Abstract
Although lung cancer has been recognized to be the deadliest type of cancer, a good prognosis and efficient treatment depend on early detection. Medical practitioners’ burden is reduced by deep learning techniques, especially Deep Convolutional Neural Networks (DCNN), which are essential in automating the diagnosis and classification of diseases. In this study, we use a variety of medical imaging modalities, including X-rays, WSI, CT scans, and MRI, to thoroughly investigate the use of deep learning techniques in the field of lung cancer diagnosis and classification. This study conducts a comprehensive Systematic Literature Review (SLR) using deep learning techniques for lung cancer research, providing a comprehensive overview of the methodology, cutting-edge developments, quality assessments, and customized deep learning approaches. It presents data from reputable journals and concentrates on the years 2015–2024. Deep learning techniques solve the difficulty of manually identifying and selecting abstract features from lung cancer images. This study includes a wide range of deep learning methods for classifying lung cancer but focuses especially on the most popular method, the Convolutional Neural Network (CNN). CNN can achieve maximum accuracy because of its multi-layer structure, automatic learning of weights, and capacity to communicate local weights. Various algorithms are shown with performance measures like precision, accuracy, specificity, sensitivity, and AUC; CNN consistently shows the greatest accuracy. The findings highlight the important contributions of DCNN in improving lung cancer detection and classification, making them an invaluable resource for researchers looking to gain a greater knowledge of deep learning’s function in medical applications.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Cancer refers to the growth of abnormal tissues that are unwanted, it’s uncontrollable, and it spreads speedily in the body; if it is not treated well at the start, it spreads and affects other body organs also. In the health sector, the use of modern technology has contributed a lot, especially in the detection of lungs cancer. It helps the doctors to identify as well as properly treat a disease. Many deaths occur in the whole world due to lung cancer and due to this, it is one of the deadliest diseases known in the world. In 2020, according to the research, approximately 2.21 million cases were detected, and 1.8 million mortalities were caused by lungs cancer (Sharma 2022). The report presented by the World Health Organization (WHO) in 2020, shows that lungs cancer is the deadliest among all kinds of cancers, that is said according to the death rate that is calculated as 1.80 million (World Health Organization 2022). Figure 1 shows the details of extinction because of cancer in 2020 according to WHO Lungs cancer is one of those diseases in which early-stage diagnosis and disease management play a crucial role in proper treatment.
Just like other cancers, the early detection of lungs cancer is mandatory due to which the chances of survival increase (Pathak et al. 2018). A large number of people affected by lung cancer cannot survive due to the delay in early detection, the overall survival rate of the patient is five years which is less than 20% (Roointan et al. 2019). Age is not a vital prognostic factor when it comes to the survival of patients (Hurria and Kris 2003). Both males and females fall prey to it. Men are more prone to lungs cancer than females According to research, the death rate in men due to lungs cancer is higher than females. (Chen et al. 2016). Factors contributing to increased lungs cancer cases include tobacco, smoke, viral infection, ionizing radiation (Cook et al. 1993; Esposito et al. 2010) air pollution, and unhealthy lifestyle. (Strak et al. 2017) History of chronic obstructive pulmonary disease (COPD) is the main factor that contributes to lungs cancer (Parris et al. 2019). The drastic increase in vehicles and bad smoking habits also play a crucial role. As tobacco is a primary factor, driving lungs cancer trend (Parascandola and Xiao 2019) so, the death toll due to lung cancer can be reduced by controlling tobacco usage (Field et al. 2013). Its symptoms include fatigue, difficulty in breathing, and persistent cough (Corner 2005; Mayo Clinic 2022). Apart from common symptoms, its symptoms vary from person to person, making its diagnosis quite tricky. It might be asymptomatic, and a person may have cancer without any sign (Quadrelli et al. 2015). Lack of symptoms at early stages leads to a late diagnosis of lungs cancer (Goebel et al. 2019). One of the most important cornerstones of human civilization is maintaining one's health, hence modern approaches to medical issues are required. The amount of information available in the form of lab tests, research papers, clinic reports, and other documents has increased due to advancements in the biomedical area (Riad Alharbey et al. 2022).
A lot of research has been performed by many researchers in different fields so that the accurate prediction and classification of lungs cancer can be increased. In recent studies, Initial screening of disease is performed by exhaled breath analysis which is non-invasive and inexpensive (Nardi-Agmon and Peled 2017). Different methods are used for the prediction of lungs cancer. In the detection process X-rays, CT, and MRI& PET scans are most used.
The classification of lungs cancer in its early stages (Basak and Nath 2017), and the chance of survival of the patient is opposite and inversely proportional to the disease. Tumor size determines the cancer stage. The cancer stage is measured by its spread in the body. The more the spread, the higher the stage. Mostly it’s not quite visible in the early stages; so, detection in the early stage is difficult (Das et al. 2020). But it’s quite easy to deal with it in the early stages as the disease progresses, it becomes complex to cope with it Fig. 2. Depicts the stages of cancer.
Analysis of visual images is an efficient way to investigate the lungs tissues identify the stages of lungs cancer and classify these stages. However, it is difficult to categorize it by stages.However, by the usage of advanced deep learning methods, lungs cancer can be classified accurately. Figure 2 effectively depicts the progression of lungs cancer and categorizes it into stages. Deep learning algorithms are implemented to identify different types of lung cancer and categorize them. The most important and effective method to diagnose and the treatment of lung cancer is made possible by the initial step of disease detection within the lung tissue. Subsequently, various classifiers are used to accurately classify the identified cases into their respective stages Fig. 3. Depicts how classification and prediction of lungs cancer.is performed using deep learning.
Different therapies like chemotherapy and radiotherapy are performed for their treatment, but advanced lungs cancer is quite complex. CT is widely and commonly used for the detection purpose of lungs tumors, but it’s been closely observed that small nodules are mostly not predictive for lungs cancer (Horeweg et al. 2014). These are comparatively tricky and complicated to detect and treat. When it comes to classifying benign & malignant lesions CT has limited ability (Lardinois et al. 2003) Small lesions, having minimal contact with the chest wall are complicated and dealt with technically (Middleton et al. 2006). Chest wall structure, small blood vessels, airway walls, pulmonary structures (Lu et al. 2015), and tissues that are pretty like nodule makes the detection difficult, so it’s rather difficult to perform biopsy that leads to detection. A biopsy is performed for the evaluation of nodules (Lowe et al. 1998). The disease’s complexity and poor CT resolution sometimes lead to re-biopsy. Nodules are classified according to their type, size, and growth rate; it’s essential because it sets the direction of treatment. Nodules can be classified as the cavity, calcified & non-calcified (El-Baz et al. 2013), or as solid, non-solid, partially solid, and calcified (Massimo 2012). A partially solid tumor is a combination of non-solid and solid, the solidified lung tumors consist of a solid internal core. Lung nodules can also be classified as Benign and Malignant (Dhara et al. 2016; Silvestri MD et al. n.d.; Wu et al. 2020; Zhang et al. 2019)
Deep learning is contributing tremendously to healthcare (Esteva et al. 2019; Miotto et al. 2018; Mittal and Hasija 2019). Deep learning paves the way for fast and accurate detection and diagnosis of diseases (Mishra et al. 2020), leading to precise and exact treatment. Research shows that the algorithm of deep learning has shown its significance in the prediction, detection, and diagnosis with the classification of lung cancer; pulmonary nodules are closely observed. By incorporating different deep-learning approaches, the tumor, and the nodule features are captured and classified (Wu and Qian 2019). Big Data refers to incredibly massive data collections that are amenable to analysis to identify trends and patterns. Deep Learning is one method for data analysis that can be utilized to discover abstract patterns in large amounts of data (Gheisari et al. 2017). Advanced data representations and knowledge can be extracted with the help of deep learning (DL). Highly efficient DL methods aid in uncovering more buried information (Gheisari et al. 2023). CAD (Computer-aided design) is used for the screening of cancer in the early stages (Traverso et al. 2017). It is proven as a helping hand for doctors and radiologists (Yuan et al. 2006). State-of-the-art methods have been designed to develop automated processes. Coherent Anti-Stroke Raman Scattering (CARS) technique is used for sensitive investigations (Müller and Zumbusch 2007), and is also there to scan the lungs that capture the molecular movement and produce an image that helps to detect diseases accurately. Researchers have defined different classification models to detect & classify lungs cancer automatically (Nasrullah et al. 2019).
