A narrative review on CT-based evaluation and prediction of lung nodule growth: current status and future directions

Introduction

Background

In 2022, 2.48 million new cases of lung cancer were diagnosed, accounting for 12.4 per cent of all new cases of cancer globally, regaining its position as the most common cancer worldwide. Lung cancer is also the leading cause of cancer deaths, with an estimated 1.8 million deaths from lung cancer, accounting for 18.7 per cent of all global cancer deaths in 2022 (1). The predominant reason for the high mortality rate associated with lung cancer is that it is often detected at an advanced stage. Therefore, early diagnosis and treatment are crucial. Studies have shown that early diagnosis can significantly increase the 5-year relative survival rate from 6% for advanced lung cancer to 33% for intermediate-stage lung cancer and 60% for early-stage lung cancer (2). The increased use of computed tomography (CT) has significantly improved the detection of lung nodules (3). Although 95% of lung nodules are benign, they are still the main manifestation of early lung cancer (4). Therefore, the primary aim of lung nodule diagnosis is to differentiate the benignity, malignancy and invasiveness of lung nodules as early as possible, to achieve early diagnosis and treatment of lung cancer, and to improve patients’ prognoses while reducing the phenomenon of overdiagnosis and treatment.

Rationale and knowledge gap

As shown in Figure 1, one of the recognized critical features associated with the probability of lung cancer is the growth rate of lung nodules, which reflects the potential malignancy of the tumor (5). The growth patterns of lung nodules vary between histological types. Benign pathologic types typically include focal interstitial fibrosis, inflammatory exudates, and pneumonitis, many of which resolve within three months with follow-up or anti-inflammatory treatment (6,7). In contrast, persistent pure ground-glass nodules often represent precancerous lesions of invasive adenocarcinoma, involving gradually into part-solid nodules with an increased risk of minimally invasive or invasive adenocarcinoma. Malignant subsolid nodules (SSNs) generally show slower growth than solid nodules. In a meta-analysis of 2,898 SSNs, the combined growth incidence was 22% (8). Among subsolid pulmonary nodules, pure ground-glass nodules demonstrate a more indolent growth pattern, with median growth periods significantly differing at 7.7 years compared to 2.0 years for part-solid nodules, based on a study including 361 SSNs with long-term follow-up (9). Even after long-term stabilization, pure ground-glass nodules may still perform interval growth and carry the potential for progression to lung adenocarcinoma (10). It is due to the complexity of lung nodule growths that while guidelines exist for defining lung nodule growth and recommending follow-up programs, these guidelines may not comprehensively cover all types of lung nodule growth, potentially leading to over-screening, patient anxiety, unnecessary radiation exposure, and even over-treatment (11-14), or misdiagnosis of fast-growing and spaced-growing nodules, resulting in delayed treatment (15). Additionally, there is variability among radiologists in image interpretation, and clinicians may subjectively stratify patients during diagnosis and treatment, with low adherence to expert consensus or guideline recommendations for nodule management (16).

Figure 1 A 65-year-old male patient. Chest CT images and the final pathological result image. The dates of the imaging examinations corresponding to (A-F) are (A) 2018-01-18; (B) 2019-04-22; (C) 2020-04-13; (D) 2021-02-26; (E) 2021-12-02; (F) 2022-08-13. Staining method: hematoxylin-eosin. CT, computed tomography.

Objective

Therefore, there is an urgent and significant clinical need for a comprehensive exploration of nodule imaging data, accurate assessment, and prediction of changes in lung nodule growth to facilitate precise risk stratification. As the predominant imaging modality for screening, diagnosis and monitoring lung nodules, CT offers unparalleled advantages. This article reviewed current research progress in CT-based imaging methodologies for evaluating lung nodule growth. This article reviews the research progress of CT-based evaluation and growth prediction models for pulmonary nodules. We present this article in accordance with the Narrative Review reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0344/rc).

Methods

Literature search

A comprehensive literature search was performed in the Web of Science, PubMed, Cochrane Library, and EMBASE databases to identify relevant studies. The search strategy aimed to retrieve articles related to pulmonary nodule growth and CT-based diagnosis, using a combination of keywords and synonyms, including “pulmonary nodule”, “lung nodule”, “solid nodule”, “part-solid nodule”, and “subsolid nodule”. These terms were combined with “growth” or “follow-up”, as well as “CT” or “computed tomography”. Medical Subject Headings (MeSH) terms were applied where appropriate to enhance search sensitivity. The search was limited to original research articles and meta-analyses published in English between January 1, 2020 and February 3, 2026. Detailed information is provided in Table 1.

Discussion

Definition of lung nodule growth

Lung nodule growth is predominantly defined as an increase in diameter or volume observed on CT examination images across multiple time points. Changes in nodule density or mass can also be used to indicate the growth of pulmonary nodules, especially SSNs. Definitions of lung nodule growth in guidelines and studies are presented in Table 2.

Definition of growth based on diameter

Regarding diameter measurements, Fleischner 2017 (35) and the 2015 British Thoracic Society guideline (36) specify that an increase of two mm in lung nodule diameter across two CT scans indicates actual nodule growth. Lung-RADS [2022] (37) and NCCN [2023] (38) define lung nodule growth as an increase in nodule diameter exceeding 1.5 mm across two CT scans. Advances in CT resolution and measurement tools have reduced measurement variability, improving accuracy and reproducibility. Fleischner 2017 specifies that results should be rounded to single digits, while Lung-RADS 2022 specifies accuracy to one decimal place. That is, from the perspective of both guidelines, it is reasonable to report growth when the diameter increases by 1.5 mm. While diameter-based measurements achieve appropriate accuracy, they are most applicable to nodules with round or round-like shapes and regular nodules.

