
==== Front
Health Sci Rep
Health Sci Rep
10.1002/(ISSN)2398-8835
HSR2
Health Science Reports
2398-8835
John Wiley and Sons Inc. Hoboken

10.1002/hsr2.70025
HSR270025
Systematic Review
Systematic Review
Deep learning techniques for Alzheimer's disease detection in 3D imaging: A systematic review
ZIA‐UR‐REHMAN et al.
Zia‐ur‐Rehman http://orcid.org/0009-0004-7218-757X
1
Awang Mohd Khalid 1
Ali Ghulam 2
Faheem Muhammad 3 Muhammad.faheem@uwasa.fi

1 Faculty of Informatics and Computing Universiti Sultan Zainal Abidin (UniSZA) Terengganu Malaysia
2 Department of Computer Science University of Okara Okara Pakistan
3 School of Technology and Innovations University of Vaasa Vaasa Finland
* Correspondence Muhammad Faheem, School of Technology and Innovations, University of Vaasa, Vaasa, Finland.
Email: Muhammad.faheem@uwasa.fi

18 9 2024
9 2024
7 9 10.1002/hsr2.v7.9 e7002508 7 2024
04 5 2024
13 8 2024
© 2024 The Author(s). Health Science Reports published by Wiley Periodicals LLC.
https://creativecommons.org/licenses/by/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by/4.0/ License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Abstract

Background and Aims

Alzheimer's disease (AD) is a degenerative neurological condition that worsens over time and leads to deterioration in cognitive abilities, reduced memory, and, eventually, a decrease in overall functioning. Timely and correct identification of Alzheimer's is essential for effective treatment. The systematic study specifically examines the application of deep learning (DL) algorithms in identifying AD using three‐dimensional (3D) imaging methods. The main goal is to evaluate these methods' current state, efficiency, and potential enhancements, offering valuable insights into how DL could improve AD's rapid and precise diagnosis.

Methods

We searched different online repositories, such as IEEE Xplore, Elsevier, MDPI, PubMed Central, Science Direct, ACM, Springer, and others, to thoroughly summarize current research on DL methods to diagnose AD by analyzing 3D imaging data published between 2020 and 2024. We use PRISMA (Preferred Reporting Items for Systematic Reviews and Meta‐Analyses) guidelines to ensure the organization and understandability of the information collection process. We thoroughly analyzed the literature to determine the primary techniques used in these investigations and their findings.

Results and Conclusion

The ability of convolutional neural networks (CNNs) and their variations, including 3D CNNs and recurrent neural networks, to detect both temporal and spatial characteristics in volumetric data has led to their widespread use. Methods such as transfer learning, combining multimodal data, and using attention procedures have improved models' precision and reliability. We selected 87 articles for evaluation. Out of these, 31 papers included various concepts, explanations, and elucidations of models and theories, while the other 56 papers primarily concentrated on issues related to practical implementation. This article introduces popular imaging types, 3D imaging for Alzheimer's detection, discusses the benefits and restrictions of the DL‐based approach to AD assessment, and gives a view toward future developments resulting from critical evaluation.

3D imaging
Alzheimer's disease
deep learning
supervised learning
machine learning
source-schema-version-number2.0
cover-dateSeptember 2024
details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.4.8 mode:remove_FC converted:18.09.2024
Zia‐ur‐Rehman , Awang MK , Ali G , Faheem M . Deep learning techniques for Alzheimer's disease detection in 3D imaging: a systematic review. Health Sci Rep. 2024;7 :e70025. 10.1002/hsr2.70025
==== Body
pmc1 INTRODUCTION

Dementia is a collection of diseases with increasing signs and symptoms, with Alzheimer's disease (AD) being the most common. AD is a medical condition defined by a gradual decline in thinking and memory functions. It is a prevalent condition amongst older individuals, constituting around 60%–80% of dementia cases. The recurrence rate of AD is substantial; however, no remedy is available. Over five million individuals globally have dementia, with 70% being AD people. The global frequency of AD was 30 million in 2006 but is expected to triple by 2050. 1 The intensity of AD could lead to the death of the person. During the early phase, memory impairments are relatively minor. 1 Figure 1 shows the different signs/indications of AD. However, as AD progresses, victims gradually lose their capacity to communicate and interact with their environment. In the later phases of the disease, this can ultimately result in a complete loss of senses for a long time.

Figure 1 Different sings of Alzheimer's.

Initial detection of the disease facilitates appropriate patient treatment, therefore reducing the detrimental impact of AD and slowing the progression towards dementia. 2 Figure 2 shows the different stages of AD.

Figure 2 Alzheimer's stages from left to right are normal, mild, and severe. 3

Magnetic resonance imaging (MRI) can differentiate between gray matter (GM) and white matter (WM). The hippocampus is a vital brain area for memory and learning. 4 The hippocampus shrinks dramatically during mild cognitive impairment (MCI) to AD. 5 MRI is the most commonly utilized method for neuroimaging due to its lack of radioactive tracer substances or hazardous gamma rays, distinguishing it from other forms of imaging. 6 Treatment could stop the development of disease symptoms. Therefore, researchers in this field apply deep learning (DL) approaches. 7 , 8 DL is data‐hungry; as more data is added, DL algorithms become more effective and surpass conventional approaches like the human brain. 9 , 10 Earlier studies have shown a leading use of two‐dimensional (2D) images with DL techniques. Due to spatial information loss in the brain's cube, 2D MRI scan slicing has drawbacks. 11 Previously, using 2D data, which converts three‐dimensional (3D) MRI images into 2D slices and uses them for feature extraction, was typical. Converting 3D to 2D loses feature detail. Straight 3D feature extraction may reveal information about the features because 3D input improves system efficiency. 12 The three planes of 3D neuroimaging MRI are shown in Figure 3.

