A new predictive scoring system based on clinical data and computed tomography features for diagnosing EGFR-mutated lung adenocarcinoma

Original Article

A new predictive scoring system based on clinical data and computed tomography features for diagnosing EGFR-mutated lung adenocarcinoma

Y. Cao, MD*, H. Xu, MD*

doi: http://dx.doi.org/10.3747/co.25.3805



We aimed to develop a new EGFR mutation–predictive scoring system to use in screening for EGFR-mutated lung adenocarcinomas (lacs).


The study enrolled 279 patients with lac, including 121 patients with EGFR wild-type tumours and 158 with EGFR-mutated tumours. The Student t-test, chi-square test, or Fisher exact test was applied to discriminate clinical and computed tomography (ct) features between the two groups. Using a principal component analysis (pca) model, we derived predictive coefficients for the presence of EGFR mutation in lac.


The EGFR mutation–predictive score includes sex, smoking history, homogeneity, ground-glass opacity (ggo) on imaging, and the presence of pericardial effusion. The pca predictive model took this form:Model scores ranged from 79 to 147. The area under the receiver operating characteristic curve was 0.752 [95% confidence interval (ci): 0.697 to 0.801] in the lac population at the optimal cut-off value of 109, and the sensitivity and specificity were 68.4% (95% ci: 60.5% to 75.5%) and 74.4% (95% ci: 65.6% to 81.9%) respectively.


The EGFR mutation risk scoring system based on clinical data and ct features is noninvasive and user-friendly. The model appears to frame a positive predictive value and was able to determine the value of repeating a biopsy if tissue is limited.

KEYWORDS: Computed tomography, epidermal growth factor receptor, lung cancer, adenocarcinoma


The cancer-related lethality rate for lung and bronchus cancer is high worldwide, accounting annually for approximately 0.16 million deaths in the United States1 and 0.61 million deaths in China2. Because of different treatment methods, lung cancers are classified into two major categories: non-small-cell lung cancer (nsclc) and small-cell lung cancer. Lung adenocarcinomas (lacs) represent a major proportion of the nsclc group3.

Earlier studies have demonstrated that pathology classification has no significant effect on treatment and prognosis in most subtypes of nsclc (neuroendocrine tumours excepted)4,5. However, advances in research and continuous improvements in gene test technology have revealed that not all nsclcs respond to treatment in a consistent manner. Increasing attention has been paid to the individual treatment of some particular lung cancer subtypes, among which the most studied is targeted therapy for lung cancer with gene mutations6. Therapy with an epidermal growth factor receptor (egfr) tyrosine kinase inhibitor, which is aimed explicitly at EGFR mutations in lung cancer, delays progression in patients with EGFR mutations7,8. Compared with conventional chemotherapy, therapy with an egfr tyrosine kinase inhibitor is associated with fewer complications and reduced toxicity8,9. Nevertheless, therapy with a tyrosine kinase inhibitor is not effective in the treatment of wild-type nsclc10.

EGFR mutations are among the most common genetic mutations in lung cancer, and they represent the most frequent mutation in lac11. For genotyping the tumour, histologic samples of plasma or another body fluid must be obtained1214. In some cases, obtaining those materials is challenging15, and testing has always been problematic. Identification of the EGFR mutation status in nsclc without a molecular examination would be beneficial for choosing treatment.

Some studies have attempted to determine the relationship between molecular findings and certain disease features such as sex, smoking status, the presence of ground-glass opacity (ggo) on imaging, and other features1618. However, most of the reports described only the features with statistical significance, whose clinical value is limited. We retrospectively analyzed a cohort of East Asian patients with lac who had been tested for EGFR mutations in our hospital. We hoped to develop a scoring system for EGFR mutation screening in lac, thus providing supplementary diagnostic information for new patients from whom tissue specimens cannot be obtained.


Patient Selection

This retrospective study was approved by the institutional review board. Informed consent was waived. Results of consecutive EGFR mutation tests in lung cancer were obtained from the hospital’s pathology database.

Between January 2014 and October 2016, 365 patients underwent EGFR mutation testing. Patients were excluded if they had no computed tomography (ct) imaging data within 1 month before surgery; if they had no non-contrast ct data or contrast-enhanced ct data available; if pathology had confirmed that the lesion was not lac; if multiple tumours were present in the lung, such that the relationship between the tumour and the pathology results could not be independently determined; if tumour boundaries could not be distinguished; and if other types of mutations or the EGFR T790M mutation was present. As a result, 86 patients were excluded.