Definition of growth based on volume

According to the 2025 European Society of Thoracic Imaging (ESTI) nodule management recommendations, volume is the preferred size measure for pulmonary nodules (21). Manual diameter measurements for nodal assessment are prone to be influenced by nodal margins. In contrast, semi-automated volumetric measurements largely mitigate this effect (34,39). Moreover, the progression of subsolid pulmonary nodules does not all manifest as a diameter increase alone. In a study by Qi et al., 110 pure ground-glass nodules showed various growth patterns: 28 pure ground-glass nodules performed increases in both diameter and volume, while 24 pure ground-glass nodules showed increases in volume only (22). Assessing the growth of indeterminate nodules via volumetric screening has demonstrated higher sensitivity than diameter measurement (24), suggesting a preference for semi-automated volumetric methods in determining nodule size and growth.

Lung nodule volumetry and volume doubling time (VDT) are integral to management protocols such as those in the NELSON (40), Danish Lung Cancer Screening Trial (17), UK Lung Cancer Screening Trial (19), Pittsburg Lung Screening studies (18), Hungary screening program (25), and Tianjin Lung Cancer Screening (20). Fleischner 2017 reports the promising use of volumetric methods in assessing lung nodule growth. The British Thoracic Society nodule management program defines nodule growth as a 25% increase in nodule volume across two CT images, and a VDT of fewer than 400 days triggers further investigation of indeterminate lung nodules. Advances in diagnostic systems with artificial intelligence (AI) have improved measurement accuracy and consistency in volumetric assessments, potentially lowering the threshold for defining lung nodule growth. Qi et al. measured a 95% confidence interval (CI) of −5.34% to 13.08% for AI volume measurements, so a 20% increase in volume was considered to represent the growth of pure ground-glass nodule (22). Bartlett et al. studied 100 small (volume <150 mm³) non-metastatic pulmonary nodules and proposed a 15% volume increase threshold as an indicator of actual growth (26), showcasing higher variability in optimal volume thresholds across different nodule types and sizes. VDT, the time required for a nodule to increase in volume by a factor of one, can be calculated based on screening intervals and size variations, and growth can be characterized in terms of growth rate rather than just a fixed diameter or volume threshold, with the following formula:

VDT=t×log2/log(V2/V1)[1]

t, interval between two CT examinations; V, nodal volume; V1, nodal volume of the first CT examination; V2, nodal volume of the second CT examination.

Mottram first proposed the concept of exponential growth of cancer at a constant rate in 1935 (41). Studies have demonstrated that the exponential model can effectively represent overall growth and solid component expansion in lung nodules, supporting the rationality of using VDT to assess nodule growth (42). Moreover, VDT has proven valuable in distinguishing between benign and malignant lung nodules, differentiating histologic subtypes of lung adenocarcinoma, and evaluating patient prognosis (31,43). Horeweg et al., analyzing 9,681 non-calcified nodules from 7,155 participants in the NELSON Screening Group undergoing CT screening, found significant differences in lung cancer risk based on VDT categories (≤400, 400–600, and ≥600 days), with risks of 9.9%, 4.0%, and 0.8%, respectively for nodules sized 100–300 mm³ (44). Although 400 days is a commonly used VDT threshold, it is not absolute. Yankelevitz et al. Classified VDTs as 60—fast, 120—moderate, and 240—slow (29). Jiang et al. categorized peripheral non-small cell lung cancer (NSCLC) into fast and slow-growing groups based on a 60-day VDT (30). However, studies have shown that the exponential model may not fully represent the growth patterns of malignant nodules. There is an overlap in VDT values between benign and malignant nodules, limiting its utility in distinguishing malignant growth (23).

Definition of growth based on mass

Mass measurements reflect both the volume and density of lung nodules. In SSNs, progression can manifest solely as increased density or the emergence of solid components, critical indicators for evaluating SSNs growth. The Fleischner Society 2017 Guidelines suggest consideration of resection if a solid component develops. Furthermore, Lung-RADS 2022 recommends that the emergence of solid components within pure ground-glass nodules shifts nodules from category 2 to 4A/B. de Hoop et al. demonstrated that mass measurements of lung nodules can detect ground-glass nodules growth earlier and with less variability than volume or diameter measurements (45). Changes in the growth of SSNs are better characterized by mass measurements.

A 2022 study by Liao et al. showed that the rate of mass increase sensitively reflected the growth of SSNs associated with lung cancer compared to diameter and volume increases. When mass increase exceeds a 25% threshold, SSNs perform significant growth and are more likely to progress to lung cancer (27). Shi et al. define lung nodule growth as an increase of at least 30% in nodule volume or mass (46). Mass doubling time (MDT) has also proven valuable in predicting tumor invasiveness and pathological types (5). Although assessing changes in the mass of pulmonary nodules can reflect concomitant variations in both volume and density, there remains a lack of well-established criteria defining thresholds for significant mass increase or MDT. Consequently, its clinical application remains limited.