Figure 3 A three‐dimensional brain magnetic resonance imaging shows three planes (sagittal, coronal, and axial) from left to right. 13

The majority of previous research on AD detection used 2D images, with very little research using 3D images. 14 The latest developments in neural network topologies, data enhancement methods, and powerful GPUs allow 3D deep learning to interpret voxel clinical data. Therefore, throughout the past decade, there has been a significant increase in the use of 3D deep learning in various medical imaging techniques. In this study, we provide a comprehensive analysis of the uses of DL approaches with 3D imaging for diagnosing AD and potential areas for further research. As far as we know, research articles are available online. However, this is the first review study explicitly focusing on DL techniques used with 3D imaging for AD detection.

1.1 Search strategy

The research examines the acknowledgment of efforts to detect AD in several scientific databases, including IEEE Xplore, Elsevier, MDPI, PubMed Central, Science Direct, ACM, Springer, and other electronic library databases. The analysis focuses on papers that include keywords such as “deep learning in Alzheimer's,” “Alzheimer's diagnosis using deep learning,” “Alzheimer's detection using deep learning with 3D imaging,” and “Alzheimer's diagnosis and prediction using supervised deep learning.” We searched for research articles regarding AD detection using DL, which can be up to 25 pages in Google Scholar. Using the keywords mentioned above and reputable journals, we queried the Google Scholar database to include all publications, including those not found in the specified repositories. In this review article, we examine articles from 2020 to 2024. This research method follows the rules in Figure 4 of the PRISMA guidelines, ensuring the information collection process is organized and understandable. Moreover, Table 1 provides comprehensive information about the criteria for including and excluding data in PRISMA. This table thoroughly summarizes the criteria for deciding whether to review or exclude an item from the examination. At the onset of the study, we gathered papers from online database searches and chose 413 articles for the review.

Figure 4 PRISMA flow diagram of the article selection procedure.

Table 1 The table describes the inclusion and exclusion criteria for selected research papers.

	Inclusion criteria	Exclusion criteria	
Research nature	Original research articles	Thesis, communication letters, white papers, editorials, and reports	
Language applied	Research papers are presented in English.	Duplicate and non‐English research articles	
Publication years	Articles published from 2020 to 2024 (for applications and results analysis part)	Not related to the subject of the review	
Intervention	DL and 3D images	Traditional and arithmetical methods	
Source of articles	Articles published in academic journals and conferences.	Research papers that insufficiency of information.	
Settings	DL in medical imaging.	Not in medical imaging settings.	
Region	Not restricted to an exact region.	NA	
Abbreviations: 3D, three‐dimensional; DL, deep learning; NA, not available.

John Wiley & Sons, Ltd.

Following a thorough analysis, we selected 87 publications for review. Among these, 31 articles presented different ideas, descriptions, and explanations of theories and models, while the remaining 56 articles mainly focused on application problems and current breakthroughs. Our dedication to delivering a sophisticated and innovative assessment has prompted us to focus on works published in recent years intentionally.

1.2 Previous research gaps

Previous research on the application of DL to find AD in 3D images mainly focused on using CNN models and other DL models to get features and do classification tasks. However, further research is necessary to explore the capabilities of more recent structures, primarily designed to handle 3D MRI images. It is also essential to do more research using heterogeneous data fusion approaches. These involve combining data from different imaging types, like structural magnetic resonance imaging (sMRI), functional MRI (fMRI), and positron emission tomography (PET) imaging. This integration aims to improve the reliability and precision of the algorithms used for detecting AD. Enhanced GPUs, suitable system hardware, and adequately designed algorithms can increase implementation reliability. The comprehensibility of DL techniques in this field requires further research, limiting their use in healthcare. Addressing these inconsistencies can enhance the efficiency of detection and treatment methods for AD.

1.3 Contributions

The review article has the following contributions: ▪ Implementation of DL methods with 3D imaging to detect AD from 2020 to 2024.

▪ 3D images help identify and characterize brain deterioration by providing a full view of the brain area.

▪ We used 3D imaging to improve spatial resolution and understand complicated structures from many perspectives, which aids in accurately detecting AD.

▪ We explored 3D imaging with DL techniques to identify AD and its stages, using different datasets, imaging modalities, and core findings.

This review further discusses popular neuroimaging modalities, 3D imaging with DL approaches for Alzheimer's detection, discussion, challenges in AD, conclusion, and future extensions in this area.

2 BRAIN IMAGING MODALITIES FOR ALZHEIMER'S

According to previous research, numerous imaging procedures exist to help identify AD. However, the most prevalent and well‐acknowledged methods among experts are PET and MRI. The text concisely explains how MRI works, including its many patterns and PET imaging. It also discusses how both techniques are used in AD classification.

2.1 MRI imaging

MRI is a significant advancement in clinical imaging. MRI provides excellent spatial detail and superficial tissue resolution, allowing for a 3D tomographic view and the ability to show how the body changes constantly. 6 Figure 5 presents a variety of MRI series images.

Figure 5 Different magnetic resonance imaging (MRI) sequence images were left to right: first (A) functional MRI (fMRI), (B) diffusion tensor imaging (DTI), (C) diffusion‐weighted imaging (DWI), and the last (D) T1‐W MRI. 15

The application of sMRI in the detection of AD is essential. This noninvasive imaging technique provides accurate information about the actual structure and condition of the brain as a whole. 16 T1‐weighted MRI is utilized for structural evaluation and T2 for diagnostic investigation. Circulation variations in fMRI measure activity in the brain. Scientists often use fMRI to study neurological diseases such as bipolar disorder, and they are adopting it for further disease diagnosis. 17 Diffusion tensor imaging (DTI) is an anisotropic diffusion technique employed in MRI to evaluate the organization of the brain's axonal tissue network. The last figure displays the maximum diffusivity of white‐matter fibers in this plane. 18

2.2 PET imaging

Researchers often utilize 18‐fluoro‐deoxyglucose (FDG) PET in addition to MRI to identify AD due to its capacity to improve training methods. Currently, PET imaging with molecules is available for use in healthcare. PET is commonly used for cell characterization, classification, therapeutic management, and cerebral glucose metabolism detection. PET frequently demonstrates the development of AD, including the phases of MCI. 19 Figure 6 Alzheimer's stages on PET images.