Using a deep learning algorithm, other techniques have been made to read and learn data representation from the unorganized (raw) data. Inner body details are examined, and valuable information is extracted from this data. Deep learning models, algorithms, and methods play a tremendous role in increasing accuracy and decreasing error in the classification of lungs cancer. Deep learning-based automatic segmentation is better than manual in many aspects (Liu et al. 2021) Deep learning helps to avoid misclassification, reduces error rate, provides high-quality images, and accurately predicts cancer. False-positive nodules are filtered out using different classifiers (Jiang et al. 2022). Accurate and high-quality images are directly proportional to the radiologist’s fast and accurate diagnostic decision. Deep learning methods are also incorporated to predict lungs cancer (Banerjee and Das 2021). Training images are provided, and features are extracted automatically. Comparatively, deep learning costs less than conventional CAD frameworks. Deep learning offers HD representation of the given input data, making the detection and identification process efficient and helping the radiologist. The image’s pixel directly contributes to cancer detection, as cancerous and non-cancerous areas are determined on the base of pixels. So, to diagnose accurately and the classification of disease, deep learning assists medical professionals in serving the healthcare system better. It helps to make accurate decisions regarding the disease. CNN design consists of multiple tiers, one of which being Convolutional Layers (CLs). By employing different kinds of convolution filters, the CL layers can extract distinct information from the images of cancer cells that are supplied to them (Manjula Devi et al. 2023). Under the methodology that is being described, the first step in the process is image processing, where preprocessing methods are used to improve the quality of medical images. The improved pictures next go through segmentation, which is an essential stage in identifying pertinent areas within lung imaging. Following identification, the regions are subjected to feature extraction, a process that involves the extraction of significant features to identify crucial patterns suggestive of lung cancer. The classification step, which uses a complex architecture called the Deep Convolutional Neural Network (DCNN), is where the classification process is most centrally located. To dynamically learn hierarchical features, this DCNN is composed of several convolutional layers, each of which has filters, activation functions, and pooling operations. Dense layers known as fully connected layers are another component of the architecture that handles the high-level characteristics that the convolutional layers have learned. The final output shows the findings of the categorization, which differentiates between various lung cancer classifications. The DCNN is an effective technique for accurately classifying lung cancer because its convolutional layers are essential for automatically learning complex patterns.
This study involves the following contributions to the field of medical science especially to the detection of lungs cancer in its early stages:
-
Provides the solution for the detection of lungs cancer in the field of healthcare.
-
Discussed different existing techniques and procedures.
-
Implemented deep learning algorithm and compared with the existing machine learning algorithms and compared the performance with the developed algorithm.
-
The designed technique is implemented on a large dataset and shows how to classify the features.
-
It shows the usability of Convolutional Neural Networks (CNN) in the field of artificial intelligence.
-
For future research it provides useful implementation and development techniques for the early detection of cancer disease.
The other parts of the literature survey are defined in the following sections. Imaging techniques for lungs cancer detection are presented in Section 2. Section 3 includes the latest trends in lungs cancer detection, Section 4 offers the Research methodology of the opted research and the process of selecting research articles is given in Section 4. Section 5 refers to the deep learning contribution towards lungs cancer classification. Section 6 refers to the Literature sources, and the research community’s contribution to the current field covering primary techniques and models used to classify, detect, and predict lungs cancer. Results that are obtained from the selected and extracted data are presented in Section 7. Section 8 refers to the conclusion in which the state-of-the-art deep learning contribution towards lungs cancer classification is presented.
2 Imaging techniques for detection of lungs cancer
Different screening approaches are employed for the identification and screening of lung cancer (Schaefer-Prokop and Prokop 2002). These aid in the doctor's ability to see internal bodily processes and to gain an understanding of how internal organs function. To check for lung anomalies. There are several multimodality imaging techniques including positron emission tomography (PET), computer tomography (CT), Ultrasound, chest radiography (X-Ray), and magnetic resonance imaging (MRI) scans (Laal 2013; Tariq Hussain n.d.) Fig. 4 shows a few techniques of Imaging.
3 Latest trends in lungs cancer detection
The primary purpose of the designed research is to demonstrate the methods and strategies employed in the deep learning categorization of lung cancer. Deep Learning algorithm is the most recent technique that helps medical professionals diagnose diseases and helps radiologists find difficult-to-diagnose conditions like lung cancer. The chosen articles demonstrate the most recent deep learning algorithms and their efficacy in the prediction and categorization of cancer. This paper presents different techniques defined in deep learning algorithms and concepts for this (Fig. 5).
Convolutional Neural Network (CNN) consists of multiple layers. Convolutional layer that extracts features of image pooling layer which selects the feature. The third one is the fully connected or FC layer, its work is to combine those extracted features. Recurrent Neural Network (RNN) is suitable for sequential data and is mostly used for audio, video, and text. Deep Belief Network (DBN) consists of multiple RBMs. These are probabilistic generative models. DBN has many variants. Support Vector Machine (SVM) is a statistical theory-based algorithm. Artificial Neural Network (ANN) is structured just like human brains in which neurons are involved, that’s why it is known as a biological-inspired network. Deep Neural Network (DNN) is a new and advanced technique in the field of artificial intelligence, as it can also work for a complex nonlinear relationship. DNA-binding proteins have a close relationship with several human disorders, including AIDS, cancer, and asthma (Ali et al. 2022b). DBP-DeepCNN would be beneficial in developing more promising therapeutic approaches for the management of chronic diseases (Ali et al. 2022a) while patients with chronic depressive illness experience confusion in their social lives (Gheisari 2016). Integrating CC technology with wireless body area networks WBANs systems to create sensor-cloud infrastructure (S-CI) is helping the healthcare industry by enabling early detection of diseases and real-time patient monitoring (Masood et al. 2023, 2018a) while patient privacy should preserved (Masood et al. 2018b). If a deep learning model is developed well, it may help prevent misdiagnosis and waste of time (Javed et al. 2023). Deep machine learning could be applied to the initial processing of images, the segmentation of images to emphasize the diagnostic objects under investigation, and the classification of these objects to ascertain their benign or malignant nature (Jamshaid Iqbal Janjua et al. 2022). It is challenging to predict human diseases, especially cancer, in order to deliver more effective and timely care. Cancer is a potentially deadly disease that affects the human body's many organs and systems (Abbas et al. 2023b)
4 Research methodology
For the selected research, a mapping study “Classification (lungs cancer)” analysis is chosen as a research methodology. Figure 6 illustrates the mapping process that has been followed. It consists of three steps that are as follows:
-
Step-I: Study Preparation
-
Step-II: Conduct of Study
-
Step-III: Analysis and Results of the Study
In this study, a mapping study methodology is employed to conduct a systematic exploration of the literature on lung cancer classification using deep learning approaches. The mapping process, illustrated in Fig. 5, is structured into three distinct steps: Study Preparation, Conduct of Study, and Analysis and Results of the Study. The main contribution of this paper is described below.
-
Conducted a comprehensive Systematic Literature Review (SLR) using deep learning approaches, which included a detailed analysis of pertinent literature in the field of lung cancer.
-
Categorized and synthesized the overall methodologies observed in the literature, offering readers a systematic overview of the strategies adopted in the domain of deep learning for lung cancer detection and analysis
-
Outlined the current state-of-the-art and the latest advancements in deep learning methodologies applied to lung cancer research, providing insights into cutting-edge techniques and emerging trends in the field.
-
Conducted a comprehensive Quality Assessment of the approaches used in the examined papers, guaranteeing a strong assessment framework to gauge the validity and dependability of the deep learning methods applied in lung cancer research.
-
Provided a comprehensive overview of deep learning methodologies specifically tailored to lung cancer research, consolidating the collective knowledge and advancements in the area for the benefit of researchers, practitioners, and stakeholders.
Study preparation, the first step, entails defining the research's scope, creating inclusion and exclusion criteria, and choosing a search technique. The implementation of the literature search, data extraction, classification, and synthesis of pertinent literature are then included in the Conduct of Study phase. Lastly, analyzing the identified literature, gauging the caliber of the included research, and extracting significant findings to guide the systematic review are all part of the Analysis and Results of the Study phase.