For the remaining 279 patients, clinical and pathology data were collected. Clinical data included age, sex, tumour staging, and smoking status (the definition of a “nonsmoker” was never having smoked). Tumour staging reflected the 7th edition of the American Joint Committee on Cancer staging manual19.

Histologic Evaluation and Molecular Analysis

Per routine procedure at our hospital, after histology specimens were obtained, they were formalin-fixed and then stained with hematoxylin–eosin. Immunohistochemical analysis was used to assess tumours that could not be diagnosed by routine procedures. Two pathologists reviewed the pathology specimens and recorded the results separately. For all disagreements about diagnosis, the two pathologists discussed the findings to obtain a final consensus. Diagnostic criteria were based on the 2015 World Health Organization classification of lung tumours20.

A fluorescence polymerase chain reaction diagnostic kit (Amoy Diagnostics, Xiamen, P.R.C.) was used to perform an EGFR gene mutation analysis for exons 18–21 in the specimens. The test result was determined by the cycle threshold score. The result was considered to be strongly positive at a cycle threshold score between 0 and 26. A cycle threshold score between 26 and 29 indicated weak positive expression. If the score was greater than 29, the result was considered negative.

CT Imaging

Two ct systems (Sensation 16 and Somatom Definition: Siemens Medical Systems, Erlangen, Germany) were involved in the retrospective study. Images were obtained at 5 mm section thickness, without a gap, using the mediastinal algorithm in both ct systems as a “mediastinal window,” and at 1 mm section thickness, without a gap, using the high-frequency algorithm as a “lung window.” The main imaging parameters were 120 kV and 100 mA (Sensation 16) and 120 kV and 100–400 mA with dose modulation (Somatom Definition).

We selected Ultravist (Ultravist 300: Bayer Pharma, Berlin, Germany) as the contrast medium. Contrast-enhanced images were collected with a 45 s delay after intravenous injection of the contrast medium. The injection rate was 3.0 mL/s, and the dose was based on the patient’s weight.

CT Interpretation

All images were reviewed by two independent board-certified thoracic radiologists. Both radiologists were unaware of the pathology diagnosis and the EGFR test results. All features were evaluated on cross-sectional images, including the lung window and the mediastinal window. Discrepancies between the radiologists were resolved by discussion. The final measurements and counts were obtained by averaging the results from the two radiologists.

In addition to morphologic features, we recorded the tumour disappearance rate (tdr), relative enhancement, lobulation, pleural retraction, calcification, bubble-like lucency, air bronchography, pneumonia-like consolidation, ggo, vessel convergence sign, spiculation, pericardial effusion, pleural effusion, and mediastinal lymph node metastasis.

These definitions of the imaging features were used:

  • tdr = 1 – MaxDmediastinal × MinDmediastinal / (MaxDlung × MinDlung), where MaxDmediastinal and MinDmediastinal represent the measurements of the longest and shortest diameters of the tumour in millimeters in the mediastinal window; and MaxDlung and MinDlung represent the measurements of the longest and shortest diameters of the tumour in millimeters in the lung window21. The cut-off value for tdr was 0.5.

  • ■ Relative enhancement = (Apost – Apre) / Eart, where Apost represents the attenuation of the tumour in the contrast-enhanced image; Apre represents the attenuation of the tumour in the non-contrast image; and Eart indicates the attenuation of the descending aorta in the contrast-enhanced image18.

  • ■ Lobulation = portion of the tumour surface exhibiting a slightly wavy appearance, with the exception of regions abutting the pleura, where “lobulation” was calculated based on numbers so as to divide the patients into two groups.

  • ■ Pneumonia-like consolidation = a homogenous or heterogeneous opacity lesion in the lung

    Unlike ggo, pneumonia-like consolidation effaces the blood vessel shadows, and occasionally an air bronchogram can be displayed without any bacterial or obstructive pneumonia16.

  • ■ Spiculation = cut-off of the burr diameter, where the cut-off was defined so as to divide the patients into two groups. That cut-off was 2 mm.