Figure 6 A positron emission tomography (PET) image of Alzheimer's disease (AD), cognitively normal (CN), and mild cognitive impairment (MCI) patients from left to right.

Regular PET investigations are not feasible due to the radioactive elements employed in scanning. FDGPET is the most commonly used PET for detecting AD. The PET method requires immediate imaging after injecting the radioactive tracer substances, making it time‐sensitive. The device captures photons, creating 3D images of the subject. 20 Electroencephalograms, MRIs, and PETs are standard neuroimaging techniques used to study brain function and make diagnoses. Since MRI and PET became unaffordable, EEG became the preferred diagnostic method. 21

3 ALZHEIMER'S DIAGNOSIS WITH DL APPROACHES

DL's untapped potential is gaining significant interest in clinical studies. DL enables algorithms using computation and multi‐processing levels to acquire a collection of attributes from source data and classify the result based on the learned data. 22 DL is suggested for initial AD detection due to its superior image accuracy for classification compared to the classic machine learning approach. 23 , 24 Unsupervised learning uses training data without labels. The following are the most commonly used unsupervised learning models: ▪ Autoencoder (AE)

▪ Deep Belief Network (DBN)

▪ Generative adversarial networks (GANs)

▪ Restricted Boltzmann Machines (RBMs)

▪ Variational Autoencoders (VAEs)

In supervised DL, methods are learned to map input information into output labels using pairs of inputs and outputs from the training process. The core models and their variants are primarily used as; ▪ Convolutional Neural Network (CNN)

▪ Recurrent Neural Network (RNN)

▪ Long short‐term memory (LSTM)

▪ Gated Recurrent Unit (GRU)

▪ Capsule Networks

▪ Transformers

Here, we aim to define transformers as models that can be classified into supervised or unsupervised deep learning, depending on the specific application. Figure 7 shows the different steps of DL methods researchers used for their work.

Figure 7 Different steps of deep learning algorithms. 25

The following DL with 3D image data for AD detection is described in detail.

3.1 3D imaging for AD detection

DL techniques use 3D images to evaluate and retrieve data for AD diagnosis. We present the details of various models used with 3D imaging in diverse domains. Table 2 presents the supervised DL with 3D imaging, which has the data set name, imaging modality, model, and critical findings.

Table 2 AD diagnosis research using supervised DL with 3D imaging.

Reference	Year	Data set	Image type	Method	Key findings	
Begum and Selvaraj 26	2023	ADNI	fMRI, MRI, DTI, and PET	DCNN	Feature investigation and classification	
Parmar et al. 27	2020	ADNI	fMRI	3D‐CNN	Extracted spatiotemporal features	
Tuan et al. 28	2022	ADNI and OASIS	MRI	CNN, XGBoost, SVM	Categorize AD based on the segmented tissues	
Tufail et al. 29	2022	ADNI	MRI, PET	3D‐CNN	Model engineering has a reduced effect on classification	
Tomassini et al. 30	2021	OASIS and ADNI	sMRI	Convolutional Long Short‐Term Memory	it can be used for other mental illnesses	
Ayyar et al. 31	2021	ADNI	sMRI	3D‐CNN	Hippocampal ROIs of the brain	
Pardakhti and Sajedi 32	2020	ADNI	MRI	3D‐CNN	Estimation of brain age	
Chakraborty et al. 33	2021	ADNI	MRI	3DCNN	Good performance measures	
Folego et al. 34	2020	ADNI	MRI	3D‐CNN	Automatic and relatively fast	
Chen et al. 35	2022	ADNI‐1 and ADNI‐2	MRI	Multiview‐slice Attention and 3D‐CNN	Slice‐level attention	
Balboni et al. 36	2022	Three hospital datasets	MRI	Transfer learning	Enhanced hippocampal segmentation feature	
Rogeau et al. 37	2024	ADNI, FTLDNI	18F‐FDG PET	3D VGG16	Excellent accuracy	
İsmail and Dalveren 38	2024	ADNI	fMRI	3D‐CAPSNET and RNN	Spatiotemporal information for AD	
Lin et al. 39	2024	ADNI	MRI	Pre‐trained ResNet	Enhances the method's training cost	
Yang et al. 40	2024	ADNI2	PET, MRI	Contrastive Masked Vim Autoencoder	The model improves by 2.7% AUC performance	
Alp et al. 41	2024	ADNI	MRI	Vision Transformer	Excellent classification accuracy	
Zhang et al. 42	2021	ADNI	MRI	Densely connected + attention mechanism	Excellent classification accuracy	
Nasina and Reddy 43	2020	ADNI	MRI	3D‐CNN	Good performance	
Zhang et al. 44	2021	ADNI	sMRI	3D Residual Self‐Attention DNN	Better Accuracy and generalizability	
Dhinagar et al. 45	2021	ADNI, OASIS	MRI	3D‐CNN	Generalized superior on unseen MRI	
Nawaz et al. 46	2020	ADNI	MRI	3D‐CNN	Good for imbalance data	
Gao et al. 47	2022	ADNI	sMRI	3D‐MgNe	Reduce the number of parameters	
Narazani et al. 48	2022	ADNI	PET, MRI	3D‐CNN	PET captures