4.1 Research objectives
The main purpose of this research is to provide the scientific community with a systematic step-by-step review of the current research on lungs cancer by using the technique known as deep learning just like the recurrent neural networks (RNN), deep belief network (DBN), support vector machine (SVM), convolution neural networks (CNN) and the deep neural networks (DNN) etc.
4.2 Research questions
As part of this procedure, the questions related to this research are listed in Table 1 and are defined step by step to provide a more thorough understanding of the investigation. These research questions are accompanied by their motivations.
4.3 Search scheme
Following databases and scientific resources have been searched to get and gather the most relevant research papers and articles IEEE Digital Library, Springer, Elsevier, ACM Digital Library, Science Direct, and Google Scholar are the main repositories that were used to get the most relevant research articles.
4.4 Search string
The following search string was used to conduct the automatic search in the selected databases/scientific sources.
(“Classification” OR “Detection” OR “Prediction” OR “Diagnosis” OR “Analysis”) AND (“Lungs Cancer” OR “Lung Cancer” OR “Pulmonary Nodule” OR “Lungs Tumor” OR “Lung Nodule”) AND “Deep Learning” also known as “Deep Neural Network” alternate “DNN” also written as “DL”
4.5 Study selection procedure
The selection procedure is focused on identifying and recognizing those articles that effectively meet the goal of the study. These articles have been searched and gathered from different sources, so if the article is present in more than one source it is counted just once. After comprehensively investigating and examining titles, abstracts, and keywords, each paper is evaluated and its candidature in a study is determined. The search string is considered in deciding the inclusion and exclusion criteria. Duplicates are removed and articles not observing the search string are excluded.
4.6 Inclusion & exclusion principles for the research studies
For the chosen research Table 2 listed the inclusion and exclusion principles. Articles from journals focused on the classification of lungs cancer where deep learning algorithms, are incorporated, and published between 2015-2024 are collected While Articles that are focused on other types of cancer and do not incorporate deep learning are not included.
Research articles are collected from different geographical locations through a combination of online databases, we gathered articles for our study from various geographic places. Using precise terms and search parameters linked to our research topic, we conducted in-depth searches on academic databases including IEEE, Nature, Google Scholar, etc. Table 3 depicts the geographical locations of the articles selected for the study.
The research process is conducted according to the given flow diagram in Fig. 6, which depicts the steps of gathering the research material, from identifying articles to selecting articles for further analysis.
It starts by gathering articles from well-reputed databases. Then the overall number is calculated. After that, the duplicate articles are removed, and initial screening is performed.
Articles that are not in the English language are excluded, and further assessment is performed in this step different criteria of exclusion are applied. The articles that are not from journals are excluded, conference papers, papers published before 2015, papers that are focused on other types of cancer, and articles that are not focused on deep learning are excluded. After excluding the papers with justification, 66 articles were chosen for further investigation.
Figure 7 displays a graphical depiction of the scientific databases where the search term was used, and articles were chosen.
5 Quality assessment of study
Quality assessment is important in systematic reviews of literature as it determines the quality of the study that is included.
-
1)
The solution to the problem is presented clearly in the paper. The answer could be yes (+ 1), No (0), somehow (0.5)
-
2)
The contribution of the paper regarding the issue “Classification of lung cancer using deep learning is presented clearly. The answer could be yes (+ 1), No (0), somehow (0.5)
-
3)
Limitations and future study are presented and defined clearly the answer could be yes (+ 1), No (0), somehow (0.5)
-
4)
Result parameters are presented clearly, the answer could be yes (+ 1), No (0), somehow (0.5)
Table 4 presents a detailed Quality assessment score. In which selected articles along with their reference number are presented. These are evaluated based on the solution of the problem, contribution, limitation, future work, and results. Each question possesses one score. A total of 4, each article is evaluated and graded.
Table 5 presents a summary of the total scores. There is one paper that possesses a score of 2, 7 articles with 2.5. There are 13 articles with a score of 3, 25 articles with a score of 3.5, and 20 articles with a score of 4
6 Literature sources
To look over and explore the detection, classification, and prediction of lung cancer using deep learning, 66 relevant articles published by reliable sources were examined.
To ensure the credibility of the systematic literature review, credible journals are selected as sources and data collection. Moreover, a large number of adequate literature surveys exist for that kind of review Fig. 8 portrays the year-wise detail of collected articles.
It’s evident that the selected articles are from 2015 to 2024 and the chart represents that 2020 is the maximum contributing year because the maximum number of articles falls in 2020.
The nature of the review was to present the work done on the topic of the classification of lungs cancer using different deep-learning methodologies. According to the collected data, it is observed that the most frequently used method is the convolutional Neural Network) CNN. Convolution neural networks (CNN) is also a technique used to solve deep learning problems; it consists of multiple layers. CNN is contributing a lot to Image processing and computer vision (Liu et al. 2019a). It is to be noted that the article review was quite prejudiced to the articles published (2015–2024), and the articles that have “Deep Learning” used in titles. It is observed that multiple data sets have been used in research.
This shows that many diversities in data when it comes to training and testing on CNN. MRI, thoracic surgery data, X-ray, CT, PET image data, CARS images, breathing data, thoracic MR images, Whole Slide Imaging (WSI), Lymph Node Slides, H & E slides, and Histopathological. Table 6 presents the summary of the selected research articles. The datasets used in the research, and their descriptions are presented. The parameters that decide the form of classification, applied method or approach, and the feature extraction techniques used are provided after thoroughly examining the selected research articles.
Cancer image data is collected and presented in forms. It is observed that a CT scan is the most frequently used type of data.
The management of input image sizes is a critical consideration in deep learning field for lung CT-scan processing. Some models require a specific input size, which calls for preprocessing operations like scaling or cropping to fit the data into this preset dimension. Such modifications, however, run the danger of information loss or image distortion, which reduces the model's effectiveness. In contrast, to solve these problems more advanced and sophisticated methods are developed. Models can adjust to different image sizes using techniques like padding or spatial pyramid pooling, but they may also introduce noise or artifacts. The idea of picture pyramids also produces numerous image copies at varying scales, facilitating feature capture at varied levels of detail. Fully Convolutional Networks (FCNs) are also specially designed to support arbitrary input sizes. The decision between these methods depends on the particular deep learning architecture and the actual requirements of the task of medical image analysis, taking into account elements like computational complexity and the requirement to adapt to various input dimensions.
The process of lungs cancer/nodules detection or classification goes as follows: -
Pre-Processing: it is the first step which is used to take inputs in the form of MRI, Thoracic Surgery Data, X-ray, CT, PET Image Data, Breathing Data, Thoracic MR Images, Whole Slide, Imaging (WSI), Lymph Node Slides, Histopathological Cancer Images, CARS Images, H&E. Different Image processing/ feature extraction techniques are applied on the Input. Table 7 lists the various data types used in the reviewed studies.
