Prediction of the benign and malignant nature of masses in COPD background based on Habitat-based enhanced CT radiomics modeling: A preliminary study

Abstract

BACKGROUND:

It is difficult to differentiate between chronic obstructive pulmonary disease (COPD)-peripheral bronchogenic carcinoma (COPD-PBC) and inflammatory masses.

OBJECTIVE:

This study aims to predict COPD-PBC based on clinical data and preoperative Habitat-based enhanced CT radiomics (HECT radiomics) modeling.

METHODS:

A retrospective analysis was conducted on clinical imaging data of 232 cases of postoperative pathological confirmed PBC or inflammatory masses. The PBC group consisted of 82 cases, while the non-PBC group consisted of 150 cases. A training set and a testing set were established using a 7:3 ratio and a time cutoff point. In the training set, multiple models were established using clinical data and radiomics texture changes within different enhanced areas of the CT mass (HECT radiomics). The AUC values of each model were compared using Delong’s test, and the clinical net benefit of the models was tested using decision curve analysis (DCA). The models were then externally validated in the testing set, and a nomogram of predicting COPD-PBC was created.

RESULTS:

Univariate analysis confirmed that female gender, tumor morphology, CEA, Cyfra21-1, CT enhancement pattern, and Habitat-Radscore B/C were predictive factors for COPD-PBC ( $P<$ 0.05). The combination model based on these factors had significantly higher predictive performance [AUC: 0.894, 95% CI (0.836–0.936)] than the clinical data model [AUC: 0.758, 95% CI (0.685–0.822)] and radiomics model [AUC: 0.828, 95% CI (0.761–0.882)]. DCA also confirmed the higher clinical net benefit of the combination model, which was validated in the testing set. The nomogram developed based on the combination model helped predict COPD-PBC.

CONCLUSION:

The combination model based on clinical data and Habitat-based enhanced CT radiomics can help differentiate COPD-PBC, providing a new non-invasive and efficient method for its diagnosis, treatment, and clinical decision-making.

Keywords

Nomograms retrospective studies antigen CYFRA21.1 area under curve radiomics tomography X-ray computed clinical decision-making pulmonary disease chronic obstructive

1. Introduction

With the progress of the global COVID-19 pandemic, the volume of lung digital radiography (DR), and computed tomography (CT) examinations has increased sharply, leading to a significant increase in the detection rate of lung cancer. At present, lung cancer is mainly divided into peripheral bronchogenic carcinoma (PBC) and central lung cancer (CLC), with CLC mostly located around the pulmonary hilum, accompanied by enlarged mediastinal lymph nodes and worsening respiratory symptoms, making it easier to distinguish. PBC is a more common lung tumor located below the third-level bronchus and above the respiratory bronchioles, with adenocarcinoma (95%), small cell lung cancer (3%), and large cell lung cancer (0.7%) being the main histological types. PBC, mostly originates from below the subsegmental bronchus, the peripheral zone of the lung, and beneath the pleura, making it in dilemma to differentiate from inflammatory masses in the background of chronic obstructive pulmonary disease (COPD) because of similar imaging features (irregular shapes, short spikes, etc.). With the worsening of atmospheric pollution, smoking addiction, accelerated industrial processes, and the discharge of biochemical waste, the incidence of COPD has been increasing year by year, reaching 8.4%. Meanwhile, COPD has a higher probability of concurrent various chronic inflammatory or PBC masses [1, 2]. Because China has a population of 1.4 billion people, the number of existing PBC cases is much higher than the world average. It has been reported that the malignant nature of PBC varies greatly, with a high rate of brain and adrenal gland metastasis. To date, it is still not possible to accurately predict the invasiveness and long-term prognosis of PBC, so early differentiation of COPD-PBC and surgical treatment is of great clinical significance. But due to the high number of tuberculosis patients in China, there are many cases of COPD combined with obsolete pulmonary tuberculosis masses, making it very difficult to differentiate these masses from PBC. Besides, COPD patients have poor lung function and tolerance, making it difficult to undergo traditional endobronchial ultrasound (EBUS) biopsy and CT-guided puncture, and frequent CT follow-up can lead to the overcrowding of public health resources and excessive radiation damage [3, 4, 5]. It has been reported that quantitative analysis of the imaging characteristics of locally advanced solid tumors can improve the surgical resection rate and survival benefits. Radiomics is an emerging tool for tumor diagnosis and prognosis prediction in recent years, and it has important value in the prognosis analysis of known lung nodules, liver and kidney tumors, etc. HECT radiomics is a comprehensive evaluation of the changes in imaging texture parameters in different enhanced areas of tumors on enhanced CT/MR or after neoadjuvant radiotherapy and chemotherapy, and it has important value in the treatment of liver, kidney, and pancreatic tumors [6, 7]. Therefore, in the accurate prediction of COPD-PBC, we have also introduced the relatively novel HECT radiomics, which improves the efficiency of the prediction model by analyzing the differences in radiographic texture associated with enhanced CT over time and region, providing strong data support for the screening and decision-making of high-risk populations for COPD-PBC. Our team found no related research has been reported in MEDLINE (Fig. 1).