AD pathologies better

	
Zuo et al. 49	2024	Huanhu Hospital, China	Eye‐tracking	Multilayered comparison CNN	Consistent validity in classifying AD patients and NC	
Zhang et al. 50	2023	ADNI	MRI, PET	Attention‐based 3D‐CNN	Emphasis on main brain regions linked to the disease.	
Turrisi et al. 51	2023	ADNI	MRI	3D‐CNN	Data augmentation approach plays an important part in performance	
Ullanat et al. 52	2021	ADNI	MRI	3D ResNet‐18	Effectiveness of models in identifying AD‐related brain areas.	
Zheng et al. 53	2023	ADNI	MRI	3D EfficientNet	High classification accuracy in identifying AD and detecting MCI progression.	
Kong et al. 54	2022	ADNI	PET, MRI	3D‐CNN	Get better multi‐modal feature	
Agarwal et al. 55	2022	ADNI	MRI	End‐to‐end‐learning deep CNN	Complicated MR image features that might enable early identification	
Tufail et al. 29	2022	ADNI	MRI, PET	3D‐CNN	Integrating augmentation approaches can decrease classification performance.	
Etminani et al. 56	2022	ADNI	PET	3D, DL	Every neurological disease needs a posterior cingulate cortex	
Biswas and Gin 57	2023	OASIS and ADNI	MRI	Random forest, KNN, Gradient boost, and decision tree.	Highest accuracy for WM volume using the OASIS data set	
Wu et al. 58	2022	OASIS	MRI	3D Transfer learning	The classification time is abridged.	
Dyrba et al. 59	2021	ADNI	MRI	3D‐CNN	Hippocampus shrinkage was the best AD indicator.	
Kang et al. 60	2023	ADNI	MRI	3D Deep Convolutional GANs	The model prevents overfitting due to insufficient sMRI data.	
Zhang et al. 61	2021	ADNI	MRI	3D‐CNN	Increases training speed and reduces boundary noise	
Tufail et al. 62	2022	ADNI	PET	CNN	Superior performance on the multiclass classification	
Soliman et al. 63	2020	ADNI	MRI	3D‐CNN	Develop general features for detecting AD biomarkers.	
Kumari et al. 64	2022	ADNI	MRI	3D‐CNN and modified owl search	Improving the weights of the suggested model	
Nithya et al. 65	2023	ADNI	MRI	3D‐CNN	Good prediction accuracy compared to other models	
Akan et al. 66	2024	ADNI	MRI	Vision Transformers and Bi‐LSTM	Good classification accuracy.	
Tomassiini et al. 30	2021	ADNI	sMRI	ConvLSTM	Offer a less data set‐specific method.	
Chen et al. 67	2024	ADNI, OASIS, and AIBL,	MRI	3D‐CNN, Longitudinal Transformer	Efficiently merge spatiotemporal features for classification.	
Khan et al. 68	2024	ADNI	MRI and PET	Dual‐3DM3‐AD method	Segmentation, efficiently decreasing complexity	
Ningsen et al. 69	2024	Kaggle	MRI	Multi‐Order 3D U‐NET, CNN	Considerably enhanced the segmentation accuracy.	
Kim et al. 70	2024	ADNI	MRI and PET	Multimodal 3D DL	Discovering the importance of ROI‐based techniques in refining effectiveness	
Erdemir 71	2024	ADNI	MRI	3D LSTM method	Little sensitivity and F1 score in detecting the shift from mild to AD	
Uyguroğlu et al. 72	2024	ADNI and AIBL	MRI	CNN and 3D angular orientations	Innovative usage of distinctive angles of the full data set by new perspectives	
Ning et al. 73	2024	ADNI	MRI and PET	3D‐CNN	Variations in cortical areas can detect	
Castellano et al. 74	2024	ADNI	MRI and PET	Multi‐modal and Uni‐modal methods	Concentrations on important Alzheimer's associated regions for its detection	
Rahman et al. 75	2024	OASIS‐3	CT	Transfer learning method	Extremely correct in identifying pelvic cracks	
Dharwada et al. 76	2024	ADNI	sMRI	3D Transfer learning	Weights in every ensemble method are tuned for enhanced accuracy	
Li et al. 77	2024	ADNI	MRI	New Med‐3D Transfer technique	Gives a dynamic platform for encountering real‐time problem‐solving	
Saad and Hasan 78	2024	ADNI	MRI	3D Vision Transformers	Understand the brain parts that greatly influence solving classification issues	
Hu et al. 79	2024	ADNI	MRI	3D‐SEConvNeXt model	Attaining excellent results in early AD diagnosis tasks	
Abbreviations: CT, computed tomography; DTI, diffusion tensor imaging; fMRI, functional magnetic resonance imaging; MRI, magnetic resonance imaging; sMRI, structural magnetic resonance imaging; PET, positron emission tomography.

John Wiley & Sons, Ltd.

Different methods have been applied in the article, and each has its pros and cons regarding its structural nature. Table 3 shows the algorithms used in this article and their advantages and limitations.

Table 3 DL methods and their advantages and disadvantages.

Methods	Advantages	Disadvantages	
3D‐CNN	Possibly will get 3D data from brain scans.	The training process is expensive in terms of different resources.	
RNN	Suitable for continuous 2D photographs	The gradient vanishes and then strengthens rapidly.	
AE	Capable communication abilities. It helps the decrease of dimensionality and is guileless to use.	Restrictions in adaptableness.	
DBN	Deliver better‐ordered feature procurement.	Require significant computing resources for training procedures.	
GANs	Produce correct distributions of data.	Tough to train and susceptible to method disintegration, which restricts generator efficiency.	
DNN	Can control several datasets. Minimal feature engineering limits	The timing of training procedures is problematic, and weights are hard to comprehend.	
Transformers	Competently control sequential data.	Challenging and expensive in training.	
Caps Net	Improve the ordered learning of representations.	It is more computationally costly than CNNs.	
RBM	Extremely expressive and favorable to rational reasoning.	The training process is resource‐intensive.	
Abbreviations: AE, Autoencoder; 3D CNN, three‐dimensional convolutional neural network; DBN, Deep Belief Network; DL, deep learning; GANs, generative adversarial networks; RBM, Restricted Boltzmann Machines; RNN, recurrent neural networks.

John Wiley & Sons, Ltd.

3.2 Performance comparison of DL algorithms for Alzheimer's detection

When diagnosing Alzheimer's using 3D scans, different DL methods demonstrate a range of performance and advantages. CNNs are highly effective because they can identify spatial patterns in image data, making them widely preferred for clinical imaging tasks. Due to their specific structure for sequential data, we do not commonly employ LSTMs, RNNs, and GRUs for this objective. However, we can integrate them with CNN models to boost their effectiveness. GANs and VAEs are valuable methods for producing artificial 3D scans, improving data set training processes, and strengthening models' adaptability. Current research uses fewer RBMs and DBNs as more sophisticated designs emerge. Capsule networks, known for their ability to maintain spatial correlations, have potential but are currently in the early stages of research. Researchers are now investigating the possibility of transformers in 3D healthcare imaging, renowned for their ability to handle sequences and attention‐intensive tasks. This is mainly due to their impressive feature retrieval capability. CNNs, enhanced by GANs and VAEs, currently provide the highest level of performance in the AD field utilizing 3D imaging.