Statistical Analysis

Most of the statistical analyses were performed using the SPSS software application (version 21.0: IBM, Armonk, NY, U.S.A.). The receiver operating characteristic curve was analysed using the MedCalc software application (version 15.8: MedCalc Software, Ostend, Belgium). Categorical variables were analyzed using the chi-square or Fisher exact test, as appropriate. Continuous variables were analyzed using the Student t-test. A p value less than 0.05 was considered statistically significant. Collinearity diagnosis was applied to detect collinearity between the features. If no collinearity was noted between the parameters, the parameters were loaded in series into a multivariable logistic regression model. Otherwise, pca was used to solve the problem. The new EGFR mutation predictive score was derived by multiplying the regression of the pca coefficient for the significant features by 10 and then rounding to the nearest integer. All significant features were thereafter calculated together in both groups, using the resulting numbers as scores. The final data were used to generate the receiver operating characteristic curve and to calculate the area under the curve. The Youden index was used to calculate the optimal cut-off.


The differences between the two radiologists were, for the presence of calcification, 1 in 279 patient exams (0.36%); for the presence of air bronchogram, 3 in 279 patient exams (1.08%); for vessel convergence sign, 6 in 279 patient exams (2.15%); and for pneumonia-like consolidation, 5 in 279 patient exams (1.79%).

Patient Characteristics

Of the 279 patients with lac involved in the analysis, 158 harboured an EGFR mutation: exon 18 mutation in 7 patients (4.43%), exon 19 mutation in 67 patients (42.41%), exon 20 mutation in 7 patients (4.43%), and exon 21 mutation in 73 patients (46.20%). The remaining 4 patients each had 2 mutations: an exon 21 mutation and an exon 19 mutation in 2 patients, and an exon 21 mutation and an exon 20 mutation in 2 patients. Of the 279 patients overall, 150 underwent lung resection or open lung biopsy, 125 underwent ct-guided biopsy, and 4 underwent fibre-optic bronchoscopy–guided biopsy. EGFR mutations were more common in women (n = 90, p < 0.001), nonsmokers (n = 120, p < 0.001), and patients with less-severe emphysema (n = 155, p = 0.002). However, no significant differences in age or tumour staging were noted between the wild-type group and the EGFR mutation group (Table i).

TABLE I Clinical features of the patients


CT Imaging Evaluation

In the EGFR mutation group, most of the tumours appeared homogenous in the contrast-enhanced image (p < 0.001). Ground-glass opacity appeared more common in patients with EGFR-mutated tumours (30 of 158) than in those with wild-type tumours (12 of 121, p = 0.036). Pericardial effusion was more frequently observed in patients with EGFR-mutated tumours (10 of 158, p = 0.026). No significant differences in location, attenuation, tdr, relative enhancement, pneumonia-type, lobulation, boundary, shape, spiculation, calcification, air bronchogram, cavitation, vascular convergence sign, pleural indentation, pleural effusion, and mediastinal lymph node metastasis were noted between the groups (Table ii).

TABLE II Computed tomography imaging evaluation


Collinear diagnosis revealed collinearity between emphysema and smoking history. We therefore used pca to obtain the coefficients for each feature. Using the previous statistical results, we divided the parameters of each feature into two categories. We set the parameters that tended to diagnose EGFR mutation to 2; the remaining parameters were set to 1 (Table iii). The original pca predictive scoring model took this form:

TABLE III The EGFR-mutation predictive score model, based on principal components analysis coefficient


To facilitate clinical application, we multiplied the pca coefficients of the significant feature by 10 and rounded to the nearest integer (Table iii). The pca predictive model thus took this form:

The receiver operating characteristic curve (Figure 1) used the final score as a variable (range: 79–148) and the genetic diagnosis as the dependent variable (1 = EGFR-mutated, 0 = wild-type). The area under the curve was 0.752 (95% ci: 0.697 to 0.801). The maximum value of the Youden index was 0.427, which corresponded to an optimal cut-off value of 109. When the calculated score was greater than 109, we tended to diagnose EGFR mutation. The sensitivity and specificity were 68.4% (95% ci: 60.5% to 75.5%) and 74.4% (95% ci: 65.6% to 81.9%) respectively.



FIGURE 1 Receiver operating characteristic (ROC) of the EGFR mutation predictive score. AUC = area under the curve.


Molecular analysis has always been the best choice for diagnosing EGFR mutation status. However, its application has been limited for various reasons. Some patients might not be able to undergo biopsy or surgery. Some medical centres might not have access to molecular profiling technology to perform the test. In addition, for some histology specimens, sufficient material might not be present to perform the test. We sought to develop an alternative method that is useful in patients who are unable to undergo molecular studies.