Figure 1.

The results of quantitative reference analysis using the keyword “COPD-PBC” indicate that rare study was conducted on radiomics model for predicting COPD-PBC since 2003, but most studies have focused on the molecular mechanisms of PBC and chemotherapy.

2. Materials and methods

2.1 Clinical data

Retrospective analysis of clinical and enhanced CT data of 291 COPD patients confirmed by pathology as PBC or inflammatory masses from Xiangyang Hospital of Traditional Chinese Medicine from February 2018 to June 2023. Inclusion criteria: PBC or inflammatory masses confirmed by biopsy or postoperative pathology; chest enhanced CT examination within 1 month before surgery; primary PBC; COPD history of more than 3 years and peripheral lung masses larger than 2.5 cm; exclusion criteria: PBC recurrence or COPD combined with other tumors; cases with poor image quality (more scanning artifacts or frequent patient movement artifacts) or lost to follow-up (Suspected mass has been found but not followed up or biopsied, or PBC suspected through biopsy but not followed up with subsequent surgery, or PBC confirmed by biopsied pathology refuse to follow up and provide baseline data, etc.); patients with contrast agent allergies, etc. Clinical baseline data include age, BMI, diabetes, hypertension, alcohol use, smoking history, NLR, PLR, PBC Volume, PBC location, NSE, CA125, gender, tumor morphology, CEA, Cyfra21-1, CT enhancement pattern, etc. (Fig. 2) [8, 9].

Figure 2.

The enrolled inclusion and exclusion criteria and case grouping method of this research.

2.2 CT inspection method

Using Siemens Definition AS 320-layer spiral CT and Toshiba Aquiliond 64-row spiral CT for lung scanning. Preparations before the examination: introduce relevant precautions for the examination, train the patient’s breathing, and remove metal foreign objects within the scanning range. Examination method: the patient lies supine on the examination bed, with the head first, instruct the patient to take a deep breath and hold their breath, then perform chest CT scanning to collect raw data. Scanning parameters: Simens Definition AS: tube voltage 120 kV, tube current 280 mA, layer thickness 5 mm, spacing 5 mm, collimator width 0.6 mm, observation window level $-$ 700 Hu, window width 1500 Hu; Toshiba Aquiliond: tube voltage 120 kV, tube current 280 mA, layer thickness 5 mm, spacing 5 mm, collimator width 1.5 mm. The scanning range is from above the thoracic inlet to below the diaphragm, first performing plain scanning and then enhanced scanning. Important body parts that do not need to be scanned are covered with lead aprons to reduce radiation damage. Through a high-pressure injector, non-ionic contrast agent is injected into the patient’s elbow vein at a rate of 2.5–3.5 mL/s, with a dosage of 60–90 mL (calculated as 1.5 mL/kg of body weight). Arterial phase and venous phase scanning are performed at 20–30 s and 50–70 s after the injection of the contrast agent. After the scanning is completed, the obtained raw images (in DICOM format) are uploaded to the PACS system. The relationship between the lesion and the surrounding lung tissue is analyzed through multi-planar reconstruction using Siemens post-processing software [10, 11].

2.3 CT routine feature analysis

Two senior radiologists with over 15 years of work experience read the film by a double-blind method and negotiated a conclusion. The CT features include: tumor location (peripheral zone, subpleural or diaphragmatic side, etc.); tumor morphology (lobulated, round or irregular); tumor volume; growth pattern (expansive, infiltrative, satellite nodules or gas-containing necrosis, etc.); degree of enhancement and enhancement pattern (inflow, plateau or outflow). According to the different enhancement areas inside the enhanced CT mass and the surrounding environment, it is divided into areas of uneven enhancement (or weak enhancement), areas of uniform enhancement (or obvious enhancement), and tumor surrounding zone, and then HECT radiomics analysis is performed [12, 13].

2.4 CT image segmentation and HECT radiomics feature extraction

All enhanced CT images (plain scan, arterial and venous phase) of all lung nodules were exported as DICOM format and imported into 3Dslicer software for volume of interest (VOI) delineation and habitat feature extraction. VOIs were manually delineated by two experienced radiologists (8 and 15 years of experience) independently, without prior knowledge or consultation of pathological results. In the arterial phase, VOIs were delineated in the maximum axial image along the edge of the nodule, away from surrounding blood vessels, pleura, and fibrotic calcification tissue. Then, based on different enhancing regions within the nodule, VOIs were divided into VOIA, corresponding to heterogeneous enhancing areas (or weak enhancing areas), VOIB, corresponding to homogeneous enhancing areas (or significantly enhancing areas), and VOIC, corresponding to a 3–10 mm micro-invasive area around the tumor. Radiomic texture parameters were extracted using the 3Dslicer Radiomics module. First, the original data were resampled and normalized (using the Nearest Neighbor method for interpolation, with a resampling interval of 1 $\times$ 1 $\times$ 1). Then, filters (LoG, Wavelet, LBP) were selected to extract texture parameters and generate the Radscore. The consistency between two repeated VOIs was evaluated using the intra-class correlation coefficient (ICC) (ICC $\geqslant$ 0.7 indicates good consistency, and this study required ICC $>$ 0.75). Each habitat region of each nodule obtained 859 sets of features, resulting in a total of 859 $\times$ 3 sets of radiomic features. These features were then imported into the 3.0 version of the imaging research software from United Imaging Corporation (based on python3.7.7 $+$ Mysql 5.7 $+$ uni $+$ HBuilder X $+$ listpip $+$ Navicat 11 $+$ Django $+$ nodejs) for habitat radiomic calculation: valuable texture parameters were selected based on VOIA-VOIB-VOIC using the LASSO algorithm and ten-fold cross-validation to generate three sets of Radiomics Radscores (Radscore A: VOIA; Radscore B: VOIB; Radscore C: VOIC) (Fig. 3) [14, 15].

Figure 3.

The simple technology roadmap of PBC lesion delineation and radiomics extraction using enhanced CT in this research.

2.5 Statistical analysis

Using R 4.13 version and United Imaging Corporation research platform (https://www.uii-ai.com/) perform statistical analysis, all continuous variables are subjected to normal distribution and homogeneity of variance tests. Quantitative data that conform to normal distribution are represented by X $\pm$ S, and $t$ -tests are used for inter group comparisons; Non normally distributed measurement data are represented by ranges or median, and rank sum tests are used for inter group comparisons; The categorical variables are represented as examples (%), and inter group comparisons are conducted using chi square test or Fisher’s exact probability test. The training and testing sets was established by a 7/3 ratio and time cutoff. In the training set, the differences in clinical data and radiomics characteristics of the two groups were analyzed using logistic regression, and then multiple predictive models were established; the predictive performance of the model was compared using the area under curve (AUC) of receiver operating characteristic (ROC) using Delong’s test, and test it on the test set. Then the clinical net benefit of the model was evaluated using decision curve (DCA), and finally a nomogram and calibration curve was created using R Studio. $P<$ 0.05 indicates a statistically significant difference [16, 17].

3. Results

3.1 Comparison of clinical data and conventional CT image features

Among the 232 patients, there were 82 cases of PBC, 49 cases were female, and 33 cases were male, with an age range of (59.7 $\pm$ 5.7) years. The tumors were located in the peripheral zone in 39 cases, under the pleura in 19 cases, or next to the diaphragm in 24 cases. There were 150 cases in the non-PBC group, 61 cases were female, and 89 cases were male, with an age range of (56.6 $\pm$ 13.1) years. The tumors were located in the peripheral zone in 95 cases, under the pleura in 22 cases, or next to the diaphragm in 33 cases. There were statistical differences in gender, tumor morphology, CEA, Cyfra21-1, CT enhancement pattern, and Habitat-Radscore B/C between the two groups ( $P<$ 0.05), while there were no statistical differences in tumor volume, location, age, NSE, CA125, diabetes, hypertension, NLR, PLR, smoking and drinking history between the two groups ( $P>$ 0.05) (Tables 1 and 2).

Table 1
Comparison of clinical baseline data between PBC group and non-PBC group

Baseline characteristics	PBC group ( $N=$ 82)		Non-PBC group ( $N=$ 150)		$P$ value
Age	59.5	$\pm$ 6.08	59.9	$\pm$ 5.53	0.63
BMI	26.2	$\pm$ 5.3	25.2	$\pm$ 5.0	0.24
Diabetes	23/59		39/111		0.73
Hypertension	39/43		74/76		0.79
Alcohol use	35/47		78/72		0.17
Smoking history	27/55		35/115		0.11
FEV1/FVC	64.88	$\pm$ 9.42	66.03	$\pm$ 9.75	0.47
White blood cell count	7.86	$\pm$ 0.92	8.02	$\pm$ 1.06	0.35
Red blood cell count	4.14	$\pm$ 1.03	4.06	$\pm$ 1.17	0.66
Hemoglobin	120.6	$\pm$ 11.7	119.7	$\pm$ 11.5	0.68
Neutrophils	0.65	$\pm$ 0.10	0.63	$\pm$ 0.09	0.07
Lymphocyte	0.38	$\pm$ 0.13	0.35	$\pm$ 0.11	0.13

^*There was no significant difference in clinical baseline data between the two groups of patients.

Table 2

Logistic regression analysis results of clinical model based on clinical characteristics for predicting the COPD-PBC, ^{*P <} 0.05

Clinical model	Univariate analysis		Multivariate analysis
Factors	$P$	Hazard ratio	$P$	Hazard ratio
NLR	0.09	1.35 (0.96–1.89)
PLR	0.11	1.18 (0.96–1.46)
PBC volume	0.12	1.01 (0.99–1.02)
PBC location	0.15	1.37 (0.89–2.09)
NSE	0.11	1.08 (0.98–1.19)
CA125	0.22	0.98 (0.96–1.01)
Gender	0.04^*	1.96 (1.03–3.77)
Tumor morphology	0.04^*	1.35 (1.01–1.82)
CEA	0.04^*	1.29 (1.01–1.68)	0.04^*	1.35 (1.01–1.81)
Cyfra21-1	0.01^*	3.81 (1.70–8.51)	0.01^*	3.10 (1.35–7.11)
CT enhancement pattern	0.01^*	1.98 (1.42–2.78)	0.01^*	1.80 (1.25–2.61)

^*There were statistically significant differences in CEA, Cyfra21-1 and CT enhancement pattern between the two groups using multivariate logistic regression ( $P<$ 0.05).

Table 3

Logistic regression analysis results of radiomics model based on MR characteristics for predicting the COPD-PBC, ^{*P <} 0.05

Radiomics model	Univariate analysis		Multivariate analysis
Factors	$P$	Hazard ratio	$P$	Hazard ratio
Radscore B	0.01^*	4.79 (2.28–10.08)	0.01^*	4.24 (1.87–9.64)
Radscore C	0.01^*	2.59 (1.83–3.64)	0.01^*	2.52 (1.76–3.61)

^*There were statistically significant differences in Radscore A/B between the two groups using multivariate logistic regression ( $P<$ 0.05).

Table 4

Logistic regression analysis results of combined model based on valuable factors (clinical data: gender, tumor morphology, CEA, Cyfra21-1; imaging data: CT enhancement pattern, Radscore B/C) mentioned above for predicting the COPD-PBC, ^{*P <} 0.05

Combined model	Univariate analysis		Multivariate analysis
Factors	$P$	Hazard ratio	$P$	Hazard ratio
Gender	0.04^*	1.96 (1.03–3.77)
Tumor morphology	0.04^*	1.35 (1.01–1.82)
CEA	0.04^*	1.29 (1.01–1.68)
Cyfra21-1	0.01^*	3.81 (1.70–8.51)	0.01^*	5.02 (1.86–13.50)
CT enhancement pattern	0.01^*	1.98 (1.42–2.78)	0.02^*	1.70 (1.09–2.66)
Radscore B	0.01^*	4.79 (2.28–10.08)	0.01^*	4.28 (1.66–11.06)
Radscore C	0.01^*	2.59 (1.83–3.64)	0.01^*	2.83 (1.83–4.37)

^*There were statistically significant differences in Cyfra21-1, CT enhancement pattern and Radscore B/C between the two groups using multivariate logistic regression ( $P<$ 0.05).

Figure 4.

Delong nonparametric curves of the training set (a) and the test set (b). The area under the ROC curve of the combined model of the two groups is the largest; which confirms that the combined model has the best predictive performance. Clinical data model (A); Radiomics model (B) and Combined model (C).

3.2 Habitat-Radiomics feature selection and establishment of multiple models

Radscore A was found to be more difficult to delineate in both groups, with poor consistency (ICC: 0.69) and no significant differences ( $P=$ 0.08), so it was discarded. Radscore B/C showed statistically significant differences in both groups and had higher predictive value ( $P<$ 0.05). In the training set, a clinical model, radiomics model, and combination model were established using the valuable clinical data and habitat-Radiomics Radscore mentioned above. Delong’s test confirmed that the combined model had higher predictive performance [AUC: 0.894, 95% CI (0.836–0.936)] than the clinical model [AUC: 0.758, 95% CI (0.685–0.822)] and radiomics model [AUC: 0.828, 95% CI (0.761–0.882)]. The same results were validated in the test set [AUC: the combined model-0.881, the clinical model-0.727, the radiomics model-0.806]. DCA also confirmed that the combination model had higher clinical net benefit. The nomogram and calibration curve generated based on the combination model obtained a good clinical application evaluation (Figs 4–6 and Tables 3 and 4).

Figure 5.

The higher clinical net benefits of the combined model was confirmed in the two groups by DCA of training set (a) and test set (b) using R software.

Figure 6.

The nomogram prediction tool based on the risk factors of the combined model was used clinically (a. nomogram, b. calibration). Namely, each risk factor is scored, added together, and the final risk value is calculated.

4. Discussion

PBC is the most common malignant tumor of the lungs in China. Early diagnosis and timely surgery significantly benefit patients, with a higher five-year survival rate. COPD is a common respiratory disease characterized by respiratory symptoms and airflow limitation, and it has a high global disease burden. COPD patients have low immunity and some lung interstitial fibrosis, making them susceptible to bacterial, tuberculous and viral pneumonia (tumour-like lesion), which can be prolonged and difficult to heal. In addition, the formation process of tumors in the context of COPD is complex, making it difficult to distinguish between benign and malignant tumors. Conventional imaging diagnostic characteristics and methods (enlarged mediastinal lymph nodes, satellite lesions, pleural traction, vascular penetration, etc.) are difficult to use for differentiation. In clinical practice, when suspicious masses are found in COPD patients, respiratory physicians often recommend frequent CT scans or biopsies, which significantly increase the pressure on patients’ psychology and family economy [18, 19, 20, 21]. Non-invasive radiomics techniques play an important role in the diagnosis and risk stratification of malignant tumors, and have been successfully applied in the artificial intelligence identification of lung malignancies and breast cancer. Radiomics converts high-throughput image features into high-dimensional exploitable feature space data through automated analysis, providing valuable information for disease diagnosis and prognosis prediction. Habitat-based Radiomics can obtain radiographic texture change data in different enhancement areas of high-quality tumors, indirectly reflecting the nature, blood flow perfusion, and malignancy of the tumor, providing strong data support for the diagnosis and treatment of PBC [22, 23, 24, 25]. Therefore, the habitat-based radiomics based prediction tool developed in this study has important clinical value for the differentiation of benign and malignant tumors in COPD. Our team found the combined model generated based on clinical data-habitat radiomics score has high predictive efficiency, reaching [AUC: 0.894, 95% CI (0.836–0.936)]; Cyfra21-1 [AUC: 0.668, 95% CI (1.70–8.51)], Radscore B [AUC: 0.706, 95% CI (2.28–10.08)], and Radscore C [AUC: 0.790, 95% CI (1.83–3.64)] present higher diagnostic weight, which helps to improve the precise diagnosis of COPD-PBC and change clinical decision-making.

Based on the summary of previous studies, this study newly discovered that factors such as gender, tumor morphology, CEA, Cyfra21-1, CT enhancement pattern, and Radscore B/C are influencing factors for predicting COPD-PBC. In China, the proportion of male smokers and drinkers is high, and the working environment is poor with high levels of stress. Generally, the proportion of COPD in males is higher than females, but the symptoms of COPD in females are usually more severe than in males. Additionally, females are more likely to be affected by secondhand smoke. If a female has COPD without obvious inducements and a tumor found, the more possibility of malignancy should be considered. Tumor morphology is a classic indicator for evaluating malignancy potential, with irregular shape, unclear boundaries, short spines, and pleural traction indicating a higher degree of malignancy, as well as higher rates of metastasis and recurrence. CEA and Cyfra21-1 are classic indicators for diagnosing PBC, and elevated levels in COPD patients not only indicate the degree of malignancy of the tumor but also affect postoperative recovery [26, 27, 28, 29]. In this study, CT enhancement pattern also has a high feature weight coefficient. COPD-PBC tends to grow rapidly, have a larger volume, uneven density, and may have hollow necrotic areas. The tumor may have internal blood vessels that are unevenly distributed, leading to obvious outflow-type enhancement, which usually indicates a higher degree of malignancy and the possibility of malignancy. Habitat-Radscore B/C is a predictive factor for COPD-PBC discovered for the first time in this study. This high-quality data based on the changes in different enhancement areas of enhanced CT helps analyze more details of the tumor and enhances tumor heterogeneity. It has been reported that dynamic time-enhanced curves help differentiate the malignancy potential of tumors, but there may be errors in interpreting the curves and defining the enhanced area by the human eye. However, tumor living environment radiomics has unique diagnostic advantages. It uses the differences in radiomics characteristics of different enhanced areas within the tumor to provide high-throughput information to quantify the tumor’s rich blood supply growth area and composition, thereby revealing the malignancy potential of the tumor and guiding clinical decision-making. Currently, Habitat-CT or MR radiomics has been proven to predict the survival period of ovarian tumors, which is significantly better than the predictive efficacy of static radiomics [30, 31, 32]. Therefore, this study used tumor living environment-CT Radscore to model the prediction of COPD-PBC, and the predictive efficacy was higher than that of previous research models, possibly because it provided high-throughput detailed imaging information of COPD-PBC and quantified the nature of the tumor.

4.1 Limitations

This is a single center study, with limited case data and possible selection bias in the results. More prospective data (such as pathological subtypes, ki67, etc.) are needed to further validate the predictive model’s effectiveness externally; In addition, this study will require multi center participation and multiple machine validation models in the future to verify the accuracy of the results. Our team is also preparing to use pathomics and artificial intelligence to enrich this research.

5. Conclusion

In conclusion, the clinical-radiomics prediction model based on the habitat-CT Radscore has high clinical value in the diagnosis of COPD-PBC. It may reduce unnecessary punctures and surgeries, and provide data support for subsequent treatment decision changes and follow-up.

Ethics statement

The experimental protocol was established according to the ethical guidelines of the Declaration of Helsinki and was approved by the Human Ethics Committee of Xiangyang Hospital of Traditional Chinese Medicine (Issue no. 1302X [2017]). Written informed consent was obtained from individual or guardian participants.

Availability of data and materials

All data generated or analysed during this study are included in this published article.

Competing interests

The authors declare no conflicts of interest.

Author contributions

WanZhao Zuo and Jing Li conceived and drafted the manuscript. MingYan Zuo contributed to the literature review. Miao Li and Xing Cai are responsible for the quality control of the statistics. Shuang Zhou and Xing Cai revised the manuscript critically for important intellectual content. Xing Cai approved the final version to be published and agreed to act as guarantor of the work. WanZhao Zuo, Jing Li and MingYan Zuo contributed equally to this work.

Funding

The study was supported by the “323” Public Health Project of the Hubei Health Commission and Xiangyang Hospital of Traditional Chinese Medicine (XYY2022-323).

Footnotes

Acknowledgments

The authors gratefully acknowledge Miao Li and Shuang Zhou for their assistance with translating the references and the application of the improved statistical technology.

References

Sun

Xiong

. From COPD to lung cancer: Mechanisms linking, diagnosis, treatment, and prognosis. Int J Chron Obstruct Pulmon Dis. 2022; 17: 2603-2621. Published 2022 Oct 17. doi: 10.2147/COPD.S380732.

Teng

Xia

Cao

, et al. Lung cancer burden and trends from 2000 to 2018 in China: Comparison between China and the United States. Chin J Cancer Res. 2023; 35(6): 618-626. doi: 10.21147/j.issn.1000-9604.2023.06.06.

Hou

, et al. Cigarette Smoke Induced Lung Barrier Dysfunction, EMT, and Tissue Remodeling: A Possible Link between COPD and Lung Cancer. Biomed Res Int. 2019; 2019: 2025636. Published 2019 Jun 24. doi: 10.1155/2019/2025636.

Mascalchi

Luconi

. Lung cancer screening, emphysema, and COPD. Chest. 2021; 159(5): 1699-1700. doi: 10.1016/j.chest.2021.01.040.

Zhou

Xiang

Lin

, et al. Long-term trends of lung cancer incidence and survival in southeastern China, 2011–2020: A population-based study. BMC Pulm Med. 2024; 24(1): 25. Published 2024 Jan 10. doi: 10.1186/s12890-024-02841-0.

Refaee

Ibrahim

, et al. The emerging role of radiomics in COPD and lung cancer. Respiration. 2020; 99(2): 99-107. doi: 10.1159/000505429.

Wang

Karalis

Elamir

, et al. Delta Radiomic Features Predict Resection Margin Status and Overall Survival in Neoadjuvant-Treated Pancreatic Cancer Patients. Ann Surg Oncol. Published online December 27, 2023. doi: 10.1245/s10434-023-14805-5.

Inoue

Johnson

Walkoff

, et al. Lung cancer screening using clinical photon-counting detector computed tomography and energy-integrating-detector computed tomography: A prospective patient study. J Comput Assist Tomogr. 2023; 47(2): 229-235. doi: 10.1097/RCT.0000000000001419.

Tang

Ramis-Cabrer

Curull

, et al. Markers of stroma in lung cancer: Influence of COPD. Arch Bronconeumol (Engl Ed). 2021; 57(2): 130-137. doi: 10.1016/j.arbres.2020.09.005.

10.

Wang

Liu

Wang

, et al. A prognostic model of non-small cell lung cancer with a radiomics nomogram in an eastern chinese population. Front Oncol. 2022; 12: 816766. Published 2022 Jun 14. doi: 10.3389/fonc.2022.816766.

11.

Zhou

Dong

, et al. CT-based radiomic nomogram for the prediction of chronic obstructive pulmonary disease in patients with lung cancer. Acad Radiol. 2023; 30(12): 2894-2903. doi: 10.1016/j.acra.2023.03.021.

12.

Szigeti

Szabó

Korom

, et al. Radiomics-based differentiation of lung disease models generated by polluted air based on X-ray computed tomography data. BMC Med Imaging. 2016; 16: 14. Published 2016 Feb 11. doi: 10.1186/s12880-016-0118-z.

13.

Chen

Xiong

Zhou

, et al. CT radiomics-based long-term survival prediction for locally advanced non-small cell lung cancer patients treated with concurrent chemoradiotherapy using features from tumor and tumor organismal environment. Radiat Oncol. 2022; 17(1): 184. Published 2022 Nov 16. doi: 10.1186/s13014-022-02136-w.

14.

Wei

Zhang

Shi

, et al. The value of CT radiomics features to predict visceral pleural invasion in ⩽ 3 cm peripheral type early non-small cell lung cancer. J Xray Sci Technol. 2022; 30(6): 1115-1126. doi: 10.3233/XST-221220.

15.

Wang

Tang

Chen

, et al. Joint use of the radiomics method and frozen sections should be considered in the prediction of the final classification of peripheral lung adenocarcinoma manifesting as ground-glass nodules. Lung Cancer. 2020; 139: 103-110. doi: 10.1016/j.lungcan.2019.10.031.

16.

Sha

Zhang

, et al. An MRI-based radiomics nomogram for detecting cervical esophagus invasion in hypopharyngeal squamous cell carcinoma. Cancer Imaging. 2023; 23(1): 120. Published 2023 Dec 15. doi: 10.1186/s40644-023-00642-y.

17.

Zhang

Luan

Liu

, et al. Comparison of MRI radiomics-based machine learning survival models in predicting prognosis of glioblastoma multiforme. Front Med (Lausanne). 2023; 10: 1271687. Published 2023 Nov 30. doi: 10.3389/fmed.2023.1271687.

18.

Wei

Chen

Liu

, et al. Outcomes of lobectomy on pulmonary function for early stage non-small cell lung cancer (NSCLC) patients with chronic obstructive pulmonary disease (COPD). Thorac Cancer. 2020; 11(7): 1784-1789. doi: 10.1111/1759-7714.13445.

19.

Young

Scott

. Should we be screening for COPD? – looking through the lens of lung cancer screening. Expert Rev Respir Med. 2023; 17(9): 753-771. doi: 10.1080/17476348.2023.2259800.

20.

Roy

Rheault

Pigeon

, et al. Lung cancer resection and postoperative outcomes in COPD: A single-center experience. Chron Respir Dis. 2020; 17: 1479973120925430. doi: 10.1177/1479973120925430.

21.

Kridin

Comaneshter

Batat

Cohen

. COPD and lung cancer in patients with pemphigus – a population based study. Respir Med. 2018; 136: 93-97. doi: 10.1016/j.rmed.2018.02.005.

22.

Warkentin

Al-Sawaihey

Lam

, et al. Radiomics analysis to predict pulmonary nodule malignancy using machine learning approaches. Thorax. Published online January 9, 2024. doi: 10.1136/thorax-2023-220226.

23.

Jiang

Zeng

Zuo

, et al. Leveraging multimodal MRI-based radiomics analysis with diverse machine learning models to evaluate lymphovascular invasion in clinically node-negative breast cancer. Heliyon. 2023; 10(1): e23916. Published 2023 Dec 18. doi: 10.1016/j.heliyon.2023.e23916.

24.

Wang

Liu

, et al. Habitat-based radiomics enhances the ability to predict lymphovascular space invasion in cervical cancer: A multi-center study. Front Oncol. 2023; 13: 1252074. Published 2023 Oct 26. doi: 10.3389/fonc.2023.1252074.

25.

Verma

Hill

Statsevych

, et al. Stable and discriminatory radiomic features from the tumor and its habitat associated with progression-free survival in glioblastoma: A multi-institutional study. AJNR Am J Neuroradiol. 2022; 43(8): 1115-1123. doi: 10.3174/ajnr.A7591.

26.

Carr

Jacobson

Lynch

, et al. Features of COPD as predictors of lung cancer. Chest. 2018; 153(6): 1326-1335. doi: 10.1016/j.chest.2018.01.049.

27.

V A B Subramoniam

Mathew

. Detection of COPD and Lung Cancer with electronic nose using ensemble learning methods. Clin Chim Acta. 2021; 523: 231-238. doi: 10.1016/j.cca.2021.10.005.

28.

Zhang

Zhao

, et al. Molecular and clinicopathological characteristics of lung cancer concomitant chronic obstructive pulmonary disease (COPD). Int J Chron Obstruct Pulmon Dis. 2022; 17: 1601-1612. Published 2022 Jul 14. doi: 10.2147/COPD.S363482.

29.

Yang

Zeng

Chen

, et al. Multi-Layer Perceptron Classifier with the Proposed Combined Feature Vector of 3D CNN Features and Lung Radiomics Features for COPD Stage Classification. J Healthc Eng. 2023; 2023: 3715603. Published 2023 Nov 3. doi: 10.1155/2023/3715603.

30.

Cai

Wang

Zhou

, et al. CT-based radiomics nomogram for predicting visceral pleural invasion in peripheral T1-sized solid lung adenocarcinoma. Am J Cancer Res. 2023; 13(12): 5901-5913. Published 2023 Dec 15.

31.

Zou

Mao

Zhao

, et al. Intratumoural and peritumoural CT-based radiomics for diagnosing lepidic-predominant adenocarcinoma in patients with pure ground-glass nodules: A machine learning approach. Clin Radiol. 2024; 79(2): e211-e218. doi: 10.1016/j.crad.2023.11.003.

32.

Zhang

Chen

You

. Decoding intratumoral heterogeneity: Clinical potential of habitat imaging based on radiomics. Radiology. 2023; 309(3): e232047. doi: 10.1148/radiol.232047.

Prediction of the benign and malignant nature of masses in COPD background based on Habitat-based enhanced CT radiomics modeling: A preliminary study

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1. Introduction

2.1 Clinical data

2.3 CT routine feature analysis

2.4 CT image segmentation and HECT radiomics feature extraction

3. Results

3.1 Comparison of clinical data and conventional CT image features

Table 1 Comparison of clinical baseline data between PBC group and non-PBC group

4.1 Limitations

5. Conclusion

Ethics statement

Availability of data and materials

Competing interests

Author contributions

Funding

Footnotes

Acknowledgments

References

Table 1
Comparison of clinical baseline data between PBC group and non-PBC group