4 DISCUSSION

Current studies 15 , 17 show that DL approaches are becoming more helpful in identifying AD. These studies highlight the potential of DL algorithms, especially CNNs, in interpreting medical imaging data like MRI, PET, and CT images. Such algorithms can accurately distinguish AD from normal controls and the early stages of AD by automatically extracting and learning features from massive datasets. Researchers also combine images and clinical and genetic data to improve detection effectiveness. However, big, well‐annotated datasets, model interpretation, and comprehensive validation among varied populations are challenges. Despite these challenges, DL‐based approaches are becoming more popular for timely and precise AD diagnosis, ensuring enhanced patient outcomes with earlier management. This research highlights significant advancements in DL techniques to identify AD on time using 3D imaging modalities. DL algorithms have recently seen a substantial increase in AD diagnosis and prediction, demonstrating promising outcomes in many areas. 3D imaging improves visualization and perception of depth, facilitating healthcare diagnosis, architectural design, and simulations. Its comprehensive spatial data allows for better assessments and evaluations than 2D imaging. In future studies, better research, standardized imaging techniques, and model improvements' interpretability can help address these shortcomings.

5 CHALLENGES

Although DL techniques have yielded positive results, identifying AD still faces several issues that require attention. The data augmentation and transfer learning models have stopped overfitting problems in the research and group of information. However, issues with generalization can still happen when there are not enough data items. 80 The data set will be expanded. However, its efficiency remains uncertain, and more studies will continue in this domain. Supervised DL has mitigated this issue and decreased the reliance on professional expertise, but additional research is necessary. Despite DL methods' impressive achievements, AD identification faces several restrictions and challenges. 81 A comprehensive understanding of the depth method, benchmarking platform, and other relevant factors is necessary to determine the ideal balance of several biomarkers. Several fusion approaches have the potential to contribute to research on AD. 82 , 83 Still, there are obstacles to overcome when using DL‐based techniques 84 , 85 , 86 , 87 , 88 , 89 , 90 , 91 to tackle AD.

6 CONCLUSION AND FUTURE DIRECTIONS

The early identification of AD remains a challenge, and computer scientists are continuously investigating potential solutions. This work presents the biomarkers associated with AD, imaging modalities, and 3D imaging applications for AD identification. DL techniques have shown success in using multiple data modalities. Regarding classification algorithms, CNN is among those most often employed and outperforms other deep models in this area. Even though most issues in AD classification remain unsolved, the problem of overfitting the data set requires solutions. Given the shortage of healthcare data, unsupervised and self‐monitoring approaches are becoming prominent areas of study in medical imaging. To advance precision treatment in AD management, robust DL algorithms that can extrapolate within diverse populations use multifaceted data integration for disease identification and easily integrate into clinical operations through transparent and comprehensible decision‐support platforms.

AUTHOR CONTRIBUTIONS

Zia‐ur‐Rehman: Writing—original draft; software; data curation; resources; conceptualization. Mohd Khalid Awang: Formal analysis; validation; investigation; supervision. Ghulam Ali: Conceptualization; writing—review and editing; methodology; project administration. Muhammad Faheem: Methodology; data curation, formal analysis; visualization, review and editing.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

ETHICS STATEMENT

We used the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guidelines for our study. This study did not include any actual human or animal subjects. We assess all studies that undergo review according to ethical principles.

TRANSPARENCY STATEMENT

The lead author Muhammad Faheem affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

ACKNOWLEDGMENTS

The authors would like to thank their affiliated universities for providing research funding to accomplish this research work.

DATA AVAILABILITY STATEMENT

Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study. Data that was analyzed is available from the corresponding author upon reasonable request.
==== Refs
REFERENCES