Our study was carried out in two parts. In the first part, we evaluated the relationship between the clinical and ct features associated with EGFR mutation. In lac, EGFR mutation was associated with 6 features. In the second part, we used pca to establish a predictive scoring model for EGFR mutation.

Our study focused on an East Asian population. Compared with patients of other ethnicities, East Asian patients with nsclc have the highest rates of EGFR mutation22,23. We demonstrated that the EGFR mutation rate was statistically higher in women and nonsmokers. The EGFR mutation rate in our group was 56.63%, and female patients accounted for 56.96% of the population, which was consistent with findings in other studies18,24. Smoking history is also a frequently studied epidemiologic feature. Our result was similar to that in earlier studies25,26.

In addition to the foregoing demographic and epidemiologic features, our study also identified emphysema status as an important predictor. Results of the Goddard scoring system27 revealed that the EGFR mutation rate was greater for patients having the mild type of emphysema (59.16%) than for those having the severe type (17.65%). We observed a correlation between smoking history and emphysema. The Goddard scoring system has certain advantages in clinical practice. The system requires only the ct image to be scored. If smoking history is not available, the Goddard score can provide some useful information.

We demonstrated that homogeneity and the presence of ggo and pericardial effusion differed statistically depending on mutation status. In the present study, wild-type tumours were more heterogeneous after contrast than were EGFR-mutated tumours. Research about this feature is limited28. Homogeneity before contrast was also studied, and no significant difference was noted by mutation status. It is possible that, compared with EGFR-mutated tumours, wild-type tumours cause more necrosis, making them appear more heterogeneous.

In the study of EGFR mutations in lac, ggo is one of the most common imaging features17,18,24. It appears more frequently in EGFR-mutated lac, and that finding has not caused much controversy. In the present study, pericardial effusion was observed more frequently in patients with EGFR mutation, and that condition is always caused by pericardial metastasis. Prior studies have demonstrated that, compared with EGFR-mutated lac tumours, those with the ALK rearrangement are more often associated with pleural or pericardial metastases29. Research focused on the relationship between pericardial metastasis and EGFR mutation is limited, and future studies should focus more on this feature.

In addition to a lack of statistically significant features, most of our results are consistent with prior research, with the exception of pneumonic-type lesions. Pneumonic-type lesions were previously known as consolidation-type lung cancer and were typically misdiagnosed as inflammatory pulmonary consolidation. A few studies demonstrated that patients with pneumonic-type lesions experience a significantly increased incidence of EGFR mutations, independent of sex, histologic type, and smoking history16. That research focused on patients before surgery, and our study focused on biopsy patients. In addition, biopsy specimens are considered highly sensitive for EGFR detection30,31. However, differences in patient selection might explain the different results.

In the first part of the study, we identified numerous significant features, but still found it difficult to reach a final diagnosis. Our challenge was how to integrate the parameters. In the second part of the study, we used clinical data and ct imaging to establish a practical predictive scoring system that could identify EGFR mutation. This EGFR mutation predictive scoring system comprises 6 features, including sex, smoking status, homogeneity, emphysema status, pericardial effusion status, and ggo on imaging. Those features were selected from among 24 clinical factors and ct features of lac. The scoring system can differentiate wild-type or EGFR-mutated tumours with moderate accuracy based on the resulting score. The system uses pca coefficients to convert the features of lac into qualitative data. The area under the curve was 0.752 in the lac population, and the sensitivity and specificity were 68.4% (95% ci: 60.5% to 75.5%) and 74.4% (95% ci: 65.6% to 81.9%) respectively.

Our scoring system is easily applied and does not require extensive clinical testing. It can provide complementary information for patients who cannot undergo genetic testing or for patients treated at clinical centres with limited access to molecular profiling. When false-negative molecular results are suspected, the scoring system can be used to assist in the diagnosis and to determine whether a repeat biopsy is necessary.

Our study has some limitations. First, the number of patients was not large, which could affect the accuracy of the prediction model. Second, because of geographic constraints, we focused on an East Asian population, and the results might therefore not have significant implications for other ethnic groups. Third, the scanning parameters were not exactly same in the two ct systems, which might have contributed bias.


Although considerable research has focused on EGFR mutation in lac, the present study is the first to use a scoring system to predict EGFR mutation status in an East Asian population. Earlier research studied Caucasian populations32. Prospective studies will have to determine whether the EGFR mutation scoring system can be used to inform treatment decisions or to predict tumour genotype.


