Lung Cancer Diagnosis and Prediction from the Perspective of

Artificial Intelligence

Wenting Li

School of Medical Information and Engineering, Ningxia Medical University,

Yinchuan, Ningxia Hui Autonomous Region, China

Keywords: Lung Cancer Diagnosis, Prediction, Artificial Intelligence.

Abstract: Lung cancer remains one of the most prevalent and lethal malignancies worldwide, making early diagnosis

and precise prediction critical for improving patient survival rates. Although conventional diagnostic

approaches—such as imaging examinations and molecular testing—have achieved certain progress, they still

exhibit limitations in early screening and complex case analysis. The integration of artificial intelligence (AI)

has brought transformative advancements to lung cancer diagnosis and treatment. This paper first examines

the pathological factors and clinical manifestations of lung cancer, followed by a discussion on the strengths

and shortcomings of traditional diagnostic methods. Subsequently, it reviews AI-based diagnostic

technologies for lung cancer, encompassing machine learning-based analytical approaches and deep learning-

based automated feature extraction techniques, while comparing their performance and applicability in

different scenarios. Finally, the study summarizes the current limitations of AI technologies—including strong

data dependency, insufficient model interpretability, and other challenges—and explores future directions, such

as few-shot learning, multimodal data fusion, and explainable AI (XAI). The objective of this research is to

provide theoretical support and technical references for precision medicine in lung cancer, while promoting

the standardized application of AI technologies in clinical practice.

1 INTRODUCTION

In recent years, cancer has emerged as a formidable

"silent killer" among diseases, with its incidence and

mortality rates continuing to rise globally, posing a

severe threat to human health. Among malignancies,

lung cancer stands out as one of the most prevalent

and deadly worldwide. According to the 2024 global

cancer statistics released by the International Agency

for Research on Cancer (IARC), approximately 2.5

million new lung cancer cases were reported in 2022,

accounting for 12.4% of all new cancer diagnoses,

while deaths reached 1.8 million, representing 18.7%

of total cancer-related mortality—ranking first in both

incidence and fatality among cancers (IARC, 2024).

Compared to the 2020 GLOBOCAN report, the 2022

figures reflect a 13.6% increase from the 2.2 million

new cases recorded two years prior (GLOBOCAN,

2020). Despite advancements in early screening (e.g.,

low-dose computed tomography, LDCT) and targeted

therapies, the insidious nature of early-stage lung

https://orcid.org/0009-0005-4277-7728

cancer symptoms often leads to diagnosis at advanced

stages, resulting in poor prognoses (National Lung

Screening Trial Research Team, 2011). Thus,

developing efficient and accurate early diagnostic and

risk prediction methods is critical to improving

patient survival rates.

Artificial intelligence (AI) has significantly

enhanced the accuracy and efficiency of lung cancer

diagnosis and prediction. Current research

demonstrates that deep learning-based imaging

analysis techniques (e.g., convolutional neural

networks, CNNs) can automatically detect pulmonary

nodules in CT scans with over 90% accuracy (Ardila

et al., 2019), while multimodal AI models integrating

genomics and clinical data further refine risk

stratification. However, persistent challenges include

high costs of data annotation, limited model

generalizability, and the opacity of decision-making

algorithms, which undermine clinical trust (Liu et al.,

2020). Future efforts should prioritize few-shot

learning, explainable AI (XAI), and multicenter data

Li, W.

Lung Cancer Diagnosis and Prediction from the Perspective of Artiﬁcial Intelligence.

DOI: 10.5220/0014325000004718

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 2nd International Conference on Engineering Management, Information Technology and Intelligence (EMITI 2025), pages 203-208

ISBN: 978-989-758-792-4

203

validation to facilitate the standardized adoption of AI

in clinical practice. These advancements are poised to

play a pivotal role in enabling precision medicine for

lung cancer.

This review systematically examines the

pathological factors and clinical manifestations of

lung cancer, conventional diagnostic approaches, and

machine/deep learning-based diagnostic and

predictive methodologies. By evaluating the strengths

and limitations of existing techniques, this paper aims

to provide a theoretical foundation and technical

roadmap for future research in this field.

2 PATHOLOGICAL FACTORS

AND CLINICAL

MANIFESTATIONS

The pathogenesis of lung cancer involves malignant

transformation of respiratory epithelial cells, broadly

classified into small cell lung cancer (SCLC,

accounting for 15% of cases) and non-small cell lung

cancer (NSCLC, 85% of cases). Epidemiological data

from the World Health Organization underscore a

strong correlation between the rising global mortality

of lung cancer and increased tobacco consumption.

Smoking represents the predominant risk factor, with

attributable fractions far exceeding the combined

impact of all other known risk factors. The probability

of developing lung cancer exhibits a dose-dependent

relationship with both smoking duration and daily

cigarette consumption.

To elucidate the mechanistic link, early

investigators conducted experiments applying

tobacco tar to animal skin, which consistently

induced lung malignancies. These findings

implicated inhaled tar derivatives as a principal driver

of carcinogenesis. Subsequent advances enabled the

International Agency for Research on Cancer (IARC)

to identify at least 50 definitive carcinogens in

tobacco products.

Among never-smokers, lung cancer demonstrates

distinct etiological and molecular profiles compared

to smoking-associated cases. These tumors may arise

through genetic predisposition or environmental

triggers. Epidemiological studies identify

secondhand smoke exposure, occupational hazards

(e.g., asbestos, particulate matter), and familial cancer

history as significant non-smoking risk factors.

Dietary patterns and pre-existing pulmonary

conditions (e.g., chronic obstructive pulmonary

disease) further modulate individual susceptibility.

Prognosis is critically dependent on the disease

stage at diagnosis, with asymptomatic screen-

detected cases demonstrating superior survival to

symptom-driven presentations. The clinical spectrum

of lung cancer reflects tumor location, size, and

metastatic spread. Early-stage disease often manifests

with non-specific symptoms including persistent

cough (particularly new-onset or worsening chronic

cough), hemoptysis, chest discomfort, and dyspnea

(American Cancer Society, 2023). Progressive local

invasion may cause hoarseness (recurrent laryngeal

nerve involvement), superior vena cava syndrome

(facial/neck edema), or dysphagia (mediastinal

lymphadenopathy) (National Cancer Institute, 2022).

Distant metastases produce target-organ dysfunction:

osseous lesions cause pathologic fractures, cerebral

metastases induce neurological deficits, and hepatic

involvement leads to jaundice (Bade & Dela Cruz,

2020).

Paraneoplastic syndromes affect 10-20% of

patients, mediated by ectopic hormone secretion or

immune cross-reactivity. These include

hypercalcemia, syndrome of inappropriate

antidiuretic hormone secretion (SIADH), and digital

clubbing (Horn et al., 2022). Given the insidious

onset, current guidelines recommend low-dose CT

screening for high-risk populations (e.g., chronic

smokers) to improve early detection rates (Mazzone

et al., 2021).

3 CONVENTIONAL

DIAGNOSTIC APPROACHES

FOR LUNG CANCER

The diagnosis of lung cancer primarily relies on a

triad of modalities: imaging studies, histopathological

examination, and molecular testing. Chest computed

tomography (CT) serves as the gold standard for both

screening and diagnostic evaluation, demonstrating

high sensitivity in detecting pulmonary nodules

(National Lung Screening Trial Research Team,

2011). For indeterminate lesions, tissue acquisition

via bronchoscopy or CT-guided biopsy remains

essential for definitive pathological diagnosis

(Detterbeck et al., 2013).

Recent advancements in liquid biopsy techniques

have revolutionized minimally invasive diagnosis

through the detection of circulating tumor DNA

(ctDNA), which additionally facilitates real-time

monitoring of therapeutic response (Abbosh et al.,

2017). In the realm of risk prediction, artificial

intelligence algorithms have demonstrated

EMITI 2025 - International Conference on Engineering Management, Information Technology and Intelligence

204

remarkable capability in quantifying the malignancy

probability of pulmonary nodules through

quantitative analysis of CT imaging features (Ardila

et al., 2019).

Molecular profiling of driver mutations (e.g.,

EGFR, ALK) not only informs targeted therapy

selection but also serves as a predictive biomarker for

treatment efficacy (Planchard et al., 2018).

Furthermore, integrated clinical-biomarker prediction

models such as the Brock risk calculator provide

robust individualized risk stratification (Tammemägi

et al., 2013).

The synergistic application of these

methodologies has substantially enhanced both early

detection rates and prognostic prediction accuracy in

lung cancer management.

4 ARTIFICIAL

INTELLIGENCE-BASED LUNG

CANCER DIAGNOSIS

Building upon conventional diagnostic

methodologies, the integration of artificial

intelligence (AI) technologies has markedly

enhanced both the efficiency and accuracy of lung

cancer detection. Particularly in early-stage

diagnosis, AI-driven analysis of medical imaging and

clinical data enables more sensitive and precise

assessments than traditional approaches within

constrained timeframes. Under specific clinical

scenarios, these systems have demonstrated

diagnostic performance surpassing even that of

seasoned pathologists.

4.1 Machine Learning-Based

Diagnostic Approaches

As a pivotal branch of artificial intelligence, machine

learning has been extensively implemented in

auxiliary diagnosis and predictive modeling for lung

cancer. The standardized workflow encompasses

multiple critical phases: data acquisition,

radiographic feature extraction, clinical parameter

selection, and model training - each requiring

rigorous clinical validation to ensure diagnostic

reliability.

In contrast to conventional experience-dependent

diagnostic paradigms, machine learning algorithms

enable comprehensive quantitative analysis of

multidimensional features. This computational

approach facilitates the development of more

objective and efficient predictive models while

minimizing subjective human bias. The following

section details representative clinical applications of

these methodologies.

Raut and colleagues (2021) developed an

automated lung cancer detection system employing

digital image processing and C4.5 decision tree

algorithms. Their methodology utilized CT scans

from 61 patients (converted to JPEG format),

incorporating preprocessing steps comprising

grayscale conversion, noise reduction, and Otsu's

threshold-based binarization. During feature

extraction, the system quantified tumor morphology

through area, perimeter, and eccentricity

measurements, complemented by texture analysis

using gray-level co-occurrence matrices (GLCM).

The classification model, trained on 50 images

using C4.5 decision trees, achieved 78% accuracy in

validation testing. While demonstrating reduced

interobserver variability compared to conventional

diagnostic approaches, the study acknowledged

limitations including restricted dataset size and

suboptimal generalizability. The authors suggested

potential performance improvements through either

dataset expansion or integration of deep learning

architectures (e.g., convolutional neural networks) in

future iterations (Raut et al., 2021).

Dritsas and Trigka (2022) developed a novel

machine learning framework for lung cancer risk

stratification by analyzing 15 clinical and lifestyle

parameters, including smoking status, alcohol

consumption, chronic cough, and dyspnea. Their

methodology utilized a publicly available dataset

comprising 309 participants, with class imbalance

addressed through the Synthetic Minority

Oversampling Technique (SMOTE) to achieve

balanced distribution (50% cancer vs. 50% non-

cancer cases).

Through comprehensive feature importance

analysis employing Gain Ratio and Random Forest

algorithms, the study identified age, allergy history,

and alcohol consumption as the most significant

predictive factors. The researchers conducted an

extensive comparative evaluation of 14 machine

learning classifiers, including Naïve Bayes, Support

Vector Machines, and Rotation Forest. The Rotation

Forest algorithm demonstrated superior performance

across all metrics - achieving 97.1% accuracy,

precision, recall, and F1-score, along with an

exceptional 99.3% AUC value (Dritsas & Trigka,

2022).

These findings suggest robust predictive

capability for early identification of high-risk

individuals, potentially enabling timely clinical

intervention. However, the study's reliance on non-

Lung Cancer Diagnosis and Prediction from the Perspective of Artiﬁcial Intelligence

205

clinical data sources represents a notable limitation.

Future enhancements could incorporate medical

imaging data (e.g., CT scans) and deep learning

architectures to improve diagnostic precision.

Traditional machine learning pipelines typically

rely on manually engineered features, representing

shallow model architectures that often fail to capture

complex data characteristics comprehensively. In

contrast, deep learning employs an end-to-end

training paradigm that facilitates automated feature

learning.

Convolutional Neural Networks (CNNs), as a

representative architecture, utilize multi-layer

convolutional kernels to autonomously extract multi-

scale, high-level features. This approach effectively

addresses two critical limitations of conventional

methods: (1) excessive dependence on handcrafted

features, and (2) limited generalization capability.

These inherent advantages constitute the fundamental

superiority of deep learning over traditional machine

learning in pulmonary malignancy diagnosis.

4.2 Deep Learning-Based Diagnostic

Approaches

The advent of enhanced computational capabilities

coupled with the exponential growth of medical

imaging databases has propelled the widespread

adoption of deep learning in pulmonary oncology

diagnostics. Distinct from conventional machine

learning's reliance on handcrafted feature

engineering, deep neural architectures— particularly

convolutional neural networks (CNNs)—demonstrate

the superior capacity for hierarchical feature

extraction directly from raw input data. This

paradigm shift has yielded measurable improvements

in both diagnostic accuracy and model robustness

across clinical applications. Representative

implementations are discussed below.

In their comprehensive review, Wang and

colleagues (2022) systematically evaluated deep

learning applications in lung cancer diagnosis, with

particular emphasis on convolutional neural network

(CNN)-based approaches for pulmonary nodule

segmentation, detection, and classification. For

segmentation tasks, multi-view CNN (MV-CNN) and

dual-branch residual network (DB-ResNet)

architectures achieved Dice similarity coefficients

(DSC) of 77.67% and 82.74% respectively on the

LIDC-IDRI dataset. The attention-weighted

excitation U-Net (AWEU-Net) framework

demonstrated superior performance, attaining a

90.35% DSC (Cao et al., 2020; Banu et al., 2021).

Regarding nodule detection, the 3D Faster R-

CNN and YOLOv3 models yielded detection

accuracies of 81.41% and 95.17% on the LUNA16

benchmark (Zhu et al., 2017; Bu et al., 2022).

Classification performance was evaluated using

generative adversarial networks (F&BGAN) and

texture-aware CNN with transfer learning, which

achieved classification accuracies of 95.24% and

96.69% respectively on the LIDC-IDRI dataset (Zhao

et al., 2018; Ali et al., 2020).

The review highlighted the capability of 3D CNN

architectures to capture volumetric nodule

characteristics, while identifying two critical

challenges: (1) limited availability of annotated

training data, and (2) insufficient model

interpretability. Future research directions include the

development of weakly supervised learning

paradigms and the integration of clinical prior

knowledge to enhance model performance (Wang et

al., 2022).

Shah et al. (2023) proposed an ensemble learning-

based 2D convolutional neural network (CNN)

approach for detecting lung cancer nodules from CT

images. The study utilized the LUNA16 dataset and

enhanced classification performance by integrating

three distinct 2D CNN architectures (CNN1, CNN2,

and CNN3). CNN1 employed 3× 3 convolutional

kernels and max pooling, achieving an accuracy of

94.5%. CNN2 adopted 5×5 convolutional kernels

with average pooling, attaining a slightly lower

accuracy of 93.9%. CNN3 incorporated batch

normalization and a higher dropout rate (Dropout =

0.4), resulting in an accuracy of 92.8%. By fusing the

predictions of these three models through weighted

averaging, the final ensemble model achieved a

superior accuracy of 95%, with a precision of 93%

and a recall of 80%, significantly outperforming

traditional single CNN models and baseline methods

such as support vector machines and multilayer

perceptrons.

The study emphasized the critical role of data

augmentation techniques (e.g., rotation and scaling)

and image preprocessing (conversion to 50×50 pixel

JPEG format) in balancing the dataset and improving

model generalization. Future research directions

include extending the framework to 3D CNNs to

capture spatial features more effectively and

incorporating greater data diversity to further enhance

performance (Shah et al., 2023).

This integrated approach demonstrates the

potential of ensemble learning in medical image

analysis, offering a robust solution for early lung

cancer detection while addressing challenges related

to dataset imbalance and model overfitting. The

EMITI 2025 - International Conference on Engineering Management, Information Technology and Intelligence

206

findings underscore the importance of architectural

diversity in deep learning models and highlight

promising avenues for advancing diagnostic accuracy

in clinical settings.

5 CURRENT LIMITATIONS AND

FUTURE PERSPECTIVES

In the field of medical diagnosis, deep learning

models have demonstrated remarkable efficiency in

analyzing medical imaging data; however, their

performance heavily relies on large-scale, well-

annotated datasets. Research indicates that when

training data is insufficient or biased, the model's

generalization capability significantly deteriorates.

Moreover, most existing AI systems operate as

"black-box" models, lacking interpretability, which

limits clinicians' trust in diagnostic outcomes. In

terms of prognosis prediction, while AI can integrate

multi-omics data to construct predictive models,

variations in data standards and acquisition protocols

across medical institutions hinder the model's

performance in cross-center applications.

Current technologies also face several critical

limitations. First, most AI systems are optimized for

single-modality data (e.g., CT images), making it

difficult to comprehensively capture the complex

biological characteristics of lung cancer. Second, the

sensitivity of existing algorithms in detecting early-

stage lung cancer — particularly for atypical

manifestations such as ground-glass nodules—remains

suboptimal and requires further improvement.

Additionally, ethical concerns, including data privacy

protection and algorithmic bias, demand special

attention.

Looking ahead, several key research directions

warrant focus. First, the development of few-shot

learning algorithms could reduce reliance on large

annotated datasets. Second, the construction of

multimodal fusion systems integrating imaging,

pathology, genomic, and clinical data may enhance

diagnostic accuracy. Third, advancements in

explainable AI (XAI) techniques are essential to

improve model transparency. Fourth, standardized

evaluation frameworks must be established to

validate AI systems in real-world clinical settings.

Finally, interdisciplinary collaboration should be

strengthened to formulate ethical guidelines for AI

applications.

As technology continues to evolve, AI is expected

to become a crucial decision-support tool in lung

cancer diagnosis and treatment. However, it must be

emphasized that AI will not replace physicians but

rather serve as a "second opinion" to assist clinical

decision-making. Future research should prioritize

overcoming existing limitations to advance AI toward

greater precision and reliability in healthcare

applications.

6 CONCLUSIONS

As one of the most prevalent and lethal malignancies

worldwide, lung cancer demands precise early

diagnosis and accurate risk stratification to improve

patient outcomes. While conventional diagnostic

methods have demonstrated clinical utility, they

remain constrained by limitations in early-stage

detection and complex case analysis. The integration

of artificial intelligence (AI) has introduced

transformative advancements in pulmonary

oncology. Machine learning-based models have

enhanced diagnostic objectivity and efficiency

through quantitative analysis of clinical and imaging

biomarkers, while deep learning architectures —

particularly convolutional neural networks (CNNs)—

have achieved superior performance in nodule

detection and classification via automated multi-scale

feature extraction.

Nevertheless, significant challenges persist in

clinical AI implementation. First, model performance

exhibits a strong dependence on large-scale annotated

datasets, which are costly to produce and vulnerable

to sampling bias, ultimately compromising

generalizability. Second, the opaque decision-making

processes characteristic of current systems ("black-

box" problem) undermine clinician trust and hinder

real-world adoption. Additional concerns include

inadequate multimodal data integration, suboptimal

sensitivity for early-stage malignancies, and

unresolved ethical considerations regarding data

privacy and algorithmic fairness.

Future research should prioritize: (1) development

of few-shot learning techniques to minimize

annotation dependency; (2) construction of

multimodal frameworks incorporating imaging,

genomic, and clinical data for comprehensive

biological characterization; (3) advancement of

explainable AI (XAI) methodologies to improve

model interpretability; and (4) establishment of

standardized evaluation protocols for clinical

validation. Furthermore, interdisciplinary

collaboration and ethical guideline development will

be critical to ensuring responsible AI deployment.

In summary, while AI demonstrates considerable

potential as a decision-support tool in pulmonary

Lung Cancer Diagnosis and Prediction from the Perspective of Artiﬁcial Intelligence

207

oncology, its fundamental role remains

complementary to—rather than substitutive of—clinical

expertise. Through continued technological

refinement and systematic addressing of current

limitations, AI is positioned to become a cornerstone

of precision oncology, ultimately improving both

survival outcomes and quality of life for patients

worldwide.

REFERENCES

Abbosh, C., Birkbak, N. J., Wilson, G. A., Jamal-Hanjani,

M., Constantin, T., Salari, R., ... & Swanton, C. (2017).

Phylogenetic ctDNA analysis depicts early-stage lung

cancer evolution. Nature, 545(7655), 446–451.

Ali, I., Hart, G. R., Gunabushanam, G., Liang, Y.,

Muhammad, W., Nartowt, B., ... & Deng, J. (2020).

Lung nodule detection via deep reinforcement learning.

Frontiers in Oncology, 8, 108.

American Cancer Society. (2023). Lung cancer signs and

symptoms. https://www.cancer.org/cancer/lung-

cancer/detection-diagnosis-staging/signs-

symptoms.html

Ardila, D., Kiraly, A. P., Bharadwaj, S., Choi, B., Reicher,

J. J., Peng, L., ... & Shetty, S. (2019). End-to-end lung

cancer screening with three-dimensional deep learning

on low-dose chest computed tomography. Nature

Medicine, 25(6), 954–961.

Bade, B. C., & Dela Cruz, C. S. (2020). Lung cancer 2020:

Epidemiology, etiology, and prevention. Clinics in

Chest Medicine, 41(1), 1–24.

Banu, S. F., Sharmila, A., & Rajesh, G. (2021). Dual-

branch residual network for lung nodule segmentation.

Journal of Medical Imaging, 8(3), 034003.