1 S. Gauthier PR‐N , Morais J , Webster C . World Azheimer Report 2021: Journey Through the Diagnosis of Dementia. Alzheimer's Disease International; 2021.
2 Wang X , Michaelis ML , Michaelis EK . Functional genomics of brain aging and Alzheimer's disease: focus on selective neuronal vulnerability. Curr Genomics. 2010;11 :618‐633.21629439
3 Aparna M , Rao BS . A novel automated deep learning approach for Alzheimer's disease classification. IAES Int J Artif Intell. 2023;12 :451.
4 Eberhard B , Miles W , Andersen P , et al. Local Circuits. Oxford University Press; 2006.
5 Kälin AM , Park MT , Chakravarty MM , et al. Subcortical shape changes, hippocampal atrophy and cortical thinning in future Alzheimer's disease patients. Front Aging Neurosci. 2017;9 :38.28326033
6 Khoo VS , Dearnaley DP , Finnigan DJ , Padhani A , Tanner SF , Leach MO . Magnetic resonance imaging (MRI): considerations and applications in radiotherapy treatment planning. Radiother Oncol. 1997;42 :1‐15.9132820
7 Pais M , Martinez L , Ribeiro O , et al. Early diagnosis and treatment of Alzheimer's disease: new definitions and challenges. Revi Brasil Psiquiatria. 2020;42 :431‐441.
8 Rosly R , Makhtar M , Awang MK , Hassan H , Rose A . Deep multi‐classifier learning for medical data sets. Int J Eng Trends Technol. 2020;68 :11‐15.
9 Jiang T . Deep learning application in Alzheimer disease diagnoses and prediction. 2020 4th Int Conf Artif Intell Virt Real. 2020;1 :62‐65.
10 Makhtar M , Rosly R , Awang MK , Mohamad M , Zakaria AH . A multi‐classifier method based deep learning approach for breast cancer. Int J Eng Trends Technol. 2020;1 :102‐107.
11 Feng C , Elazab A , Yang P , et al. Deep learning framework for Alzheimer's disease diagnosis via 3D‐CNN and FSBi‐LSTM. IEEE Access. 2019;7 :63605‐63618.
12 Yagis E , Citi L , Diciotti S , Marzi C , Atnafu SW , De Herrera AGS . 3D convolutional neural networks for diagnosis of Alzheimer's disease via structural mri. 2020 IEEE 33rd International Symposium on Computer‐Based Medical Systems (CBMS). IEEE; 2020:65‐70.
13 Karasawa H , Liu C‐L , Ohwada H . Deep 3D convolutional neural network architectures for Alzheimer's disease diagnosis. Intelligent Information and Database Systems: 10th Asian Conference, ACIIDS 2018, Dong Hoi City, Vietnam, March 19‐21, 2018, Proceedings, Part I 10. Springer; 2018:287‐296.
14 Krizhevsky A , Sutskever I , Hinton GE . ImageNet classification with deep convolutional neural networks. Commun ACM. 2017;60 :84‐90.
15 Sharma R , Goel T , Tanveer M , Lin C , Murugan R . Deep learning based diagnosis and prognosis of Alzheimer's disease: A comprehensive review. IEEE Trans Cogni Dev Syst. 2023;15 (3 ):1123‐1138.
16 Gomez‐Nicola D , Boche D . Post‐mortem analysis of neuroinflammatory changes in human Alzheimer's disease. Alzheimer's Res Ther. 2015;7 :42.25904988
17 Verma S , Goel T , Tanveer M , Ding W , Sharma R , Murugan R . Machine learning techniques for the Schizophrenia diagnosis: a comprehensive review and future research directions. J Amb Intell Human Comput\. 2023;14 :4795‐4807.
18 Jellison BJ , Field AS , Medow J , Lazar M , Salamat MS , Alexander AL . Diffusion tensor imaging of cerebral white matter: a pictorial review of physics, fiber tract anatomy, and tumor imaging patterns. AJNR Am J Neuroradiol. 2004;25 :356‐369.15037456
19 Reske SN , Kotzerke J . FDG‐PET for clinical use: results of the 3rd German Interdisciplinary Consensus Conference,” Onko‐PET III”, 21 July and 19 September 2000. Eur J Nucl Med. 2001;28 :1707‐1723.11702115
20 Berger A . How does it work?: Positron emission tomography. BMJ. 2003;326 :1449.12829560
21 Ieracitano C , Mammone N , Hussain A , Morabito FC . A novel multi‐modal machine learning based approach for automatic classification of EEG recordings in dementia. Neural Netwks. 2020;123 :176‐190.
22 Tanveer M , Rashid AH , Ganaie MA , Reza M , Razzak I , Hua K‐L . Classification of Alzheimer's disease using ensemble of deep neural networks trained through transfer learning. IEEE J Biomed Health Inform. 2022;26 :1453‐1463.34033550
23 Azmi MAA , Mustafa WA , Alquran H , et al. Breast cancer detection on X‐tray mammogram images. 2023 3rd International Conference on Mobile Networks and Wireless Communications (ICMNWC). IEEE; 2023:1‐5.
24 Heaton J . Ian Goodfellow, Yoshua Bengio, and Aaron Courville: Deep learning: The MIT Press, 2016, 800 pp, ISBN: 0262035618. Genet Prog Evol Mach. 2018;19 :305‐307.
25 Liguori G . What is deep learning? Simple explained. Kenovy; 2022. https://kenovy.com/what-is-deep-learning/
26 Begum AP , Selvaraj P . Alzheimer's disease classification and detection by using AD‐3D DCNN model. Bull Electric Eng Inform. 2023;12 :882‐890.
27 Parmar H , Nutter B , Long R , Antani S , Mitra S . Spatiotemporal feature extraction and classification of Alzheimer's disease using deep learning 3D‐CNN for fMRI data. J Med Imaging. 2020;7 :056001.
28 Tuan TA , Pham TB , Kim JY , Tavares J . Alzheimer's diagnosis using deep learning in segmenting and classifying 3D brain MR images. Int J Neurosci. 2022;132 :689‐698.33045895
29 Tufail AB , Ullah K , Khan RA , et al. On improved 3D‐CNN‐based binary and multiclass classification of Alzheimer's disease using neuroimaging modalities and data augmentation methods. J Healthc Eng. 2022;2022 :1302170.35186220
30 Tomassini S , Falcionelli N , Sernani P , Müller H , Dragoni AF . An end‐to‐end 3D ConvLSTM‐based framework for early diagnosis of Alzheimer's disease from full‐resolution whole‐brain sMRI scans. 2021 IEEE 34th International Symposium on Computer‐Based Medical Systems (CBMS). IEEE; 2021:74‐78.