We have read and understood Current Oncology’s policy on disclosing conflicts of interest, and we declare that we have none.


*Department of Radiology, Zhongnan Hospital of Wuhan University, Wuhan, Hubei, P.R.C..


1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin 2016;66:7–30.
cross-ref  pubmed  

2. Chen W, Zheng R, Baade PD, et al. Cancer statistics in China, 2015. CA Cancer J Clin 2016;66:115–32.
cross-ref  pubmed  

3. de Lima Lopes G, Segel JE, Tan DS, Do YK, Mok T, Finkelstein EA. Cost-effectiveness of epidermal growth factor receptor mutation testing and first-line treatment with gefitinib for patients with advanced adenocarcinoma of the lung. Cancer 2012;118:1032–9.

4. Sakashita S, Sakashita M, Sound Tsao M. Genes and pathology of non–small cell lung carcinoma. Semin Oncol 2014;41:28–39.
cross-ref  pubmed  

5. Alexander M, Wolfe R, Ball D, et al. Lung cancer prognostic index: a risk score to predict overall survival after the diagnosis of non-small-cell lung cancer. Br J Cancer 2017;117:744–51.

6. Buettner R, Wolf J, Thomas RK. Lessons learned from lung cancer genomics: the emerging concept of individualized diagnostics and treatment. J Clin Oncol 2013;31:1858–65.
cross-ref  pubmed  

7. Lee CK, Brown C, Gralla RJ, et al. Impact of egfr inhibitor in non–small cell lung cancer on progression-free and overall survival: a meta-analysis. J Natl Cancer Inst 2013;105:595–605.
cross-ref  pubmed  

8. Hoffknecht P, Tufman A, Wehler T, et al. Efficacy of the irreversible ErbB family blocker afatinib in epidermal growth factor receptor (egfr) tyrosine kinase inhibitor (tki)–pretreated non-small-cell lung cancer patients with brain metastases or leptomeningeal disease. J Thorac Oncol 2015;10:156–63.

9. Ozkan E, West A, Dedelow JA, et al. ct gray-level texture analysis as a quantitative imaging biomarker of epidermal growth factor receptor mutation status in adenocarcinoma of the lung. AJR Am J Roentgenol 2015;205:1016–25.
cross-ref  pubmed  

10. Lee JK, Hahn S, Kim DW, et al. Epidermal growth factor receptor tyrosine kinase inhibitors vs conventional chemotherapy in non–small cell lung cancer harboring wild-type epidermal growth factor receptor: a meta-analysis. JAMA 2014;311:1430–7.
cross-ref  pubmed  

11. Antonicelli A, Cafarotti S, Indini A, et al. egfr-targeted therapy for non–small cell lung cancer: focus on EGFR oncogenic mutation. Int J Med Sci 2013;10:320–30.
cross-ref  pmc  

12. Wakelee HA, Gadgeel SM, Goldman JW, et al. Epidermal growth factor receptor (EGFR) genotyping of matched urine, plasma and tumor tissue from non–small cell lung cancer (nsclc) patients (pts) treated with rociletinib [abstract 9001]. J Clin Oncol 2016;34:. [Available online at: https://meetinglibrary.asco.org/record/123411/abstract; cited 30 December 2017]

13. Reckamp KL, Melnikova VO, Karlovich C, et al. A highly sensitive and quantitative test platform for detection of nsclc EGFR mutations in urine and plasma. J Thorac Oncol 2016;11:1690–700.
cross-ref  pubmed  

14. Weber B, Meldgaard P, Hager H, et al. Detection of EGFR mutations in plasma and biopsies from non–small cell lung cancer patients by allele-specific pcr assays. BMC Cancer 2014;14:294.

15. Kim H, Chae KJ, Yoon SH, et al. Repeat biopsy of patients with acquired resistance to egfrtkis: implications of biopsy-related factors on T790M mutation detection. Eur Radiol 2018;28:861–8.

16. Liu J, Shen J, Yang C, et al. High incidence of EGFR mutations in pneumonic-type non–small cell lung cancer. Medicine (Baltimore) 2015;94:e540.

17. Hsu JS, Huang MS, Chen CY, et al. Correlation between EGFR mutation status and computed tomography features in patients with advanced pulmonary adenocarcinoma. J Thorac Imaging 2014;29:357–63.
cross-ref  pubmed  

18. Liu Y, Kim J, Qu F, et al. ct features associated with epidermal growth factor receptor mutation status in patients with lung adenocarcinoma. Radiology 2016;280:271–80.
cross-ref  pubmed  pmc  

19. Edge SB, Compton CC. The American Joint Committee on Cancer: the 7th edition of the ajcc cancer staging manual and the future of TNM. Ann Surg Oncol 2010;17:1471–4.
cross-ref  pubmed  

20. Travis WD, Brambilla E, Nicholson AG, et al. The 2015 World Health Organization classification of lung tumors: impact of genetic, clinical and radiologic advances since the 2004 classification. J Thorac Oncol 2015;10:1243–60.
cross-ref  pubmed  

21. Nakada T, Okumura S, Kuroda H, et al. Imaging characteristics in ALK fusion–positive lung adenocarcinomas by using hrct. Ann Thorac Cardiovasc Surg 2015;21:102–8.

22. Dearden S, Stevens J, Wu YL, Blowers D. Mutation incidence and coincidence in non small-cell lung cancer: meta-analyses by ethnicity and histology (mutMap). Ann Oncol 2013;24:2371–6.
cross-ref  pubmed  pmc  

23. El-Telbany A, Ma PC. Cancer genes in lung cancer: racial disparities: are there any? Genes Cancer 2012;3:467–80.
cross-ref  pubmed  pmc  

24. Ha SY, Choi SJ, Cho JH, et al. Lung cancer in never-smoker Asian females is driven by oncogenic mutations, most often involving EGFR. Oncotarget 2015;6:5465–74.
cross-ref  pubmed  pmc  

25. Hasegawa Y, Ando M, Maemondo M, et al. The role of smoking status on the progression-free survival of non–small cell lung cancer patients harboring activating epidermal growth factor receptor (EGFR) mutations receiving first-line egfr tyrosine kinase inhibitor versus platinum doublet chemotherapy: a meta-analysis of prospective randomized trials. Oncologist 2015;20:307–15.
cross-ref  pmc  

26. Kim MH, Kim HR, Cho BC, et al. Impact of cigarette smoking on response to epidermal growth factor receptor (egfr)–tyrosine kinase inhibitors in lung adenocarcinoma with activating EGFR mutations. Lung Cancer 2014;84:196–202.

27. Kitaguchi Y, Fujimoto K, Kubo K, Honda T. Characteristics of copd phenotypes classified according to the findings of hrct. Respir Med 2006;100:1742–52.
cross-ref  pubmed  

28. Zhou JY, Zheng J, Yu ZF, et al. Comparative analysis of clinicoradiologic characteristics of lung adenocarcinomas with ALK rearrangements or EGFR mutations. Eur Radiol 2015;25:1257–66.
cross-ref  pubmed  

29. Choi CM, Kim MY, Hwang HJ, Lee JB, Kim WS. Advanced adenocarcinoma of the lung: comparison of ct characteristics of patients with anaplastic lymphoma kinase gene rearrangement and those with epidermal growth factor receptor mutation. Radiology 2015;275:272–9.
cross-ref  pubmed  

30. Robertson WW, Steliga MA, Siegel ER, Arnaoutakis K. Accuracy of fine needle aspiration and core lung biopsies to predict histology in patients with non–small cell lung cancer. Med Oncol 2014;31:967.

31. De Filippo M, Saba L, Concari G, et al. Predictive factors of diagnostic accuracy of ct-guided transthoracic fine-needle aspiration for solid noncalcified, subsolid and mixed pulmonary nodules. Radiol Med 2013;118:1071–81.
cross-ref  pubmed  

32. Sabri A, Batool M, Xu Z, Bethune D, Abdolell M, Manos D. Predicting EGFR mutation status in lung cancer: proposal for a scoring model using imaging and demographic characteristics. Eur Radiol 2016;26:4141–7.
cross-ref  pubmed  

Correspondence to: Yiyuan Cao or Haibo Xu, Department of Radiology, Zhongnan Hospital of Wuhan University, Wuhan University, 169 Donghu Road, Wuhan, Hubei 430071 P.R.C. E-mail: caoyy@whu.edu.cn or haiboxu1120@hotmai.com

(Return to Top)

Current Oncology, VOLUME 25, NUMBER 2, April 2018

Copyright © 2018 Multimed Inc.
ISSN: 1198-0052 (Print) ISSN: 1718-7729 (Online)