M=V×(A+1,000)/1,000[2]

Thus, in the measurement of pulmonary nodules, diameter-based assessment remains the mainstream method recommended by current guidelines and the most fundamental measurement approach, owing to its operational simplicity and strong feasibility. Volume measurement can be achieved through various methods: initially relying on formula-based calculations, it has now evolved with the development of AI to obtain volumetric parameters by voxel-wise summation across serial sections. This AI-based approach offers greater accuracy than formula-based calculations and is particularly suitable for irregularly shaped nodules. Mass measurement further enhances the precision of assessing pulmonary nodule growth, offering unique advantages especially for SSNs. When a nodule exhibits changes only in density without significant alterations in diameter or volume, mass change can sensitively capture such early and subtle signs of growth. However, this method has not yet been widely adopted in clinical practice, primarily because its implementation is hindered by the lack of authoritative guideline support, despite exploratory studies having been conducted. Furthermore, it should be noted that differences in scanning parameters and equipment from different manufacturers may influence the above measurement results. Moreover, whether these subtle changes cause patient anxiety or lead to over-screening and overtreatment is also a concern that needs to be addressed in clinical practice.

Risk factors for lung nodule growth from an imaging perspective

In clinical practice, the most commonly used and fundamental approach for radiologists to assess lung nodule growth remains the use of conventional CT imaging features. CT signs associated with lung nodule growth are summarized in Table 3.

Nodule size

Nodules detected at baseline may have been present for years, while newly formed nodules after baseline may grow rapidly. Therefore, the size of nodules at detection can estimate their growth rate (47). Qi et al. analyzed 110 patients with pure ground-glass nodules, employing log-rank test and Cox proportional hazards regression to identify risk factors. They concluded that larger initial diameters, volumes, and masses of pure ground-glass nodules were associated with an increased likelihood of growth (22). In 2023, the meta study by Wu et al. indicated that an initially large SSNs size was found to be a risk factor affecting the incidence of SSNs growth (8).

Nodule density

The density of pulmonary nodules, expressed as CT values (mean CT value, SD of CT values) or solid component size [consolidation-to-tumor ratio (CTR)], can be used to assess and predict the growth of pulmonary nodules. He et al. found that increased lung nodule density had a higher proportion of early growth and a faster growth rate in growing nodules compared to size measurements (28). Besides, CT values indicate the extent to which the tumor has replaced lung airspace and aid in predicting the natural growth of SSNs (48). Chen et al. further conducted a prospective study demonstrating that among 250 pathologically confirmed solid nodules of NSCLC, CT attenuation value served as a key predictor of rapid growth in solid nodules (32). The study by He et al. also corroborates this finding (49). The progression of lung cancer correlates significantly with the proportion of its non-solid components. Non-solid portions of nodules typically indicate adenocarcinoma in situ (AIS) or minimally invasive adenocarcinoma (MIA), whereas solid portions usually reflect invasive components. Xia et al. found that part-solid nodules with CTR ≥50% were the most common cause of continued growth in residual lung nodules following surgery in elderly patients with multifocal ground-glass nodules (GGNs) (50).

Tumor characteristics and peritumoral signs

Tumor characteristics and peritumoral signs are associated with lung nodule growth. Xia et al. (50) investigated risk factors influencing the growth of residual nodules (RNs) in 238 surgical patients with GGNs. Their findings indicated that RN with higher roundness and the presence of vascular convergence sign were more likely to exhibit growth. Xue et al. (52) demonstrated that the tumor blood vessel diameter of ground-glass opacity (GGO) lesions was closely associated with the growth of malignant pulmonary nodules. A meta-analysis by Liang et al. (51), encompassing 19 original studies involving 2,444 patients with 3,012 SSNs, revealed that air bronchogram, well-defined borders, and lobulated margins were significantly associated with the growth of pure ground-glass nodules. Specifically, lobulated sign has been consistently linked to tumor growth in several studies, potentially reflecting uneven or environment-constricted tumor growth. Moreover, lobulated sign often indicated larger nodules, indirectly suggesting nodule size information. Furthermore, Chen et al. (32) conducted a prospective study demonstrating that among 250 pathologically confirmed solid nodules of NSCLC, deep lobulation served as a key predictor of rapid growth in solid nodules.

Predicting lung nodule growth

Driven by advances in AI, radiomics and deep learning approaches have greatly advanced the development of predictive models for lung nodule growth. The performance of the relevant predictive models is presented in Table 4.

Predicting lung nodule growth—radiomics

Traditionally, lung cancer has been understood to evolve according to an exponential growth model. However, emerging evidence suggests that the natural course of lung cancer is more complex, influenced by interactions between stem cells, tumor micro-environments.

Radiologists have traditionally relied on subjective visual assessment of tumor images, but this relies on factors such as an individual’s diagnostic level and image resolution, and there is a need for a more objective and comprehensive way to further assess tumor growth. Radiomics, introduced by Philippe Lambin in 2011, addressed this by enabling automated, reproducible analysis of numerous features extracted from images using high-throughput methods to further access the biological information from the data. Studies have demonstrated that radiomics approaches based on CT images can achieve preliminary predictions of pulmonary nodule growth trends (53). Building on this foundation, subsequent research has further enriched the input dimensions of predictive models by integrating radiomics features with conventional imaging features, thereby enhancing the predictive performance for pulmonary nodule growth risk. Tan et al. constructed a radiomics-radiological model based on CT images, achieving a growth prediction accuracy with an area under the curve (AUC) of 0.78 (54). Ma et al. integrated radiomics with radiological features to predict fast-growth (VDT ≤400 days) and slow-growth (VDT >400 days) SSNs with an AUC of 0.905 (55). On the basis of images, further clinical features were added to construct predictive models (56). Sun et al. developed a nomogram model combining radiomics, size, age, and location, achieving AUCs of 0.843 and 0.824 in training and validation sets, respectively (57). Yang et al. constructed five radiomics models using different methods, achieving all AUC predictive efficacy of no less than 0.7, but the model could only achieve growth prediction within 1 year (58). Xue et al. achieved individualized 2-year growth prediction of indeterminate small pulmonary nodules by constructing a nomogram of radiomics and clinical parameters (AUC of 0.826 and 0.843 in the internal and external test sets, respectively) (59). Wu et al. demonstrated that radiomics-based models outperformed traditional clinical models in estimating interval growth in early-stage lung adenocarcinoma (≤3 cm), highlighting their utility in predicting substantial interval and staging shift growth (60). Jin et al. (61) developed and validated a radiomics-based machine learning method that achieved good performance in predicting the progressive state (absorption or persistence) of GGNs on initial CT examination (AUC, 0.959). Yuan et al. (33) demonstrated that a model combining radiomics and radiological features was superior to a model using only radiological features in predicting the growth of SSNs.

While radiomics have made significant strides in predicting lung nodule growth by extracting quantitative information, current findings often dichotomize growth outcomes (growth/non-growth) without fully predicting future size and morphology changes, limiting the ability to visualize and quantitatively assess the nodule growth.

Predicting lung nodule growth—deep learning

Since their introduction, deep learning methods have demonstrated superior modeling performance for predicting lung nodule growth compared to traditional mathematical approaches (72,73). Research in this field has progressively evolved from “predicting growth status via classification” towards “predicting growth via image visualization”. Leveraging the powerful feature learning and inference capabilities of deep learning, these advancements effectively address previous limitations in medical image generation and growth evolution modeling. This provides radiologists with more multi-dimensional and intuitive imaging information of nodules, thereby contributing to improved precision in clinical nodule management.

In the early stages of deep learning research in the domain of lung nodule growth, the core task was to develop classification models aimed at determining whether a nodule had undergone growth according to predefined criteria. Liao et al. (27) developed and validated a deep learning-based prediction model to differentiate between growth and non-growth outcomes in subsolid pulmonary nodules, achieving AUCs of 0.858 and 0.760 in the internal and external validation sets, respectively. Tang et al. (63) constructed four deep learning convolutional models (DLCMs) to predict the growth of pure ground-glass nodules, with the best-performing model attaining AUCs of 0.79 and 0.70 on the internal and external validation sets, respectively. Wang et al. developed two distinct deep learning models: one to distinguish between resolving and non-resolving new intermediate-sized lung nodules during CT follow-up (70), and another to predict the disappearance of indeterminate pulmonary nodules (71), demonstrating the potential to reduce unnecessary follow-up scans.

In recent years, deep learning models have evolved beyond merely serving as classifiers for predicting lung nodule growth; they are now capable of generating future images of nodules, marking a significant transition from “prediction” to “visualization”. Zhao et al. (69) proposed MT-NoGNet, a dual-task network that integrates deformation and texture modeling for lung nodule growth prediction, demonstrating high efficacy in generating future nodule appearances [peak signal-to-noise ratio (PSNR): 44.30; structural similarity index (SSIM): 0.7776; Dice similarity coefficient (DSC): 0.7823]. Xiao et al. (64) developed a conditional recursive Variational Autoencodersr using data from 658 patients in the National Lung Screening Trial (NLST) dataset to predict nodule images at three follow-up time points, achieving improved predictive accuracy (DSC: 82.49%). Lee et al. (67) constructed a prediction model based on convolutional long short-term memory networks to forecast anatomical changes and tumor volume evolution, successfully predicting morphological changes of pulmonary nodules at weeks 4, 5, and 6 (mean Dice coefficients for tumor volume: 0.83, 0.82, and 0.81, respectively). In addition to further improving the performance of growth prediction models, deep learning-based growth models have also been explored for extending the prediction time horizon. Rafael-Palou et al. (65) developed an uncertainty-aware hierarchical probabilistic network that achieved a mean absolute error of 1.74 mm in tumor size prediction over a 24-month follow-up period, along with a nodule segmentation Dice coefficient of 78% and a growth prediction accuracy of 84%. Fang et al. (68) proposed a parameterized Gompertz-guided morphological autoencoder capable of generating high-quality images of pulmonary nodules at arbitrary future time points from baseline CT scans, achieving AUCs of 0.803 and 0.808 for volume and mass prediction, respectively.

Building upon this, deep learning research has further advanced to validate the growth discriminative performance of the predicted lung nodule images. Tao et al. devised a CT-based visual prediction system that could visualize and quantify nodules in a three-dimensional manner at any time point in the future, achieving an AUC of 0.857 in predicting nodule growth (62). Wang et al. (66) further integrated the predicted lung nodule images with clinical risk prediction models to explore their incremental value in risk stratification. The study developed a GP-WGAN model based on the Wasserstein generative adversarial network framework, utilizing baseline low-dose CT images to predict nodule growth at 1-year follow-up. The results demonstrated that the model significantly outperformed the Brock model in early diagnosis. Moreover, decision curve analysis revealed that reclassifying baseline nodules based on the generated nodule images led to consistent and significant improvements in risk stratification compared to direct classification of actual nodules using Lung-RADS, the Brock model, or the Lung Cancer Risk Prediction (LCRP) model.

Predicting lung nodule growth—natural language processing (NLP) model

NLP models, a class of AI models specifically designed to process and interpret human language, are demonstrating increasing utility in the integration of medical image-text information, radiology report analysis, and clinical decision support. Mao et al. (74) evaluated the capability of the latest generative pre-trained transformer (GPT)-4o model to capture dynamic changes in lung nodules using longitudinal CT images from 647 patients. Assessed by six radiologists, the model achieved a median Likert score of 4.17 (out of 5.00) for identifying changes in nodule characteristics, indicating a clinically acceptable level of perceptual accuracy in recognizing feature evolution.

Several studies have further explored the performance of NLP in specific clinical scenarios related to lung nodule growth management. In a retrospective study, Moore et al. (75) analyzed 26,545 chest CT examinations performed across three emergency departments between 2014 and 2021. They developed a three-step NLP pipeline using a RoBERTa large language model (LLM) on the SpaCy platform to identify incidental lung nodules requiring follow-up. In cases without known malignancy but with nodules present, the model achieved an overall accuracy of 93.3% in determining follow-up recommendations. Wen et al. (76) further compared the performance of two LLMs, ERNIE-4.0-Turbo-8K and GPT-4o-mini, in providing guideline-based follow-up recommendations for lung nodules, with both models achieving accuracies exceeding 92%, highlighting their potential to support clinical decision-making.

Despite these preliminary investigations, research on the application of NLP in the assessment and dynamic management of lung nodule growth remains limited. Their potential to integrate multi-source clinical text data with imaging information for the development of intelligent decision-support systems has yet to be fully realized. Future efforts should focus on conducting multicenter, large-sample prospective studies to further explore their implementation pathways in real-world clinical settings.

Predicting pulmonary nodule growth: advances and challenges

Assessing the growth potential of pulmonary nodules based on conventional imaging features remains the most fundamental and commonly used approach by radiologists. With the advancement of imaging and AI technologies, the application of radiomics and deep learning methods in the domain of pulmonary nodule growth has enabled the extraction of imaging information imperceptible to the human eye. This has led to groundbreaking progress in predicting nodule growth, demonstrating predictive performance superior to that of traditional imaging features. Furthermore, deep learning has advanced the field from binary classification of nodule growth status to the visualization of future imaging characteristics. More notably, the rapid evolution of NLP models has recently broadened the scope of research in this domain, effectively contributing to the optimization of clinical follow-up strategies and the overall management of pulmonary nodules.

Despite considerable advancements in research on pulmonary nodule growth prediction, a number of key challenges still require further investigation. First, there is a lack of uniformity in growth definition criteria. Current guidelines exhibit subtle differences in the dimensionality and precision of criteria for determining pulmonary nodule growth. This phenomenon may stem from advancements in measurement techniques and imaging equipment, which have reduced measurement variability and enriched the available quantitative indicators. This heterogeneity in growth definitions across studies, however, has hindered meaningful cross-study comparisons and pooled analyses. Hence, advancing the standardization of growth criteria using contemporary imaging measurement modalities is essential to establish a consistent framework for subsequent investigations. Second, spectrum bias. Pulmonary nodules of diverse subtypes follow different growth patterns, yet most existing studies have either enrolled a single nodule type or aggregated all types without stratification. Future investigations should emphasize stratified analyses based on pulmonary nodule subtypes to unravel their respective growth mechanisms. Furthermore, substantial progress in clinical translation remains to be achieved. Despite the emergence of numerous studies on radiomics and deep learning in this domain, large-scale, robustly validated translational outcomes have yet to be established. Future efforts should focus on bridging the critical gap between algorithmic research and clinical deployment, facilitating the real-world implementation of predictive tools. Lastly, the expansion of application scenarios warrants further investigation. While NLP models offer substantial potential for growth prediction and follow-up optimization in pulmonary nodule management, their deployment in key clinical domains such as low-dose screening and surgical timing decision-making remains underexplored. Subsequent investigations should extend their application across multiple scenarios to enhance the intelligence of whole-process pulmonary nodule management.

Conclusions

Lung nodule growth assessment has evolved from subjective morphological features to objective quantitative features, from single-dimensional to multi-dimensional assessment, from single time-point images to longitudinal CT images, and from manual physician measurements to automated predictions by AI. As one of the most well-established and extensively adopted applications of AI in medical imaging, pulmonary nodule management is experiencing substantial breakthroughs driven by these technological advances. Nevertheless, to fully validate the clinical value of these advances and optimize nodule management protocols, further efforts are needed to standardize the definition of pulmonary nodules, expand their application across diverse clinical scenarios, and facilitate the translation of predictive models from algorithm development into real-world deployment.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the Narrative Review reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0344/rc

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0344/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0344/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

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

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Li N, Tan F, Chen W, et al. One-off low-dose CT for lung cancer screening in China: a multicentre, population-based, prospective cohort study. Lancet Respir Med 2022;10:378-91. [Crossref] [PubMed]
3. Hendrix W, Rutten M, Hendrix N, et al. Trends in the incidence of pulmonary nodules in chest computed tomography: 10-year results from two Dutch hospitals. Eur Radiol 2023;33:8279-88. [Crossref] [PubMed]
4. Gould MK, Tang T, Liu IL, et al. Recent Trends in the Identification of Incidental Pulmonary Nodules. Am J Respir Crit Care Med 2015;192:1208-14. [Crossref] [PubMed]
5. Hong JH, Park S, Kim H, et al. Volume and Mass Doubling Time of Lung Adenocarcinoma according to WHO Histologic Classification. Korean J Radiol 2021;22:464-75. [Crossref] [PubMed]
6. Nicholson AG, Tsao MS, Beasley MB, et al. The 2021 WHO Classification of Lung Tumors: Impact of Advances Since 2015. J Thorac Oncol 2022;17:362-87. [Crossref] [PubMed]
7. Kim YW, Kwon BS, Lim SY, et al. Lung cancer probability and clinical outcomes of baseline and new subsolid nodules detected on low-dose CT screening. Thorax 2021;76:980-8. [Crossref] [PubMed]
8. Wu L, Gao C, Kong N, et al. The long-term course of subsolid nodules and predictors of interval growth on chest CT: a systematic review and meta-analysis. Eur Radiol 2023;33:2075-88. [Crossref] [PubMed]
9. Zhang Z, Zhou L, Min X, et al. Long-term follow-up of persistent pulmonary subsolid nodules: Natural course of pure, heterogeneous, and real part-solid ground-glass nodules. Thorac Cancer 2023;14:1059-70. [Crossref] [PubMed]
10. Lee JH, Lim WH, Hong JH, et al. Growth and Clinical Impact of 6-mm or Larger Subsolid Nodules after 5 Years of Stability at Chest CT. Radiology 2020;295:448-55. [Crossref] [PubMed]
11. Thalanayar PM, Altintas N, Weissfeld JL, et al. Indolent, Potentially Inconsequential Lung Cancers in the Pittsburgh Lung Screening Study. Ann Am Thorac Soc 2015;12:1193-6. [Crossref] [PubMed]
12. Gao W, Wen CP, Wu A, et al. Association of Computed Tomographic Screening Promotion With Lung Cancer Overdiagnosis Among Asian Women. JAMA Intern Med 2022;182:283-90. [Crossref] [PubMed]
13. Gould MK, Creekmur B, Qi L, et al. Emotional Distress, Anxiety, and General Health Status in Patients With Newly Identified Small Pulmonary Nodules: Results From the Watch the Spot Trial. Chest 2023;164:1560-71. [Crossref] [PubMed]
14. Jonas DE, Reuland DS, Reddy SM, et al. Screening for Lung Cancer With Low-Dose Computed Tomography: Updated Evidence Report and Systematic Review for the US Preventive Services Task Force. JAMA 2021;325:971-87. [Crossref] [PubMed]
15. Byrne SC, Hammer MM. Malignant Nodules Detected on Lung Cancer Screening CT: Yield of Short-Term Follow-Up CT in Showing Nodule Growth. AJR Am J Roentgenol 2022;219:735-41. [Crossref] [PubMed]
16. Yoon SH, Kim YJ, Doh K, et al. Interobserver variability in Lung CT Screening Reporting and Data System categorisation in subsolid nodule-enriched lung cancer screening CTs. Eur Radiol 2021;31:7184-91. [Crossref] [PubMed]
17. Saghir Z, Dirksen A, Ashraf H, et al. CT screening for lung cancer brings forward early disease. The randomised Danish Lung Cancer Screening Trial: status after five annual screening rounds with low-dose CT. Thorax 2012;67:296-301.
18. Wilson DO, Ryan A, Fuhrman C, et al. Doubling times and CT screen–detected lung cancers in the Pittsburgh Lung Screening Study. Am J Respir Crit Care Med 2012;185:85-9. [Crossref] [PubMed]
19. Field JK, Duffy SW, Baldwin DR, et al. UK Lung Cancer RCT Pilot Screening Trial: baseline findings from the screening arm provide evidence for the potential implementation of lung cancer screening. Thorax 2016;71:161-70. [Crossref] [PubMed]
20. Du Y, Zhao Y, Sidorenkov G, et al. Methods of computed tomography screening and management of lung cancer in Tianjin: design of a population-based cohort study. Cancer Biol Med 2019;16:181-8. [Crossref] [PubMed]
21. Prokop M, Schaefer-Prokop C, Jacobs C, et al. Aggressiveness-guided nodule management for lung cancer screening in Europe-justification for follow-up intervals and definition of growth. Eur Radiol 2026;36:122-34. [Crossref] [PubMed]
22. Qi LL, Wu BT, Tang W, et al. Long-term follow-up of persistent pulmonary pure ground-glass nodules with deep learning-assisted nodule segmentation. Eur Radiol 2020;30:744-55. [Crossref] [PubMed]
23. Zhang R, Tian P, Qiu Z, et al. The growth feature and its diagnostic value for benign and malignant pulmonary nodules met in routine clinical practice. J Thorac Dis 2020;12:2019-30. [Crossref] [PubMed]
24. Lee JH, Hwang EJ, Lim WH, et al. Determination of the optimum definition of growth evaluation for indeterminate pulmonary nodules detected in lung cancer screening. PLoS One 2022;17:e0274583. [Crossref] [PubMed]
25. Kerpel-Fronius A, Monostori Z, Kovacs G, et al. Nationwide lung cancer screening with low-dose computed tomography: implementation and first results of the HUNCHEST screening program. Eur Radiol 2022;32:4457-67. [Crossref] [PubMed]
26. Bartlett EC, Kemp SV, Rawal B, et al. Defining growth in small pulmonary nodules using volumetry: results from a "coffee-break" CT study and implications for current nodule management guidelines. Eur Radiol 2022;32:1912-20. [Crossref] [PubMed]
27. Liao RQ, Li AW, Yan HH, et al. Deep learning-based growth prediction for sub-solid pulmonary nodules on CT images. Front Oncol 2022;12:1002953. [Crossref] [PubMed]
28. He Y, Xiong Z, Zhang J, et al. Growth assessment of pure ground-glass nodules on CT: comparison of density and size measurement methods. J Cancer Res Clin Oncol 2023;149:9937-46. [Crossref] [PubMed]
29. Yankelevitz DF, Yip R, Henschke CI. Impact of Duration of Diagnostic Workup on Prognosis for Early Lung Cancer. J Thorac Oncol 2023;18:527-37. [Crossref] [PubMed]
30. Jiang X, Liu MW, Zhang X, et al. Observational Study of the Natural Growth History of Peripheral Small-Cell Lung Cancer on CT Imaging. Diagnostics (Basel) 2023;13:2560. [Crossref] [PubMed]
31. Jiang B, Lancaster HL, Davies MPA, et al. AI performance for nodule volume doubling time in the follow-up of the UKLS lung cancer screening study compared to expert consensus and histological validation. Eur J Cancer 2026;232:116137. [Crossref] [PubMed]
32. Chen J, Liu J, Qi L, et al. Influencing factors and prediction of growth heterogeneity in solid nodule non-small cell lung cancer based on artificial intelligence: a prospective study. Transl Lung Cancer Res 2025;14:4285-300. [Crossref] [PubMed]
33. Yuan C, Li H, Zhang J, et al. Prediction model for growth trends of subpleural subsolid nodules using CT radiomics and radiological features. Sci Rep 2025;15:24800. [Crossref] [PubMed]
34. Gimbel IA, Bergsma M, van de Weijer MAJ, et al. Earlier discharge from pulmonary nodule follow-up using artificial intelligence based volume measurements in computed tomography. Eur J Radiol 2025;190:112253. [Crossref] [PubMed]
35. MacMahon H, Naidich DP, Goo JM, et al. Guidelines for Management of Incidental Pulmonary Nodules Detected on CT Images: From the Fleischner Society 2017. Radiology 2017;284:228-43. [Crossref] [PubMed]
36. Callister ME, Baldwin DR, Akram AR, et al. British Thoracic Society guidelines for the investigation and management of pulmonary nodules. Thorax 2015;70.
37. Christensen J, Prosper AE, Wu CC, et al. ACR Lung-RADS v2022: Assessment Categories and Management Recommendations. Chest 2024;165:738-53. [Crossref] [PubMed]
38. Ettinger DS, Wood DE, Aisner DL, et al. NCCN Guidelines® Insights: Non-Small Cell Lung Cancer, Version 2.2023. J Natl Compr Canc Netw 2023;21:340-50. [Crossref] [PubMed]
39. Han D, Heuvelmans MA, Vliegenthart R, et al. Influence of lung nodule margin on volume- and diameter-based reader variability in CT lung cancer screening. Br J Radiol 2018;91:20170405. [Crossref] [PubMed]
40. Nair VS. Management of lung nodules detected by volume CT scanning. N Engl J Med 2010;362:757.
41. Mottram J. On the origin of tar tumours in mice, whether from single cells or many cells. J Pathol Bacteriol 1935;40:407-14.
42. de Margerie-Mellon C, Ngo LH, Gill RR, et al. The Growth Rate of Subsolid Lung Adenocarcinoma Nodules at Chest CT. Radiology 2020;297:189-98. [Crossref] [PubMed]
43. Park S, Lee SM, Kim S, et al. Volume Doubling Times of Lung Adenocarcinomas: Correlation with Predominant Histologic Subtypes and Prognosis. Radiology 2020;295:703-12. [Crossref] [PubMed]
44. Horeweg N, van Rosmalen J, Heuvelmans MA, et al. Lung cancer probability in patients with CT-detected pulmonary nodules: a prespecified analysis of data from the NELSON trial of low-dose CT screening. Lancet Oncol 2014;15:1332-41. [Crossref] [PubMed]
45. de Hoop B, Gietema H, van de Vorst S, et al. Pulmonary ground-glass nodules: increase in mass as an early indicator of growth. Radiology 2010;255:199-206. [Crossref] [PubMed]
46. Shi Z, Deng J, She Y, et al. Quantitative features can predict further growth of persistent pure ground-glass nodule. Quant Imaging Med Surg 2019;9:283-91. [Crossref] [PubMed]
47. Qiu T, Ru X, Yin K, et al. Two nomograms based on CT features to predict tumor invasiveness of pulmonary adenocarcinoma and growth in pure GGN: a retrospective analysis. Jpn J Radiol 2020;38:761-70. [Crossref] [PubMed]
48. Gao C, Li J, Wu L, et al. The Natural Growth of Subsolid Nodules Predicted by Quantitative Initial CT Features: A Systematic Review. Front Oncol 2020;10:318. [Crossref] [PubMed]
49. He Y, Xiong Z, Tian D, et al. Natural progression of persistent pure ground-glass nodules 10 mm or smaller: long-term observation and risk factor assessment. Jpn J Radiol 2023;41:605-16. [Crossref] [PubMed]
50. Xia T, Cai M, Zhuang Y, et al. Risk Factors for The Growth of Residual Nodule in Surgical Patients with Adenocarcinoma Presenting as Multifocal Ground-glass Nodules. Eur J Radiol 2020;133:109332. [Crossref] [PubMed]
51. Liang X, Liu M, Li M, et al. Clinical and CT Features of Subsolid Pulmonary Nodules With Interval Growth: A Systematic Review and Meta-Analysis. Front Oncol 2022;12:929174. [Crossref] [PubMed]
52. Xue W, Kong L, Zhang X, et al. Tumor blood vessel in 3D reconstruction CT imaging as an risk indicator for growth of pulmonary nodule with ground-glass opacity. J Cardiothorac Surg 2023;18:333. [Crossref] [PubMed]
53. Huang W, Zhang H, Ge Y, et al. Radiomics-based Machine Learning Methods for Volume Doubling Time Prediction of Pulmonary Ground-glass Nodules With Baseline Chest Computed Tomography. J Thorac Imaging 2023;38:304-14. [Crossref] [PubMed]
54. Tan M, Ma W, Sun Y, et al. Prediction of the Growth Rate of Early-Stage Lung Adenocarcinoma by Radiomics. Front Oncol 2021;11:658138. [Crossref] [PubMed]
55. Ma ZJ, Ma ZX, Sun YL, et al. Prediction of subsolid pulmonary nodule growth rate using radiomics. BMC Med Imaging 2023;23:177. [Crossref] [PubMed]
56. Gao C, Yan J, Luo Y, et al. The Growth Trend Predictions in Pulmonary Ground Glass Nodules Based on Radiomic CT Features. Front Oncol 2020;10:580809. [Crossref] [PubMed]
57. Sun Y, Ma Z, Zhao W, et al. Computed tomography radiomics in growth prediction of pulmonary ground-glass nodules. Eur J Radiol 2023;159:110684. [Crossref] [PubMed]
58. Yang R, Hui D, Li X, et al. Prediction of single pulmonary nodule growth by CT radiomics and clinical features - a one-year follow-up study. Front Oncol 2022;12:1034817. [Crossref] [PubMed]
59. Xue LM, Li Y, Zhang Y, et al. A predictive nomogram for two-year growth of CT-indeterminate small pulmonary nodules. Eur Radiol 2022;32:2672-82. [Crossref] [PubMed]
60. Wu FZ, Wu YJ, Chen CS, et al. Prediction of Interval Growth of Lung Adenocarcinomas Manifesting as Persistent Subsolid Nodules ≤3 cm Based on Radiomic Features. Acad Radiol 2023;30:2856-69. [Crossref] [PubMed]
61. Jin L, Liu Z, Sun Y, et al. Prediction of Pulmonary Ground-Glass Nodule Progression State on Initial Screening CT Using a Radiomics-Based Model. Respirology 2026;31:73-81. [Crossref] [PubMed]
62. Tao G, Zhu L, Chen Q, et al. Prediction of future imagery of lung nodule as growth modeling with follow-up computed tomography scans using deep learning: a retrospective cohort study. Transl Lung Cancer Res 2022;11:250-62. [Crossref] [PubMed]
63. Tang Y, Li M, Lin B, et al. Deep learning-assisted development and validation of an algorithm for predicting the growth of persistent pure ground-glass nodules. Transl Lung Cancer Res 2023;12:2494-504. [Crossref] [PubMed]
64. Xiao N, Qiang Y, Zhao Z, et al. Tumour growth prediction of follow-up lung cancer via conditional recurrent variational autoencoder. IET Image Processing 2020;14:3975-81.
65. Rafael-Palou X, Aubanell A, Ceresa M, et al. Prediction of Lung Nodule Progression with an Uncertainty-Aware Hierarchical Probabilistic Network. Diagnostics (Basel) 2022;12:2639. [Crossref] [PubMed]
66. Wang Y, Zhou C, Ying L, et al. Enhancing Early Lung Cancer Diagnosis: Predicting Lung Nodule Progression in Follow-Up Low-Dose CT Scan with Deep Generative Model. Cancers (Basel) 2024;16:2229. [Crossref] [PubMed]
67. Lee D, Alam SR, Jiang J, et al. Deformation driven Seq2Seq longitudinal tumor and organs-at-risk prediction for radiotherapy. Med Phys 2021;48:4784-98. [Crossref] [PubMed]
68. Fang J, Wang J, Li A, et al. Parameterized Gompertz-Guided Morphological AutoEncoder for Predicting Pulmonary Nodule Growth. IEEE Trans Med Imaging 2023;42:3602-13. [Crossref] [PubMed]
69. Zhao W, Chen Y, Xie Y, et al. Multi-task learning for predicting pulmonary nodule growth and follow-up volume. Front Oncol 2026;16:1677521. [Crossref] [PubMed]
70. Wang J, Zhang X, Tang W, et al. A multi-view CNN model to predict resolving of new lung nodules on follow-up low-dose chest CT. Insights Imaging 2025;16:138. [Crossref] [PubMed]
71. Wang J, Cai J, Tang W, et al. A comparison of an integrated and image-only deep learning model for predicting the disappearance of indeterminate pulmonary nodules. Comput Med Imaging Graph 2025;123:102553. [Crossref] [PubMed]
72. Zhang L, Lu L, Summers RM, et al. Convolutional Invasion and Expansion Networks for Tumor Growth Prediction. IEEE Trans Med Imaging 2018;37:638-48. [Crossref] [PubMed]
73. Zhang L, Lu L, Wang X, et al. Spatio-Temporal Convolutional LSTMs for Tumor Growth Prediction by Learning 4D Longitudinal Patient Data. IEEE Trans Med Imaging 2020;39:1114-26. [Crossref] [PubMed]
74. Mao Y, Xu N, Wu Y, et al. Assessments of lung nodules by an artificial intelligence chatbot using longitudinal CT images. Cell Rep Med 2025;6:101988. [Crossref] [PubMed]
75. Moore CL, Socrates V, Hesami M, et al. Using natural language processing to identify emergency department patients with incidental lung nodules requiring follow-up. Acad Emerg Med 2025;32:274-83. [Crossref] [PubMed]
76. Wen J, Huang W, Yan H, et al. Evaluation of large language models in generating pulmonary nodule follow-up recommendations. Eur J Radiol Open 2025;14:100655. [Crossref] [PubMed]