31 Ayyar MP , Benois‐Pineau J , Zemmari A , Catheline G . Explaining 3D CNNS for Alzheimer's disease classification on sMRI images with multiple ROIs. 2021 IEEE International Conference on Image Processing (ICIP). IEEE; 2021:284‐288.
32 Pardakhti N , Sajedi H . Brain age estimation based on 3D MRI images using 3D convolutional neural network. Multimed Tools Appl. 2020;79 :25051‐25065.
33 Chakraborty S , Sain M , Park J , Aich S . Early detection of Alzheimer's disease from 1.5 T MRI scans Using 3D convolutional neural network. Proceedings of International Conference on Smart Computing and Cyber Security: Strategic Foresight, Security Challenges and Innovation (SMARTCYBER 2020). Springer; 2021:15‐28.
34 Folego G , Weiler M , Casseb RF , Pires R , Rocha A . Alzheimer's disease detection through whole‐brain 3D‐CNN MRI. Front Bioeng Biotechnol. 2020;8 :534592.33195111
35 Chen L , Qiao H , Zhu F . Alzheimer's disease diagnosis with brain structural MRI using multiview‐slice attention and 3D convolution neural network. Front Aging Neurosci. 2022;14 :871706.35557839
36 Balboni E , Nocetti L , Carbone C , et al. The impact of transfer learning on 3D deep learning convolutional neural network segmentation of the hippocampus in mild cognitive impairment and Alzheimer disease subjects. Hum Brain Mapp. 2022;43 :3427‐3438.35373881
37 Rogeau A , Hives F , Bordier C , et al. A 3D convolutional neural network to classify subjects as Alzheimer's disease, frontotemporal dementia or healthy controls using brain 18F‐FDG PET. Neuroimage. 2024;288 :120530.38311126
38 Ismail A , Dalveren GGM . Use of 3D‐CAPSNET and RNN models for 4D fMRI‐based Alzheimer's disease pre‐detection. Turk J Sci Technol. 2024;19 :223‐235.
39 Lin Y , Li X , Zhang Y , Tang J . Attention‐based efficient classification for 3D MRI image of Alzheimer's disease. Biomed Signal Process Control. 2024;82 :104571.
40 Yang G , Du K , Yang Z , Du Y , Zheng Y , Wang S . CMViM: contrastive masked vim autoencoder for 3D multi‐modal representation learning for AD classification. arXiv. Preprint March 25, 2024.
41 Alp S , Akan T , Bhuiyan MS , et al. Joint transformer architecture in brain 3D MRI classification: its application in Alzheimer's disease classification. Sci Rep. 2024;14 :8996.38637671
42 Zhang J , Zheng B , Gao A , Feng X , Liang D , Long X . A 3D densely connected convolution neural network with connection‐wise attention mechanism for Alzheimer's disease classification. Magn Reson Imaging. 2021;78 :119‐126.33588019
43 Nasina SS , Reddy ARM . An integrated deep auto‐encoder based 3D‐CNN architecture to analyse brain functional units in detecting premature stage of Alzheimer's disease. J Crit Rev. 2020;7 (15 ):5821‐5828.
44 Zhang X , Han L , Zhu W , Sun L , Zhang D . An explainable 3D residual self‐attention deep neural network for joint atrophy localization and Alzheimer's disease diagnosis using structural MRI. IEEE J Biomed Health Inform. 2021c;26 :5289‐5297.
45 Dhinagar NJ , Thomopoulos SI , Owens‐Walton C , et al. 3D convolutional neural networks for classification of Alzheimer's and Parkinson's disease with T1‐weighted brain MRI. 17th International Symposium on Medical Information Processing and Analysis. SPIE; 2021:277‐286.
46 Nawaz A , Anwar SM , Liaqat R , Iqbal J , Bagci U , Majid M . Deep convolutional neural network based classification of Alzheimer's disease using MRI data. 2020 IEEE 23rd International Multitopic Conference (INMIC). IEEE; 2020:1‐6.
47 Gao Y , Huang H , Zhang L . Predicting Alzheimer's disease using 3DMgNet. arXiv. Preprint. January 12, 2022.
48 Narazani M , Sarasua I , Pölsterl S , Lizarraga A , Yakushev I , Wachinger C . Is a PET all you need? a multi‐modal study for Alzheimer's disease using 3D CNNs. International Conference on Medical Image Computing and Computer‐Assisted Intervention. Springer; 2022:66‐76.
49 Zuo F , Jing P , Sun J , Duan J , Ji Y , Liu Y . Deep learning‐based eye‐tracking analysis for diagnosis of Alzheimer's disease using 3D comprehensive visual stimuli. IEEE J Biomed Health Inform. 2024;28 (5 ):2871‐2893.
50 Zhang Y , Teng Q , He X , et al. Attention‐based 3D CNN with multi‐layer features for Alzheimer's disease diagnosis using brain images. 2023 45th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). IEEE; 2023:1‐4.
51 Turrisi R , Verri A , Barla A . The effect of data augmentation and 3D‐CNN depth on Alzheimer's DiseaseAlzheimer's disease detection. arXiv. Preprint. Sepetember 13, 2023.
52 Ullanat V , Balamurali V , Rao A . A novel residual 3‐D convolutional network for Alzheimer's disease diagnosis based on raw MRI scans. 2020 IEEE‐EMBS Conference on Biomedical Engineering and Sciences (IECBES). IEEE; 2021:82‐87.
53 Zheng B , Gao A , Huang X , Li Y , Liang D , Long X . A modified 3D EfficientNet for the classification of Alzheimer's disease using structural magnetic resonance images. IET Image Proces. 2023;17 :77‐87.
54 Kong Z , Zhang M , Zhu W , Yi Y , Wang T , Zhang B . Multi‐modal data Alzheimer's disease detection based on 3D convolution. Biomed Signal Process Control. 2022;75 :103565.
55 Agarwal D , Berbis MA , Martín‐Noguerol T , Luna A , Garcia SCP , De La Torre‐Diez I . End‐to‐end deep learning architectures using 3D neuroimaging biomarkers for early Alzheimer's diagnosis. Mathematics. 2022;10 :2575.
56 Etminani K , Soliman A , Davidsson A , et al. A 3D deep learning model to predict the diagnosis of dementia with Lewy bodies, Alzheimer's disease, and mild cognitive impairment using brain 18F‐FDG PET. Eur J Nucl Med Mol Imaging. 2022;49 (2 ):563‐584.34328531
57 Biswas R , Gini JR . Multi‐class classification of Alzheimer's disease detection from 3D MRI image using ML techniques and its performance analysis. Multimed Tools Appl. 2023;83 :33527‐33554.