Computed Tomography is used most of the time, because it’s been frequently used in our review techniques used for preprocessing includes FPSOCNN, Wavelet-Transform-Based Features, SIFT, LBP, ABF Zernike Moment, and HE, inceptionv3, LDA, ODNN, Novel Nodule Candidate Detection Method, Single-level discrete Two-Dimensional Wavelet Transform (2D-DWT), Fractional-Order Darwinian Particle Swarm Optimization, Two-Dimensional Discrete Fourier transform (2D-DFT), U-Net, Back-Propagation, Deep Learning Architectures, Gradient Descent, SIFT + LBP, Swarm Algorithms, Bipartite Undirected Graphical Models (RBMs), including Alex, Deep (ConvNet),Intensity features + SVM Support Vector Machines, Transfer Learning Networks, Hybrid Geometric Texture Feature Descriptor, FODPSO CARS, VGG16,Wiener Filter, VGG19, Three-Dimensional (3D) CNN model, CNN Region-Of-Interest (ROI), Median Filter, Gaussian Filter, Gabor Filter, Knowledge-Based Collaborative (KBC) sub-mode, ResNet-50 networks, ProNet, RadNet, UB open-source software ITK-Snap, 3D Stereoscopic Planning System, IMR, Maximum Intensity Projection Technique, Based On Lung-RADS version 1.1, Three Reconstruction Kernels. Multi-Scale Dilated Residual Representation Block Size-Related (SR-DiRes) &Multi-Mask Convolution Representation Block (ConvRB), Multi-Scale LBP, AP Algorithms, Affinity Propagation (AP) Algorithm, Deep Convolution Neural Network (DCNN) Architecture, Radiomics-Based Analysis, Hot-Spot ROI-based DCE Kinetic Analysis, Deep Dueling Q-network, Hierarchical Deep Q-Networks, Radiomics Deep Q-Network, Weighted Mean Histogram Equalization, Dense PriNet, Multi-and single model method, Pixel-Wise Segmentation based On FCN ,Multi-Model, Weakly Supervised Learning Methods, HALO Tissue Classifier Analysis Module (Random Forest Algorithm), HALO AI (CNN, VGG network),Context-Aware Feature Selection And Aggregation, Graph Regularized Sparse MTL,K-means Algorithm, Extensive Data Augmentation,LUNA16 Pulmonary Nodule Annotations, Multi-Group Patch-Based Learning System, Diffusion (VED) and a Vessel filter, Integrated Deep Learning , Region Growth Algorithm, fuzzy logic and a NN, HE hybrid learning algorithm, ACC, VEL, Correlation Analysis Algorithm, PCA, (GSEA)Gene-Set Enrichment Analysis, ReLU Activation Function, CVAE-GAN, DenseNet using CoxPH, CNN model, ImageNet, Score-CAM Maximum Intensity Projection (MIP).,Multi-Scale Convolution Image Feature, Novel 15-Layer 2D Deep CNN architecture Hyperparameter Tuning, vector Quantization(VQ) algorithm, Gaussian noise model-based collaborative Wiener Filtering (GNM-CWF),Self-Adaptive Online VQ algorithm Residual Learning Denoising Model (DR-Net),Feature fusion strategy called DCA, Curriculum Learning, And Transfer Learning, Rival Convolutional Neural Network Models (setio-CNN and Over Feat), Taguchi-based CNN, combination of Deep Residual Learning, The Trial And Error Method, ACL algorithm, Converged Search and Rescue (CSAR) Algorithm, Novel Wilcoxon Signed-Rank Gain Preprocessing, Hybrid Spiral Optimization Intelligent-Generalized Rough Set approach, AI-based noninvasive radiomics biomarkers, SVM Principal Component Analysis (PCA), Multilevel Brightness-Preserving approach,. Numerous segmentation methods and approaches are then applied.
Segmentation contributes to the feature extraction process; it prepares the extracted data for classification.
Classification: - Classification is the phase where the prepared, feature extracted and segmented data is classified into different categories that may be abnormal or normal, benign, or malignant, LUAD or LUSC. After the detection of cancer nodules or pulmonary nodules which confirm that the disease is present or not of lungs cancer further classification is performed to classify these nodules into solid, calcified, partially solid, perifissural, and speculated.
Several classifiers are used in our study which includes; Fuzzy Particle Swarm Optimization (FPSO),Confusion matrix, Forest Classifiers, Back-Propagation, SVM or Naïve Bayes classifiers, Cross Entropy loss and Transfer Learning, RMS Prop-optimization method, Modified Gravitational Search Algorithm (MGSA),Multi-Channel CNN, Deep Learning and Swarm Intelligence, DBN and CNN Deep Learning, Convolutional Network Architecture, FODPSO, CARS & Deep learning, GoogleNet Inception v3 CNN architecture, Deep Learning Approach Based On Stacked Autoencoder and Softmax, CADe systemVGG19architectureand SVM classifier, Multi-View Knowledge-Based Collaborative (MV-KBC) Deep Model, Long Short-Term Memory/LSTM, Recurrent Neural Network/RNN,CNN Cancer Risk Prediction Model, triplet, DCNN AlexNet, watershed segmentation Banarization, Auto Encoder/AE &General Adversarial Networks/GAN, Deep Belief Network/DBN, Random Forest Classifier, Deep Quality Model, GG-net architecture, Convolutional Neural Networks with a U-net architecture, Recurrent Neural Networks (RNN), 3D Deep Learning and Radiomics, Core-Ring Blocks Residual Estimation with size-related damper block deep prediction model, GG-net architecture, Conditional Random Framework, SVM anti-PD-1 response prediction by H&E,DCNN CAD system, DCE Kinetic Parameters, Convolutional Long Short Term Memory (CLSTM) Network, DCE-MRI,Value-based Reinforcement Learning Approach, DSRL, Deep Successor Q-Network, Profuse Clustering Technique (IPCT),Deep. Learning Instantaneously Trained Neural Network (DITNN), deep Q-network and hierarchical deep Q-network, Explosion-Trained Deep Learning Neural Network (DITNN),CADe / CADx models, Deep Learning Classifier (Lymphoid Follicle CNN - LFCNN),3D Neural Network, Fully Convolutional Network (FCN),FCN, ScanNet, proportion-SVM, Adaptive Hierarchical Heuristic Mathematical Model (AHHMM),K-mean algorithm, 3D fully CNN based on the V-Net architecture, Four-Channel CNN, corrective lung contour, Wilcoxon Signed Generative DL (WS-GDL)Hyper-Parameter Tuning Algorithm, Multi-Scene Deep Learning Framework (MSDLF),Normalized Spherical Sampling, LSTM algorithms, NFNet, Fast R-CNN, Intra- And Inter-Fraction Fuzzy Deep Learning (IIFDL),deep learning-based radiogenomic framework-net, EfficientNet, CoxPH and CoxCC, LdcNet, Converged Search and Rescue Algorithm, Deep Gaussian Mixture Model in region-based CNN[DGMM-RBCNN], Lung-Deep System, Novel Nodule CADeCNN, INC classification, (DFD-Net), Two-path CNN with feature fusion DFD-Nets, Curriculum learning, 3D (DCNNs), Taguchi-based CNN, Fuzzy C-Ordered Means (FCOM) with ECN, Enhanced Capsule Networks (ECN), Lung Cancer Prediction (LCP-CNN), Generative Deep Learning, Survival Neural Network model, machine learning peculated known as k-nearest, KNN neighbors and SVM on CNN Ensemble Classifier and Improved DNN.
7 Literature synthesis: unveiling patterns and insights
Deep Convolutional Neural Networks (DCNN) is the best technique to detect lungs cancer in the field of machine learning. The highest levels of accurate lung cancer case classification were continuously attained using DCNN. The ability to make accurate diagnoses, a critical component in healthcare, is shown by this improved accuracy. Additionally, DCNN demonstrated exceptional specificity, reducing the incidence of false positives. In healthcare contexts, this precision is crucial since it lessens the possibility of misdiagnosing non-cancerous patients as cancer. The remarkable sensitivity of DCNN enabled the identification of sizable actual lung cancer cases. The potential for early identification and intervention, which are essential for enhancing patient outcomes, is increased by this high sensitivity. Finally, DCNN demonstrated impressive accuracy in detecting lung cancer. Finally, DCNN demonstrated remarkable accuracy in diagnosing lung cancer, reducing the possibility of incorrect diagnosis. These findings demonstrate how well DCNN performs in automatically extracting complex patterns from medical images, which helps in the accurate and reliable identification of lung cancer. DCNN stands out as the leading machine learning technology, to improve the accuracy and reliability of lung cancer detection in clinical practice due to its exceptional performance in accuracy, specificity, sensitivity, and precision. To promote medical diagnostics in this crucial area, we encourage continued investigation and development of DCNN-based techniques.
In the study, 66 research articles were carefully examined and their methods were analyzed. In Table 8, the results of this evaluation procedure are collated and summarized. Important details including reference numbers, research methodology, and performance measures like the F1 score, accuracy, precision, sensitivity, and AUC are included in this table. Table 8 is a useful tool for comparing and assessing the research findings because these metrics are significant indicators of the efficacy and dependability of the various methodologies covered in the examined papers. The major limitation of current studies is mostly the size of the data sets. There is potential for improvement because there isn't a finished product or global standard for cancer detection and prediction. To ascertain the accuracy of these models, researchers need to gather up-to-date and new data, employ various deep learning and machine learning techniques, and combine new and old data. Early cancer detection can benefit millions of people. To detect cancer, there is no established standard or finished product. While deep learning has promising opportunities for lung cancer detection, there are important gaps that need to be filled. The generalizability of current findings is frequently problematic while accurate feature extraction is also crucial to be handled. Certain populations in the actual world may not respond well to models that were trained on their particular datasets. Furthermore, characterization may be eclipsed by an emphasis on detection. Some models are quite good at detecting nodules, but they may not give information about the tumor, which makes it more difficult to diagnose patients early and accurately. In addition, the "black box" nature of sophisticated deep learning models and worries about data security and privacy persist, making it challenging to comprehend how these models make decisions. Most of the research frequently runs into issues with feature extraction, making it difficult to identify pertinent elements that are essential for accurate prediction. These restrictions make it more difficult to do the thorough study needed to produce accurate and trustworthy detection results. In order to advance deep learning models' efficacy and dependability in lung cancer detection and eventually enhance patient outcomes and diagnostic accuracy, it is imperative that these issues be resolved. Despite these shortcomings, scientists are working hard to fill in the gaps Closing the existing gaps is critical to the future of deep learning for lung cancer detection. Researchers are focusing on tumor characterization rather than merely detection, enhancing generalizability through transfer learning, and advanced approaches. Strong data security protocols and resolving any biases in training data are also essential. Lastly, to stay up to date with the changing features of cancer, ongoing learning will be crucial. Deep learning has the potential to transform lung cancer detection and result in earlier diagnoses and better patient outcomes by overcoming these obstacles.
Different algorithms used for classification and detection purposes perform differently in terms of performance CNN surpasses the others.
8 Conclusion
Lung cancer is a serious and sometimes fatal disease that necessitates early discovery to effectively treat it. This work emphasizes how deep learning—specifically, the application of Convolutional Neural Networks, or CNN—is essential to changing the face of medical diagnosis. Deep learning techniques reduce the workload of healthcare professionals by automating the identification and categorization of lung cancer. This technological advancement improves the precision and efficacy of diagnosis. In the context of lung cancer research, this paper offers an extensive Systematic Literature Review (SLR) that makes use of deep learning approaches. It benefits researchers, practitioners, and stakeholders by classifying and synthesizing methodologies, outlining the state-of-the-art, conducting a thorough Quality Assessment, and offering a customized overview of deep learning approaches for lung cancer research. Our thorough analysis examines the several deep learning methods used in the identification and categorization of lung cancer across a range of imaging modalities, including MRIs, CT scans, and X-rays. Notably, we ensure a current overview by concentrating on the years 2015 to 2024 and obtaining information from credible journals. The study highlights the critical function that Deep Convolutional Neural Networks (DCNN) fulfill in the field of deep learning techniques. In particular, DCNN is the recommended method in the Convolutional Neural Network (CNN) architecture because of its unique multi-layered architecture, which makes direct feature learning from lung nodule images possible. The main reason why DCNN is so effective at obtaining the best accuracy is that it has the innate capacity to learn weights automatically. These models can significantly boost the efficiency and accuracy of lung cancer classification by the automatic learning of pertinent features, which improves diagnostic procedures in clinical settings. The research thoroughly assesses several performance measures, such as AUC, sensitivity, specificity, accuracy, and precision. When it comes to the classification of lung cancer, DCNN frequently performs more accurately than other algorithms. Even with these noteworthy successes, there are still difficulties, especially when managing large-volume datasets. This study presents the classification process on multiple data sets and multiple classifiers are used. The focus of the study is to present different deep learning algorithms and approaches to classify lungs cancer Nevertheless, there are still some challenges that still exist, most of them are related to the size of datasets. This study will help the researchers to better understand the existing deep-learning techniques and procedures to classify lungs cancer. This work lays the groundwork for future advancements in the field by providing a comprehensive grasp of the state-of-the-art deep learning methods for the categorization of lung cancer. Lung cancer diagnosis could be revolutionized by deep learning, however there remain obstacles because of the scale of data sets and the absence of an international standard. Current data must be gathered, deep learning and machine learning methods must be used, and both new and old data must be combined. Generalizability, characterization, data security, and privacy issues are lacking, nonetheless. Notwithstanding these obstacles, researchers are concentrating on tumor characterization, improving generalizability via transfer learning, and addressing biases in training data in an effort to close these gaps. Continual learning is also crucial to stay updated with changing cancer features.
References
Abbas S, Issa GF, Fatima A et al (2023) Fused Weighted Federated Deep Extreme Machine Learning Based on Intelligent Lung Cancer Disease Prediction Model for Healthcare 5.0. Int J Intell Syst 2023:1–14. https://doi.org/10.1155/2023/2599161
Abbas T, Fatima A, Shahzad T et al (2023) Secure IoMT for Disease Prediction Empowered With Transfer Learning in Healthcare 5.0, the Concept and Case Study. IEEE Access 11:39418–39430. https://doi.org/10.1109/access.2023.3266156
Abbas Q (2017) Lung-Deep: a Computerized Tool for detection of lung nodule patterns using deep learning algorithms detection of lung nodules patterns. Int J Adv Comput Sci Appl 8. https://doi.org/10.14569/ijacsa.2017.081015
Affonso C, de Nedjah NL (2020) Detection and classification of pulmonary nodules using deep learning and swarm intelligence. Multimedia Tools Appl 79:15437–15465. https://doi.org/10.1007/s11042-019-7473-z
Alharbey R, Kim JI, Daud A et al (2022) Indexing important drugs from medical literature. Scientometrics 127:2661–2681. https://doi.org/10.1007/s11192-022-04340-7
Ali F, Kumar H, Patil S et al (2022) DBP-DeepCNN: Prediction of DNA-binding proteins using wavelet-based denoising and deep learning. Chemom Intell Lab Syst 229:104639–104639. https://doi.org/10.1016/j.chemolab.2022.104639
Ali F, Kumar H, Patil S et al (2022) Target-DBPPred: An intelligent model for prediction of DNA-binding proteins using discrete wavelet transform based compression and light eXtreme gradient boosting. Comput Biol Med 145:105533–105533. https://doi.org/10.1016/j.compbiomed.2022.105533
Ardila D, Kiraly AP, Bharadwaj S et al (2019) Author Correction: End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography. Nat Med 25:1319–1319. https://doi.org/10.1038/s41591-019-0536-x
Asuntha A, Srinivasan A (2020) Deep learning for lung Cancer detection and classification. Multimedia Tools Appl 79:7731–7762. https://doi.org/10.1007/s11042-019-08394-3
Banerjee N, Das S (2021) Lung cancer prediction in deep learning perspective
Basak P, Nath A (2017) Detection of different stages of lungs cancer in CT-scan images using image processing techniques. Int J Innov Res Comput Commun Eng 2320–9798
Chae KJ, Jin GY, Ko SB et al (2020) Deep Learning for the Classification of Small (≤2 cm) Pulmonary Nodules on CT Imaging: A Preliminary Study. Acad Radiol 27:e55–e63. https://doi.org/10.1016/j.acra.2019.05.018
Chaunzwa TL, Hosny A, Xu Y et al (2021) Deep learning classification of lung cancer histology using CT images. Sci Rep 11. https://doi.org/10.1038/s41598-021-84630-x
Chen W, Zheng R, Baade PD et al (2016) Cancer statistics in China, 2015. CA: A Cancer J Clin 66:115–132. https://doi.org/10.3322/caac.21338
Ciompi F, Chung K, van Riel SJ et al (2017) Towards automatic pulmonary nodule management in lung cancer screening with deep learning. Scientific Reports 7. https://doi.org/10.1038/srep46479
Cook RM, Miller YE, Bunn PA (1993) Small cell lung cancer: etiology, biology, clinical features, staging, and treatment. Curr Probl Cancer 17:69–141. https://doi.org/10.1016/0147-0272(93)90010-y
Corner J (2005) Is late diagnosis of lung cancer inevitable? Interview study of patients’ recollections of symptoms before diagnosis. Thorax 60:314–319. https://doi.org/10.1136/thx.2004.029264
Coudray N, Ocampo PS, Sakellaropoulos T et al (2018) Classification and mutation prediction from non–small cell lung cancer histopathology images using deep learning. Nat Med 24:1559–1567. https://doi.org/10.1038/s41591-018-0177-5
Cui X, Zheng S, Heuvelmans MA et al (2022) Performance of a deep learning-based lung nodule detection system as an alternative reader in a Chinese lung cancer screening program. Eur J Radiol 146:110068–110068. https://doi.org/10.1016/j.ejrad.2021.110068
Das P, Das B, Dutta HS (2020) Prediction of lungs cancer using machine learning
Dhara AK, Mukhopadhyay S, Dutta A et al (2016) A Combination of Shape and Texture Features for Classification of Pulmonary Nodules in Lung CT Images. J Digit Imaging 29:466–475. https://doi.org/10.1007/s10278-015-9857-6
Doppalapudi S, Qiu RG, Badr Y (2021) Lung cancer survival period prediction and understanding: Deep learning approaches. Int J Med Informatics 148:104371. https://doi.org/10.1016/j.ijmedinf.2020.104371
El-Baz A, Elnakib A, Abou El-Ghar M et al (2013) Automatic Detection of 2D and 3D Lung Nodules in Chest Spiral CT Scans. Int J Biomed Imaging 2013:1–11. https://doi.org/10.1155/2013/517632
Elnakib A, Amer HM, Abou-Chadi FE (2020) Early Lung Cancer Detection using Deep Learning Optimization. Int J Online Biomed Eng (iJOE) 16:82. https://doi.org/10.3991/ijoe.v16i06.13657
Esposito L, Conti D, Ailavajhala R et al (2010) Lung Cancer: Are we up to the Challenge? Curr Genom 11:513–518. https://doi.org/10.2174/138920210793175903
Esteva A, Robicquet A, Ramsundar B et al (2019) A guide to deep learning in healthcare. Nat Med 25:24–29. https://doi.org/10.1038/s41591-018-0316-z
Field JK, Oudkerk M, Pedersen JH, Duffy SW (2013) Prospects for population screening and diagnosis of lung cancer. The Lancet 382:732–741. https://doi.org/10.1016/s0140-6736(13)61614-1
Gheisari M, Wang G, Alam Bhuiyan MdZ (2017) A Survey on deep learning in big data. https://doi.org/10.1109/CSE-EUC.2017.215
Gheisari M, Ebrahimzadeh F, Rahimi M et al (2023) Deep learning: applications, architectures, models, tools, and frameworks: a comprehensive survey. CAAI Trans Intell Technol. https://doi.org/10.1049/cit2.12180
Gheisari M (2016) The Effectiveness of schema therapy integrated with neurological rehabilitation methods to improve executive functions in patients with chronic depression. Health Sci J 10
Goebel C, Louden CL, Mckenna R et al (2019) Diagnosis of Non-small Cell Lung Cancer for Early Stage Asymptomatic Patients. Cancer Genom - Proteom 16:229–244. https://doi.org/10.21873/cgp.20128
Gu Y, Chi J, Liu J et al (2021) A survey of computer-aided diagnosis of lung nodules from CT scans using deep learning. Comput Biol Med 137:104806. https://doi.org/10.1016/j.compbiomed.2021.104806
Guo Y, Song Q, Jiang M et al (2021) Histological Subtypes Classification of Lung Cancers on CT Images Using 3D Deep Learning and Radiomics. Acad Radiol 28:e258–e266. https://doi.org/10.1016/j.acra.2020.06.010
Heuvelmans MA, van Ooijen PMA, Ather S et al (2021) Lung cancer prediction by Deep Learning to identify benign lung nodules. Lung Cancer 154:1–4. https://doi.org/10.1016/j.lungcan.2021.01.027
Hoang Ngoc Pham H, Futakuchi M, Bychkov A et al (2019) Detection of Lung Cancer Lymph Node Metastases from Whole-Slide Histopathologic Images Using a Two-Step Deep Learning Approach. Am J Pathol 189:2428–2439. https://doi.org/10.1016/j.ajpath.2019.08.014
Horeweg N, van Rosmalen J, Heuvelmans MA et al (2014) 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 15:1332–1341. https://doi.org/10.1016/S1470-2045(14)70389-4
Hu J, Cui C, Yang W et al (2021) Using deep learning to predict anti-PD-1 response in melanoma and lung cancer patients from histopathology images. Translational Oncol 14:100921. https://doi.org/10.1016/j.tranon.2020.100921
Hurria A, Kris MG (2003) Management of Lung Cancer in Older Adults. CA: A Cancer J Clin 53:325–341. https://doi.org/10.3322/canjclin.53.6.325
Hussein S, Kandel P, Bolan CW et al (2019) Lung and Pancreatic Tumor Characterization in the Deep Learning Era: Novel Supervised and Unsupervised Learning Approaches. IEEE Trans Med Imaging 38:1777–1787. https://doi.org/10.1109/tmi.2019.2894349
Jamshaid Iqbal Janjua, Tahir Abbas Khan, Nadeem M (2022) Chest x-ray anomalous object detection and classification framework for medical diagnosis. 2022 International conference on information networking (ICOIN). https://doi.org/10.1109/icoin53446.2022.9687110
Javed R, Abbas T, Jamshaid Iqbal Janjua et al (2023) wrist fracture prediction using transfer learning, a case study. J Popul Ther Clin Pharmacol 30. https://doi.org/10.53555/jptcp.v30i18.3161
Jena SR, George ST, Ponraj DN (2021) Lung cancer detection and classification with DGMM-RBCNN technique. 33:15601–15617. https://doi.org/10.1007/s00521-021-06182-5
Jiang H, Ma H, Qian W et al (2017) An Automatic Detection System of Lung Nodule Based on Multigroup Patch-Based Deep Learning Network. IEEE J Biomed Health Inform 22:1227–1237. https://doi.org/10.1109/JBHI.2017.2725903
Jiang W, Zeng G, Wang S et al (2022) Application of Deep Learning in Lung Cancer Imaging Diagnosis. J Healthc Eng 2022:1–12. https://doi.org/10.1155/2022/6107940
Jung H, Kim B, Lee I et al (2018) Classification of lung nodules in CT scans using three-dimensional deep convolutional neural networks with a checkpoint ensemble method. BMC Med Imaging 18. https://doi.org/10.1186/s12880-018-0286-0
Kumar V, Bakariya B (2021) Classification of malignant lung cancer using deep learning. J Med Eng Technol 45:85–93. https://doi.org/10.1080/03091902.2020.1853837
Kumar Swain A, Swetapadma A, Kumar Rout J, Kumar Balabantaray B (2024) Classification of non-small cell lung cancer types using sparse deep neural network features. Biomed Signal Process Control 87. https://doi.org/10.1016/j.bspc.2023.105485
Laal M (2013) Innovation Process in Medical Imaging. Procedia Soc Behav Sci 81:60–64. https://doi.org/10.1016/j.sbspro.2013.06.388
Lakshmanaprabu SK, Mohanty SN, Shankar K et al (2019) Optimal deep learning model for classification of lung cancer on CT images. Future Gener Comput Syst 92:374–382. https://doi.org/10.1016/j.future.2018.10.009
Lang N, Zhang Y, Zhang E et al (2019) Differentiation of spinal metastases originated from lung and other cancers using radiomics and deep learning based on DCE-MRI. Magn Reson Imaging 64:4–12. https://doi.org/10.1016/j.mri.2019.02.013
Lanjewar MG, Panchbhai KG, Charanarur P (2023) Lung Cancer detection from CT scans using modified DenseNet with feature selection methods and ML classifiers. Expert Syst Appl 119961. https://doi.org/10.1016/j.eswa.2023.119961
Lardinois D, Weder W, Hany TF et al (2003) Staging of Non–Small-Cell Lung Cancer with Integrated Positron-Emission Tomography and Computed Tomography. N Engl J Med 348:2500–2507. https://doi.org/10.1056/nejmoa022136
Li Y, Zhang L, Chen H, Yang N (2019) Lung Nodule Detection With Deep Learning in 3D Thoracic MR Images. IEEE Access 7:37822–37832. https://doi.org/10.1109/access.2019.2905574
Li Z, Zhang J, Tan T et al (2020) Deep learning methods for lung cancer segmentation in whole-slide histopathology images -- the ACDC@LungHP Challenge 2019. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2008.09352
Lin C-J, Li Y-C (2020) Lung Nodule Classification Using Taguchi-Based Convolutional Neural Networks for Computer Tomography Images. Electronics 9:1066. https://doi.org/10.3390/electronics9071066
Liu Y, Wang H, Gu Y, Lv X (2019) Image classification toward lung cancer recognition by learning deep quality model. J Vis Commun Image Represent 63:102570. https://doi.org/10.1016/j.jvcir.2019.06.012
Liu Z, Yao C, Yu H, Wu T (2019) Deep reinforcement learning with its application for lung cancer detection in medical Internet of Things. Futur Gener Comput Syst 97:1–9. https://doi.org/10.1016/j.future.2019.02.068
Liu S, Yao W (2022) Prediction of lung cancer using gene expression and deep learning with KL divergence gene selection. BMC Bioinformatics 23. https://doi.org/10.1186/s12859-022-04689-9
Liu X, Li K-W, Yang R, Geng L-S (2021) Review of deep learning based automatic segmentation for lung cancer Radiotherapy. Front Oncol 11. https://doi.org/10.3389/fonc.2021.717039
Lowe VJ, Fletcher JW, Gobar L et al (1998) Prospective investigation of positron emission tomography in lung nodules. J Clin Oncol 16:1075–1084. https://doi.org/10.1200/jco.1998.16.3.1075
Lu L, Tan Y, Schwartz LH, Zhao B (2015) Hybrid detection of lung nodules on CT scan images. Med Phys 42:5042–5054. https://doi.org/10.1118/1.4927573
Manjula Devi R, Dhanaraj RK, Pani SK et al (2023) An improved deep convolutionary neural network for bone marrow cancer detection using image processing. Inf Med Unlocked 101233. https://doi.org/10.1016/j.imu.2023.101233
Masood I, Wang Y, Daud A et al (2018) Towards Smart Healthcare: Patient Data Privacy and Security in Sensor-Cloud Infrastructure. Wirel Commun Mob Comput 2018:1–23. https://doi.org/10.1155/2018/2143897
Masood I, Wang Y, Daud A et al (2018) Privacy management of patient physiological parameters. Telematics Inform 35:677–701. https://doi.org/10.1016/j.tele.2017.12.020
Masood I, Daud A, Wang Y et al (2023) A blockchain-based system for patient data privacy and security. Multimedia Tools Appl. https://doi.org/10.1007/s11042-023-17941-y
Massimo B (2012) A classification of pulmonary nodules by CT scan. https://doi.org/10.3332/ecancer.2012.260
Masud M, Sikder N, Nahid A-A et al (2021) A Machine Learning Approach to Diagnosing Lung and Colon Cancer Using a Deep Learning-Based Classification Framework. Sensors 21:748. https://doi.org/10.3390/s21030748
Mayo Clinic (2022) Cancer - Symptoms and Causes. In: Mayo Clinic. https://www.mayoclinic.org/diseases-conditions/cancer/symptoms-causes/syc-20370588
Middleton WD, Teefey SA, Dahiya N (2006) Ultrasound-Guided Chest Biopsies. Ultrasound Q 22:241–252. https://doi.org/10.1097/01.ruq.0000237258.48756.94
Miotto R, Wang F, Wang S et al (2018) Deep Learning for healthcare: review, Opportunities and Challenges. Brief Bioinform 19:1236–1246. https://doi.org/10.1093/bib/bbx044
Mishra S, Dash A, Jena L (2020) Use of deep learning for disease detection and diagnosis. 903. https://doi.org/10.1007/978-981-15-5495-7_10
Mittal S, Hasija Y (2019) Deep Learning Techniques for Biomedical and Health Informatics. Stud Big Data 68:57–77. https://doi.org/10.1007/978-3-030-33966-1_4
Müller M, Zumbusch A (2007) Coherent anti-Stokes Raman Scattering Microscopy. ChemPhysChem 8:2156–2170. https://doi.org/10.1002/cphc.200700202
Nam JG, Park S, Hwang EJ et al (2019) Development and Validation of Deep Learning–based Automatic Detection Algorithm for Malignant Pulmonary Nodules on Chest Radiographs. Radiology 290:218–228. https://doi.org/10.1148/radiol.2018180237
Naqi SM, Sharif M, Jaffar A (2018) Lung nodule detection and classification based on geometric fit in parametric form and deep learning. Neural Comput Appl 32:4629–4647. https://doi.org/10.1007/s00521-018-3773-x
Nardi-Agmon I, Peled N (2017) Exhaled breath analysis for the early detection of lung cancer: recent developments and future prospects. Lung Cancer: Targets Ther 8:31–38. https://doi.org/10.2147/lctt.s104205
Nasrullah, Sang J, Mohammad Khursheed Alam, Xiang H (2019) Automated detection and classification for early stage lung cancer on CT images using deep learning. https://doi.org/10.1117/12.2520333
Nibali A, He Z, Wollersheim D (2017) Pulmonary nodule classification with deep residual networks. Int J Comput Assist Radiol Surg 12:1799–1808. https://doi.org/10.1007/s11548-017-1605-6
Obulesu O, Kallam S, Dhiman G et al (2021) Adaptive diagnosis of lung cancer by deep learning classification using Wilcoxon gain and generator. 2021:1–13. https://doi.org/10.1155/2021/5912051
Oh S, Im J, Kang S-R et al (2021) PET-Based Deep-Learning Model for Predicting Prognosis of Patients With Non-Small Cell Lung Cancer. IEEE Access 9:138753–138761. https://doi.org/10.1109/access.2021.3115486
Ozdemir O, Russell RL, Berlin AA (2020) A 3D Probabilistic Deep Learning System for Detection and Diagnosis of Lung Cancer Using Low-Dose CT Scans. IEEE Trans Med Imaging 39:1419–1429. https://doi.org/10.1109/tmi.2019.2947595
Pandit BR, Alsadoon A, Prasad PWC et al (2022) Deep learning neural network for lung cancer classification: enhanced optimization function. Multimedia Tools Appl. https://doi.org/10.1007/s11042-022-13566-9
Parascandola M, Xiao L (2019) Tobacco and the lung cancer epidemic in China. Translational Lung Cancer Res 8:S21–S30. https://doi.org/10.21037/tlcr.2019.03.12
Park S, Jin Lee S, Weiss E, Motai Y (2016) Intra- and Inter-Fractional Variation Prediction of Lung Tumors Using Fuzzy Deep Learning. IEEE J Translational Eng Health Med 4:4300112. https://doi.org/10.1109/JTEHM.2016.2516005
Parris BA, O’Farrell HE, Fong KM, Yang IA (2019) Chronic obstructive pulmonary disease (COPD) and lung cancer: common pathways for pathogenesis. J Thoracic Dis\ 11:S2155–S2172. https://doi.org/10.21037/jtd.2019.10.54
Pathak H, Manoj Kumar Pandey, Kaur J (2018) Detection and feature extraction of cancer nodules in lung CT image. J Emerging Technol Innov Res
Quadrelli S, Lyons G, Colt H et al (2015) Clinical Characteristics and Prognosis of Incidentally Detected Lung Cancers. Int J Surg Oncol 2015:1–6. https://doi.org/10.1155/2015/287604
Roointan A, Ahmad Mir T, Ibrahim Wani S et al (2019) Early detection of lung cancer biomarkers through biosensor technology: A review. J Pharm Biomed Anal 164:93–103. https://doi.org/10.1016/j.jpba.2018.10.017
Said Y, Alsheikhy AA, Shawly T, Lahza H (2023) Medical Images Segmentation for Lung Cancer Diagnosis Based on Deep Learning Architectures. Diagnostics 13:546–546. https://doi.org/10.3390/diagnostics13030546
Savitha G, Jidesh P (2020) A holistic deep learning approach for identification and classification of sub-solid lung nodules in computed tomographic scans. Comput Electr Eng 84:106626. https://doi.org/10.1016/j.compeleceng.2020.106626
Schaefer-Prokop C, Prokop M (2002) New imaging techniques in the treatment guidelines for lung cancer. Eur Respir J 19:71S-83S. https://doi.org/10.1183/09031936.02.00277902
Shah AA, Malik HAM, Muhammad A et al (2023) Deep learning ensemble 2D CNN approach towards the detection of lung cancer. Sci Rep 13. https://doi.org/10.1038/s41598-023-29656-z
Shakeel PM, Burhanuddin MA, Desa MI (2019) Lung cancer detection from CT image using improved profuse clustering and deep learning instantaneously trained neural networks. Measurement 145:702–712. https://doi.org/10.1016/j.measurement.2019.05.027
Shakeel PM, Burhanuddin MA, Desa MI (2020) Automatic lung cancer detection from CT image using improved deep neural network and ensemble classifier. Neural Comput Appl. https://doi.org/10.1007/s00521-020-04842-6
Shakeel PM, Burhanuddin MA, Desa MI (2022) Automatic lung cancer detection from CT image using improved deep neural network and ensemble classifier. Neural Comput Appl 7731–7762. https://doi.org/10.1007/s00521-020-04842-6
Sharma R (2022) Mapping of global, regional and national incidence, mortality and mortality-to-incidence ratio of lung cancer in 2020 and 2050. Int J Clin Oncol 27: https://doi.org/10.1007/s10147-021-02108-2
Siddiqui EA, Chaurasia V, Shandilya M (2023) Detection and classification of lung cancer computed tomography images using a novel improved deep belief network with Gabor filters. Chemom Intell Lab Syst 235:104763. https://doi.org/10.1016/j.chemolab.2023.104763
Silvestri MD GA, Tanner MD NT, Kearney P et al (n.d.) Assessment of plasma proteomics biomarker’s ability to distinguish benign from malignant lung nodules: results of the PANOPTIC (Pulmonary Nodule Plasma Proteomic Classifier). Trial Chest 154:491–500. https://doi.org/10.1016/j.chest.2018.02.012
Song Q, Zhao L, Luo X, Dou X (2017) Using Deep Learning for Classification of Lung Nodules on Computed Tomography Images. J Healthc Eng 2017:1–7. https://doi.org/10.1155/2017/8314740
Sori WJ, Feng J, Godana AW et al (2020) DFD-Net: lung cancer detection from denoised CT scan image using deep learning. Front Comput Sci 15: https://doi.org/10.1007/s11704-020-9050-z
Strak M, Janssen N, Beelen R et al (2017) Associations between lifestyle and air pollution exposure: Potential for confounding in large administrative data cohorts. Environ Res 156:364–373. https://doi.org/10.1016/j.envres.2017.03.050
Sui D, Guo M, Ma X et al (2021) Image bio-markers and gene expression data correlation framework for lung cancer radio-genomics analysis based on deep learning. Res Square (Res Square). https://doi.org/10.21203/rs.3.rs-144196/v1
Tan J, Huo Y, Liang Z, Li L (2019) Expert knowledge-infused deep learning for automatic lung nodule detection. J Xray Sci Technol 27:17–35. https://doi.org/10.3233/xst-180426
Tariq Hussain S (n.d.) The journey: from X-rays to PET-MRI. Indian J Nucl Med
Tian Q, Wu Y, Ren X, Razmjooy N (2021) A New optimized sequential method for lung tumor diagnosis based on deep learning and converged search and rescue algorithm. Biomed Signal Process Control 68:102761. https://doi.org/10.1016/j.bspc.2021.102761
Tran GS, Nghiem TP, Nguyen VT et al (2019) Improving Accuracy of Lung Nodule Classification Using Deep Learning with Focal Loss. J Healthc Eng 2019:1–9. https://doi.org/10.1155/2019/5156416
Traverso A, Lopez Torres E, Fantacci ME, Cerello P (2017) Computer-aided detection systems to improve lung cancer early diagnosis: state-of-the-art and challenges. J Phys: Conf Ser 841:012013. https://doi.org/10.1088/1742-6596/841/1/012013
Wang X, Chen H, Gan C et al (2020) Weakly Supervised Deep Learning for Whole Slide Lung Cancer Image Analysis. IEEE Trans Cybern 50:3950–3962. https://doi.org/10.1109/tcyb.2019.2935141
Wang Y-W, Chen C-J, Huang H-C et al (2021) Dual energy CT image prediction on primary tumor of lung cancer for nodal metastasis using deep learning. Comput Med Imaging Graph 91:101935–101935. https://doi.org/10.1016/j.compmedimag.2021.101935
Wang W, Liu F, Zhi X et al (2020a) An Integrated deep learning algorithm for detecting lung nodules with low-dose CT and its application in 6G-enabled internet of medical things. IEEE Internet Things J 1–1. https://doi.org/10.1109/jiot.2020.3023436
Wani NA, Kumar R, Bedi J (2023) DeepXplainer: an interpretable deep learning based approach for lung cancer detection using explainable artificial intelligence. Comput Methods Programs Biomed 243:107879. https://doi.org/10.1016/j.cmpb.2023.107879
Wankhade S, Vigneshwari S (2023) Lung cell cancer identification mechanism using deep learning approach. Soft Computing. https://doi.org/10.1007/s00500-023-08661-4
Weng S, Xu X, Li J, Wong STC (2017) Combining deep learning and coherent anti-Stokes Raman scattering imaging for automated differential diagnosis of lung cancer. J Biomed Opt 22:1. https://doi.org/10.1117/1.jbo.22.10.106017
World Health Organization (2022) Cancer. In: World Health Organization. https://www.who.int/news-room/fact-sheets/detail/cancer. Accessed 22 Jan 2024
Wu P, Sun X, Zhao Z et al (2020) Classification of lung nodules based on deep residual networks and migration learning. 2020:1–10. https://doi.org/10.1155/2020/8975078
Wu J, Qian T (2019) A survey of pulmonary nodule detection, segmentation and classification in computed tomography with deep learning techniques. J Med Artif Intell 2:8–8. https://doi.org/10.21037/jmai.2019.04.01
Xie Y, Xia Y, Zhang J et al (2019) Knowledge-based Collaborative Deep Learning for Benign-Malignant Lung Nodule Classification on Chest CT. IEEE Trans Med Imaging 38:991–1004. https://doi.org/10.1109/tmi.2018.2876510
Xu Y, Hosny A, Zeleznik R et al (2019) Deep Learning Predicts Lung Cancer Treatment Response from Serial Medical Imaging. Clin Cancer Res 25:3266–3275. https://doi.org/10.1158/1078-0432.ccr-18-2495
Yu H, Zhou Z, Wang Q (2020) Deep Learning Assisted Predict of Lung Cancer on Computed Tomography Images Using the Adaptive Hierarchical Heuristic Mathematical Model. IEEE Access 8:86400–86410. https://doi.org/10.1109/access.2020.2992645
Yuan R, Vos PM, Cooperberg PL (2006) Computer-Aided Detection in Screening CT for Pulmonary Nodules. Am J Roentgenol 186:1280–1287. https://doi.org/10.2214/ajr.04.1969
Yu-Jen Chen Y-J, Hua K-L, Hsu C-H et al (2015) Computer-aided classification of lung nodules on computed tomography images via deep learning technique. OncoTargets Ther 2015. https://doi.org/10.2147/ott.s80733
Zhang Q, Kong X (2020) Design of Automatic Lung Nodule Detection System Based on Multi-Scene Deep Learning Framework. IEEE Access 8:90380–90389. https://doi.org/10.1109/access.2020.2993872
Zhang G, Yang Z, Gong L et al (2019) Classification of benign and malignant lung nodules from CT images based on hybrid features. Phys Med Biol 64:125011. https://doi.org/10.1088/1361-6560/ab2544
Zhao X, Wang X, Xia W et al (2020) A cross-modal 3D deep learning for accurate lymph node metastasis prediction in clinical stage T1 lung adenocarcinoma. Lung Cancer (amsterdam, Netherlands) 145:10–17. https://doi.org/10.1016/j.lungcan.2020.04.014
Author information
Authors and Affiliations
Contributions
Rabia and Ali Haider have written a major part of the paper under the supervision of Tahir and Ali Daud. Tahir and Ali Daud have helped design and improve the methodology and wrote the paper initial draft with Rabia and Ali Haider. Riad and Amal have helped in improving the paper sections, such as, review methodology, datasets, performance evaluation and challenges and future directions. Ali, Amal and Riad have improved the technical writing of paper. All authors are involved in revising the manuscript critically and have approved the final version of the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Javed, R., Abbas, T., Khan, A.H. et al. Deep learning for lungs cancer detection: a review. Artif Intell Rev 57, 197 (2024). https://doi.org/10.1007/s10462-024-10807-1
Accepted:
Published:
DOI: https://doi.org/10.1007/s10462-024-10807-1