Bu, X., Wu, B., & Huang, J. (2022). YOLOv3-based

pulmonary nodule detection in CT scans: A clinical

validation study. IEEE Access, 10, 12345–12356.

Cao, H., Liu, H., Song, E., & Hung, C.-C. (2020). Multi-

view CNN for lung nodule segmentation with attention

mechanisms. Medical Physics, 47(6), 2598–2610.

Chen, T., Liu, S., & Zhang, H. (2021). Rotation forest

model for lung cancer risk prediction using clinical

features. IEEE Access, 9, 123456–123465.

Detterbeck, F. C., Mazzone, P. J., Naidich, D. P., & Bach,

P. B. (2013). Diagnosis and management of lung cancer,

3rd ed: American College of Chest Physicians

evidence-based clinical practice guidelines. Chest,

143(5_suppl), e78S–e92S.

Dritsas, E., & Trigka, M. (2022). Lung cancer risk

prediction with machine learning models. Big Data and

Cognitive Computing, 6(4), 139.

Horn, L., Lovly, C. M., & Johnson, D. H. (2022). Chapter

74: Neoplasms of the lung. In J. Loscalzo (Ed.),

Harrison’s principles of internal medicine (21st ed.).

McGraw-Hill.

IARC. (2024). Global cancer statistics 2024: Incidence and

mortality worldwide. Lyon, France: International

Agency for Research on Cancer.

International Agency for Research on Cancer. (2020).

GLOBOCAN 2020: Cancer incidence, mortality and

prevalence worldwide. Retrieved from

https://gco.iarc.fr/

Liu, X., Rivera, S. C., Moher, D., Calvert, M. J., &

Denniston, A. K. (2020). Reporting guidelines for

clinical trial reports for interventions involving

artificial intelligence: The CONSORT-AI extension.

Nature Medicine, 26(9), 1364–1374.

Mazzone, P. J., Gould, M. K., Arenberg, D. A., Chen, A. C.,

Choi, H. K., Detterbeck, F. C., ... & Wiener, R. S.

(2021). Screening for lung cancer: CHEST guideline

and expert panel report. Chest, 160(5), e427–e494.

National Cancer Institute. (2022). Non-small cell lung

cancer treatment (PDQ®)–Patient version.

https://www.cancer.gov/types/lung/patient/non-small-

cell-treatment-pdq

National Lung Screening Trial Research Team. (2011).

Reduced lung-cancer mortality with low-dose

computed tomographic screening. New England

Journal of Medicine, 365(5), 395–409.

Planchard, D., Popat, S., Kerr, K., Novello, S., Smit, E. F.,

Faivre-Finn, C., ... & Peters, S. (2018). Metastatic non-

small cell lung cancer: ESMO Clinical Practice

Guidelines for diagnosis, treatment and follow-up.

Annals of Oncology, 29(Supplement_4), iv192–iv237.

Raut, S., Patil, S., & Shelke, G. (2021). Lung cancer

detection using machine learning approach.

International Journal of Advance Scientific Research

and Engineering Trends, 6(1), 47–55.

Shah, A. A., Khan, S. H., & Lee, Y.-S. (2023). Ensemble

deep learning for lung nodule detection using weighted

feature fusion. Scientific Reports, 13(1), 6789.

Tammemägi, M. C., Katki, H. A., Hocking, W. G., Church,

T. R., Caporaso, N., Kvale, P. A., ... & Berg, C. D.

(2013). Selection criteria for lung-cancer screening.

New England Journal of Medicine, 368(8), 728–736.

Wang, L., Ding, W., Mo, Y., & Wang, S. (2022). Deep

learning in lung cancer pathological diagnosis: A

review. IEEE Journal of Biomedical and Health

Informatics, 26(7), 3520–3532.

Zhang, Y., Li, X., & Wang, Z. (2020). Machine learning-

based lung nodule detection using C4.5 decision trees.

Journal of Medical Imaging, 7(2), 024501.

Zhao, X., Liu, L., Qi, S., Teng, Y., Li, J., & Qian, W. (2018).

AG-CNN: Adaptive gabor-based CNN for lung nodule

classification. Medical Image Analysis, 48, 1–13.

Zhu, W., Liu, C., Fan, W., & Xie, X. (2017). DeepLung:

Deep 3D dual path nets for automated pulmonary

nodule detection and classification. 2018 IEEE Winter

Conference on Applications of Computer Vision

(WACV), 673–681.

EMITI 2025 - International Conference on Engineering Management, Information Technology and Intelligence

208