58 Wu H , Luo J , Lu X , Zeng Y . 3D transfer learning network for classification of Alzheimer's disease with MRI. Int J Mach Learn Cybern. 2022;13 :1997‐2011.
59 Dyrba M , Hanzig M , Altenstein S , et al. Improving 3D convolutional neural network comprehensibility via interactive visualization of relevance maps: evaluation in Alzheimer's disease. Alzheimer's Res Ther. 2021;13 :191.34814936
60 Kang W , Lin L , Sun S , Wu S . Three‐round learning strategy based on 3D deep convolutional GANs for Alzheimer's disease staging. Sci Rep. 2023;13 :5750.37029214
61 Zhang P , Lin S , Qiao J , Tu Y . Diagnosis of Alzheimer's disease with ensemble learning classifier and 3D convolutional neural network. Sensors. 2021b;21 :7634.34833710
62 Tufail AB , Anwar N , Othman MTB , et al. Early‐stage Alzheimer's disease categorization using PET neuroimaging modality and convolutional neural networks in the 2D and 3D domains. Sensors. 2022a;22 :4609.35746389
63 Soliman SA , El‐Sayed A , Salem A‐BM . Predicting Alzheimer's disease with 3D convolutional neural networks. Int J Appl Fuzzy Sets Artif Intell. 2020;10 :125‐146.
64 Kumari R , Goel S , Das S . A 3D convolutional neural network approach for diagnosing Alzheimer's disease using modified Owl search optimization technique. TENCON 2022‐2022 IEEE Region 10 Conference (TENCON). IEEE; 2022:1‐7.
65 Nithya V , Mohanasundaram N , Santhosh R . Earlier detection of Alzheimer's disease using 3D‐convolutional neural networks. Comput Syst Sci Eng. 2023;46 (2 ):2601‐2618.
66 Akan T , Alp S , Bhuiyanb MA . Vision Transformerstransformers and Bi‐LSTM for Alzheimer's disease diagnosis from 3D MRI. arXiv. Preprint. January 6, 2024.
67 Chen Q , Fu Q , Bai H , Hong Y . Longformer: longitudinal transformer for Azheimer's disease classification with structural MRIs. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. IEEE; 2024:3575‐3584.
68 Khan AA , Mahendran RK , Perumal K , Faheem M . Dual‐3DM 3‐AD: mixed transformer based semantic segmentation and triplet pre‐processing for early multi‐class Alzheimer's diagnosis. IEEE Trans Neural Syst Rehabil Eng. 2024;32 :696‐707.38261494
69 Ningsen S , Yuan Y , Lanhua Zhang ZH . Convolutional Neural Network Image Segmentation of Alzheimer's Disease Based on Multi‐Order 3D U‐NET. IOS Press; 2024:182‐188.
70 Kim SK , Duong QA , Gahm JK . Multimodal 3D Deep Learning for Early Diagnosis of Alzheimer's Disease. IEEE Access; 2024.
71 Erdemir E . Predicting Alzheimer's Disease Stage Transformations 12 Months in Advance using 3D Convolutional LSTM based on 3D Magnetic Resonance Images. Informatics Institute; 2024.
72 Uyguroğlu F , Toygar Ö , Demirel H . CNN‐based Alzheimer's disease classification using fusion of multiple 3D angular orientations. Signal Image Video Proces. 2024;272 :1‐9.
73 Ning K , Cannon PB , Yu J , et al. 3D convolutional neural networks uncover modality‐specific brain‐imaging predictors for Alzheimer's disease sub‐scores. Brain Inform. 2024;11 :5.38310619
74 Castellano G , Esposito A , Lella E , Montanaro G , Vessio G . Automated detection of Alzheimer's disease: a multi‐modal approach with 3D MRI and amyloid PET. Sci Rep. 2024;14 :5210.38433282
75 Rahman R , Yagi N , Hayashi K , Maruo A , Muratsu H , Kobashi S . Enhancing fracture diagnosis in pelvic X‐rays by deep convolutional neural network with synthesized images from 3D‐CT. Sci Rep. 2024;14 :8004.38580737
76 Dharwada S , Tembhurne J , Diwan T . An optimal weighted ensemble of 3D CNNs for early diagnosis of Alzheimer's disease. SN Comput Sci. 2024;5 :252.
77 Li Y , Tang J , Ding W , Luo J , Ahmad N , Kumar R . An assisted diagnosis of Alzheimer's disease incorporating attention mechanisms Med‐3D transfer modeling. Comput Mater Contin. 2024;78 (1 ):713‐733.
78 Saoud LS , Almarzouqi H . Explainable early detection of Alzheimer's disease using ROIs and an ensemble of 3D vision transformers. Research Square. Preprint. February 2, 2024.
79 Hu Z , Wang Y , Xiao L . Alzheimer's Disease Diagnosis By 3D‐SEConvNeXt. Springer; 2024.
80 Arbabyazd L , Shen K , Wang Z , et al. Virtual connectomic datasets in Alzheimer's disease and aging using whole‐brain network dynamics modelling. eNeuro. 2021;8 :ENEURO.0475‐20.2021.
81 Choi B‐K , Madusanka N , Choi H‐K , et al. Convolutional neural network‐based MR image analysis for alzheimer's disease classification. Curr Med Imaging Rev. 2020;16 :27‐35.31989891
82 Wang S‐H , Zhou Q , Yang M , Zhang Y‐D . ADVIAN: Alzheimer's disease VGG‐inspired attention network based on convolutional block attention module and multiple way data augmentation. Front Aging Neurosci. 2021;13 :687456.34220487
83 Zhang Y‐D , Dong Z , Wang S‐H , et al. Advances in multimodal data fusion in neuroimaging: overview, challenges, and novel orientation. Int J Inform Fusion. 2020;64 :149‐187.
84 Raza B , Aslam A , Sher A , et al. Autonomic performance prediction framework for data warehouse queries using lazy learning approach. Appl Soft Comput. 2020;91 (6 ):106216.
85 Ashraf MW , Idrus SM , Iqbal F , et al. Disaster‐resilient optical network survivability: a comprehensive survey. Photonics. 2018;5 (4 ):1‐24.
86 Faheem M , Kuusniemi H , Eltahawy B , et al. A lightweight smart contracts framework for blockchain‐based secure communication in smart grid applications. IET Generation, Transmission & Distribution. 2024;18 (3 ):625‐638.
87 Burhan M , Alam H , Arsalan A , et al. A comprehensive survey on the cooperation of fog computing paradigm‐based IoT applications: layered architecture, real‐time security issues, and solutions. IEEE Access. 2023;11 (3 ):73303‐73329.
88 Alarood AA , