.
References
1. 1. Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229–63. pmid:38572751
* View Article
* PubMed/NCBI
* Google Scholar
2. 2. Zhang Y, Vaccarella S, Morgan E, Li M, Etxeberria J, Chokunonga E, et al. Global variations in lung cancer incidence by histological subtype in 2020: a population-based study. Lancet Oncol. 2023;24(11):1206–18. pmid:37837979
* View Article
* PubMed/NCBI
* Google Scholar
3. 3. Iskender I, Kadioglu SZ, Cosgun T, Kapicibasi HO, Sagiroglu G, Kosar A, et al. False-positivity of mediastinal lymph nodes has negative effect on survival in potentially resectable non-small cell lung cancer. Eur J Cardiothorac Surg. 2012;41(4):874–9. pmid:22423060
* View Article
* PubMed/NCBI
* Google Scholar
4. 4. Kim KY, Park HL, Kang HS, Lee HY, Yoo IR, Lee SH, et al. Clinical characteristics and outcome of pathologic N0 non-small cell lung cancer patients with false positive mediastinal lymph node metastasis on FDG PET-CT. In Vivo. 2021;35(3):1829–36. pmid:33910869
* View Article
* PubMed/NCBI
* Google Scholar
5. 5. Turkmen C, Sonmezoglu K, Toker A, Yilmazbayhan D, Dilege S, Halac M, et al. The additional value of FDG PET imaging for distinguishing N0 or N1 from N2 stage in preoperative staging of non-small cell lung cancer in region where the prevalence of inflammatory lung disease is high. Clin Nucl Med. 2007;32(8):607–12. pmid:17667432
* View Article
* PubMed/NCBI
* Google Scholar
6. 6. Feng M, Yang X, Ma Q, He Y. Retrospective analysis for the false positive diagnosis of PET-CT scan in lung cancer patients. Medicine (Baltimore). 2017;96(42):e7415. pmid:29049175
* View Article
* PubMed/NCBI
* Google Scholar
7. 7. Pham T, Watanabe Y, Higuchi M, Suzuki H. Texture analysis and synthesis of malignant and benign mediastinal lymph nodes in patients with lung cancer on computed tomography. Scientific Rep. 2017;7:43209.
* View Article
* Google Scholar
8. 8. Figueiredo VR, Cardoso PFG, Jacomelli M, Santos LM, Minata M, Terra RM. EBUS-TBNA versus surgical mediastinoscopy for mediastinal lymph node staging in potentially operable non-small cell lung cancer: a systematic review and meta-analysis. J Bras Pneumol. 2020;46(6):e20190221. pmid:33111752
* View Article
* PubMed/NCBI
* Google Scholar
9. 9. El-Sherief AH, Lau CT, Wu CC, Drake RL, Abbott GF, Rice TW. International association for the study of lung cancer (IASLC) lymph node map: radiologic review with CT illustration. Radiographics. 2014;34(6):1680–91. pmid:25310423
* View Article
* PubMed/NCBI
* Google Scholar
10. 10. Flechsig P, Kratochwil C, Schwartz LH, Rath D, Moltz J, Antoch G, et al. Quantitative volumetric CT-histogram analysis in N-staging of 18F-FDG-equivocal patients with lung cancer. J Nucl Med. 2014;55(4):559–64. pmid:24525209
* View Article
* PubMed/NCBI
* Google Scholar
11. 11. Lubner M, Smith A, Sandrasegaran K, Sahani D, Pickhardt P. CT texture analysis: definitions, applications, biologic correlates, and challenges. Radiology. 2017;37(5):1483–503.
* View Article
* Google Scholar
12. 12. Flechsig P, Frank P, Kratochwil C, Antoch G, Rath D, Moltz J, et al. Radiomic analysis using density threshold for FDG-PET/CT-based N-staging in lung cancer patients. Mol Imaging Biol. 2017;19(2):315–22. pmid:27539308
* View Article
* PubMed/NCBI
* Google Scholar
13. 13. Bayanati H, E Thornhill R, Souza CA, Sethi-Virmani V, Gupta A, Maziak D, et al. Quantitative CT texture and shape analysis: can it differentiate benign and malignant mediastinal lymph nodes in patients with primary lung cancer? Eur Radiol. 2015;25(2):480–7. pmid:25216770
* View Article
* PubMed/NCBI
* Google Scholar
14. 14. Koda E, Yamashiro T, Onoe R, Handa H, Azagami S, Matsushita S, et al. CT texture analysis of mediastinal lymphadenopathy: combining with US-based elastographic parameter and discrimination between sarcoidosis and lymph node metastasis from small cell lung cancer. PLoS One. 2020;15(12):e0243181. pmid:33264372
* View Article
* PubMed/NCBI
* Google Scholar
15. 15. Wang Z-L, Yan Y, Li X-T, Li Y-L, Li Z-W, Sun Y-S. Usefulness of attenuation value on computed tomography plain scan for diagnosing enlarged mediastinal lymph nodes metastases. Quant Imaging Med Surg. 2023;13(9):5759–69. pmid:37711834
* View Article
* PubMed/NCBI
* Google Scholar
16. 16. Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med. 2016;15(2):155–63. pmid:27330520
* View Article
* PubMed/NCBI
* Google Scholar
17. 17. Viera A, Garrett J. Understanding interobserver agreement: the kappa statistic. Family Med. 2005;37(5):360–3.
* View Article
* Google Scholar
18. 18. Glazer GM, Gross BH, Quint LE, Francis IR, Bookstein FL, Orringer MB. Normal mediastinal lymph nodes: number and size according to American Thoracic Society mapping. AJR Am J Roentgenol. 1985;144(2):261–5. pmid:3871268
* View Article
* PubMed/NCBI
* Google Scholar
19. 19. Andersen MB, Harders SW, Ganeshan B, Thygesen J, Torp Madsen HH, Rasmussen F. CT texture analysis can help differentiate between malignant and benign lymph nodes in the mediastinum in patients suspected for lung cancer. Acta Radiol. 2016;57(6):669–76. pmid:26271125
* View Article
* PubMed/NCBI
* Google Scholar
20. 20. Toloza EM, Harpole L, McCrory DC. Noninvasive staging of non-small cell lung cancer: a review of the current evidence. Chest. 2003;123(1 Suppl):137S-146S. pmid:12527573
* View Article
* PubMed/NCBI
* Google Scholar
21. 21. Prenzel KL, Mönig SP, Sinning JM, Baldus SE, Brochhagen H-G, Schneider PM, et al. Lymph node size and metastatic infiltration in non-small cell lung cancer. Chest. 2003;123(2):463–7. pmid:12576367
* View Article
* PubMed/NCBI
* Google Scholar
22. 22. Fritscher-Ravens A, Bohuslavizki KH, Brandt L, Bobrowski C, Lund C, Knöfel WT, et al. Mediastinal lymph node involvement in potentially resectable lung cancer: comparison of CT, positron emission tomography, and endoscopic ultrasonography with and without fine-needle aspiration. Chest. 2003;123(2):442–51. pmid:12576364
* View Article
* PubMed/NCBI
* Google Scholar
23. 23. Ganeshalingam S, Koh D-M. Nodal staging. Cancer Imaging. 2009;9(1):104–11. pmid:20080453
* View Article
* PubMed/NCBI
* Google Scholar
24. 24. Davnall F, Yip CSP, Ljungqvist G, Selmi M, Ng F, Sanghera B, et al. Assessment of tumor heterogeneity: an emerging imaging tool for clinical practice?. Insights Imaging. 2012;3(6):573–89. pmid:23093486
* View Article
* PubMed/NCBI
* Google Scholar
25. 25. Ganeshan B, Panayiotou E, Burnand K, Dizdarevic S, Miles K. Tumour heterogeneity in non-small cell lung carcinoma assessed by CT texture analysis: a potential marker of survival. Eur Radiol. 2012;22(4):796–802. pmid:22086561
* View Article
* PubMed/NCBI
* Google Scholar
26. 26. Liu Z, Zeng W, Liu C, Wang S, Xiong Y, Guo Y, et al. Diagnostic accuracy of ultrasonographic features for lymph node metastasis in papillary thyroid microcarcinoma: a single-center retrospective study. World J Surg Oncol. 2017;15(1):32. pmid:28125992
* View Article
* PubMed/NCBI
* Google Scholar
27. 27. Vineela E, Sakalecha A, Narayanrao Suresh T. Role of sonoelastography in differentiating benign from malignant cervical lymph nodes and correlating with pathology. Cureus. 2022;14(3):e22984.
* View Article
* Google Scholar
28. 28. Varghese AP, Naik S, Asrar Up Haq Andrabi S, Luharia A, Tivaskar S. Enhancing radiological diagnosis: a comprehensive review of image quality assessment and optimization strategies. Cureus. 2024;16(6):e63016. pmid:39050319
* View Article
* PubMed/NCBI
* Google Scholar
29. 29. Meyer H-J, Schnarkowski B, Pappisch J, Kerkhoff T, Wirtz H, Höhn A-K, et al. CT texture analysis and node-RADS CT score of mediastinal lymph nodes - diagnostic performance in lung cancer patients. Cancer Imaging. 2022;22(1):75. pmid:36567339
* View Article
* PubMed/NCBI
* Google Scholar
30. 30. Wei S, Shi B, Zhang J, Li N. Differentiating mass-like tuberculosis from lung cancer based on radiomics and CT features. Transl Cancer Res. 2021;10(10):4454–63. pmid:35116302
* View Article
* PubMed/NCBI
* Google Scholar
31. 31. Lubner M, Smith A, Sandrasegaran K, Sahani D, Pickhardt P. CT texture analysis: definitions, applications, biologic correlates, and challenges. Radiographics. 2017;37(5):1483–503.morphologicfeatures
* View Article
* Google Scholar
Citation: Prakaikietikul P, Wannasopha Y, Euathrongchit J, Tantraworasin A (2025) CT features and histogram analysis of non-contrast images for differentiating malignant and benign mediastinal lymph nodes in Non-Small Cell Lung Cancer (NSCLC). PLoS ONE 20(4): e0321921. https://doi.org/10.1371/journal.pone.0321921
About the Authors:
Pakorn Prakaikietikul
Contributed equally to this work with: Pakorn Prakaikietikul, Yutthaphan Wannasopha
Roles: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Validation, Visualization, Writing – original draft, Writing – review & editing
Affiliation: Department of Radiology, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand
Yutthaphan Wannasopha
Contributed equally to this work with: Pakorn Prakaikietikul, Yutthaphan Wannasopha
Roles: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing
E-mail: [email protected]
Affiliation: Department of Radiology, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand
ORICD: https://orcid.org/0000-0002-6727-0828
Juntima Euathrongchit
Roles: Conceptualization, Supervision, Writing – review & editing
¶‡ JE and AT also contributed equally to this work.
Affiliation: Department of Radiology, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand
Apichat Tantraworasin
Roles: Conceptualization, Formal analysis, Writing – review & editing
¶‡ JE and AT also contributed equally to this work.
Affiliations: Clinical Epidemiology and Clinical Statistic Center, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand, Department of Surgery, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand
ORICD: https://orcid.org/0000-0002-7913-2799
1. Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229–63. pmid:38572751
2. Zhang Y, Vaccarella S, Morgan E, Li M, Etxeberria J, Chokunonga E, et al. Global variations in lung cancer incidence by histological subtype in 2020: a population-based study. Lancet Oncol. 2023;24(11):1206–18. pmid:37837979
3. Iskender I, Kadioglu SZ, Cosgun T, Kapicibasi HO, Sagiroglu G, Kosar A, et al. False-positivity of mediastinal lymph nodes has negative effect on survival in potentially resectable non-small cell lung cancer. Eur J Cardiothorac Surg. 2012;41(4):874–9. pmid:22423060
4. Kim KY, Park HL, Kang HS, Lee HY, Yoo IR, Lee SH, et al. Clinical characteristics and outcome of pathologic N0 non-small cell lung cancer patients with false positive mediastinal lymph node metastasis on FDG PET-CT. In Vivo. 2021;35(3):1829–36. pmid:33910869
5. Turkmen C, Sonmezoglu K, Toker A, Yilmazbayhan D, Dilege S, Halac M, et al. The additional value of FDG PET imaging for distinguishing N0 or N1 from N2 stage in preoperative staging of non-small cell lung cancer in region where the prevalence of inflammatory lung disease is high. Clin Nucl Med. 2007;32(8):607–12. pmid:17667432
6. Feng M, Yang X, Ma Q, He Y. Retrospective analysis for the false positive diagnosis of PET-CT scan in lung cancer patients. Medicine (Baltimore). 2017;96(42):e7415. pmid:29049175
7. Pham T, Watanabe Y, Higuchi M, Suzuki H. Texture analysis and synthesis of malignant and benign mediastinal lymph nodes in patients with lung cancer on computed tomography. Scientific Rep. 2017;7:43209.
8. Figueiredo VR, Cardoso PFG, Jacomelli M, Santos LM, Minata M, Terra RM. EBUS-TBNA versus surgical mediastinoscopy for mediastinal lymph node staging in potentially operable non-small cell lung cancer: a systematic review and meta-analysis. J Bras Pneumol. 2020;46(6):e20190221. pmid:33111752
9. El-Sherief AH, Lau CT, Wu CC, Drake RL, Abbott GF, Rice TW. International association for the study of lung cancer (IASLC) lymph node map: radiologic review with CT illustration. Radiographics. 2014;34(6):1680–91. pmid:25310423
10. Flechsig P, Kratochwil C, Schwartz LH, Rath D, Moltz J, Antoch G, et al. Quantitative volumetric CT-histogram analysis in N-staging of 18F-FDG-equivocal patients with lung cancer. J Nucl Med. 2014;55(4):559–64. pmid:24525209
11. Lubner M, Smith A, Sandrasegaran K, Sahani D, Pickhardt P. CT texture analysis: definitions, applications, biologic correlates, and challenges. Radiology. 2017;37(5):1483–503.
12. Flechsig P, Frank P, Kratochwil C, Antoch G, Rath D, Moltz J, et al. Radiomic analysis using density threshold for FDG-PET/CT-based N-staging in lung cancer patients. Mol Imaging Biol. 2017;19(2):315–22. pmid:27539308
13. Bayanati H, E Thornhill R, Souza CA, Sethi-Virmani V, Gupta A, Maziak D, et al. Quantitative CT texture and shape analysis: can it differentiate benign and malignant mediastinal lymph nodes in patients with primary lung cancer? Eur Radiol. 2015;25(2):480–7. pmid:25216770
14. Koda E, Yamashiro T, Onoe R, Handa H, Azagami S, Matsushita S, et al. CT texture analysis of mediastinal lymphadenopathy: combining with US-based elastographic parameter and discrimination between sarcoidosis and lymph node metastasis from small cell lung cancer. PLoS One. 2020;15(12):e0243181. pmid:33264372
15. Wang Z-L, Yan Y, Li X-T, Li Y-L, Li Z-W, Sun Y-S. Usefulness of attenuation value on computed tomography plain scan for diagnosing enlarged mediastinal lymph nodes metastases. Quant Imaging Med Surg. 2023;13(9):5759–69. pmid:37711834
16. Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med. 2016;15(2):155–63. pmid:27330520
17. Viera A, Garrett J. Understanding interobserver agreement: the kappa statistic. Family Med. 2005;37(5):360–3.
18. Glazer GM, Gross BH, Quint LE, Francis IR, Bookstein FL, Orringer MB. Normal mediastinal lymph nodes: number and size according to American Thoracic Society mapping. AJR Am J Roentgenol. 1985;144(2):261–5. pmid:3871268
19. Andersen MB, Harders SW, Ganeshan B, Thygesen J, Torp Madsen HH, Rasmussen F. CT texture analysis can help differentiate between malignant and benign lymph nodes in the mediastinum in patients suspected for lung cancer. Acta Radiol. 2016;57(6):669–76. pmid:26271125
20. Toloza EM, Harpole L, McCrory DC. Noninvasive staging of non-small cell lung cancer: a review of the current evidence. Chest. 2003;123(1 Suppl):137S-146S. pmid:12527573
21. Prenzel KL, Mönig SP, Sinning JM, Baldus SE, Brochhagen H-G, Schneider PM, et al. Lymph node size and metastatic infiltration in non-small cell lung cancer. Chest. 2003;123(2):463–7. pmid:12576367
22. Fritscher-Ravens A, Bohuslavizki KH, Brandt L, Bobrowski C, Lund C, Knöfel WT, et al. Mediastinal lymph node involvement in potentially resectable lung cancer: comparison of CT, positron emission tomography, and endoscopic ultrasonography with and without fine-needle aspiration. Chest. 2003;123(2):442–51. pmid:12576364
23. Ganeshalingam S, Koh D-M. Nodal staging. Cancer Imaging. 2009;9(1):104–11. pmid:20080453
24. Davnall F, Yip CSP, Ljungqvist G, Selmi M, Ng F, Sanghera B, et al. Assessment of tumor heterogeneity: an emerging imaging tool for clinical practice?. Insights Imaging. 2012;3(6):573–89. pmid:23093486
25. Ganeshan B, Panayiotou E, Burnand K, Dizdarevic S, Miles K. Tumour heterogeneity in non-small cell lung carcinoma assessed by CT texture analysis: a potential marker of survival. Eur Radiol. 2012;22(4):796–802. pmid:22086561
26. Liu Z, Zeng W, Liu C, Wang S, Xiong Y, Guo Y, et al. Diagnostic accuracy of ultrasonographic features for lymph node metastasis in papillary thyroid microcarcinoma: a single-center retrospective study. World J Surg Oncol. 2017;15(1):32. pmid:28125992
27. Vineela E, Sakalecha A, Narayanrao Suresh T. Role of sonoelastography in differentiating benign from malignant cervical lymph nodes and correlating with pathology. Cureus. 2022;14(3):e22984.
28. Varghese AP, Naik S, Asrar Up Haq Andrabi S, Luharia A, Tivaskar S. Enhancing radiological diagnosis: a comprehensive review of image quality assessment and optimization strategies. Cureus. 2024;16(6):e63016. pmid:39050319
29. Meyer H-J, Schnarkowski B, Pappisch J, Kerkhoff T, Wirtz H, Höhn A-K, et al. CT texture analysis and node-RADS CT score of mediastinal lymph nodes - diagnostic performance in lung cancer patients. Cancer Imaging. 2022;22(1):75. pmid:36567339
30. Wei S, Shi B, Zhang J, Li N. Differentiating mass-like tuberculosis from lung cancer based on radiomics and CT features. Transl Cancer Res. 2021;10(10):4454–63. pmid:35116302
31. Lubner M, Smith A, Sandrasegaran K, Sahani D, Pickhardt P. CT texture analysis: definitions, applications, biologic correlates, and challenges. Radiographics. 2017;37(5):1483–503.morphologicfeatures
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
© 2025 Prakaikietikul et al. This is an open access article distributed under the terms of the Creative Commons Attribution License: http://creativecommons.org/licenses/by/4.0/ (the “License”), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
Objective
To evaluate the diagnostic value of CT features and histogram analysis in distinguishing between malignant and benign mediastinal lymph nodes in patients with non-small cell lung cancer (NSCLC).
Method
This retrospective study analyzed non-contrast chest CT images from 40 NSCLC patients, comprising 80 pathology-proven mediastinal lymph nodes (46 benign, 34 metastasis). Morphologic features, including size, shape, margins, and internal composition, were independently assessed by two radiologists. Histogram analysis was conducted using the Synapse Vincent system with six parameters: mean attenuation, mean positive pixel (MPP), standard deviation (SD), skewness, kurtosis, and entropy. Statistical analysis included the Mann-Whitney test for continuous data, Fisher’s exact test for categorical data, and receiver-operating characteristic (ROC) curve analysis to assess diagnostic accuracy, with statistical significance set at p < 0.05.
Results
Malignant lymph nodes demonstrated significantly larger sizes (p < 0.001), ill-defined margins (p = 0.024), irregular shapes (p < 0.001), and the presence of necrotic areas (p < 0.001). A nodal size cutoff of 13.0 mm and volume of 1.632 ml were strongly associated with malignancy, yielding high diagnostic accuracy with sensitivities of 70.6% and 73.5% and specificities of 95.7% and 87.0%, respectively. Significant differences were observed between benign and malignant lymph nodes in several CT histogram parameters, including mean attenuation (p = 0.004), skewness (p = 0.041), kurtosis (p = 0.005), and entropy (p < 0.001). The integrating all CT histogram parameters yielded an area under the curve (AUC) of 0.870 for differentiating between benign and malignant lymph nodes.
Conclusion
The combination of morphologic CT features and CT histogram analysis offers a robust method for differentiating malignant from benign mediastinal lymph nodes in NSCLC patients, potentially enhancing diagnostic accuracy and informing treatment strategies.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer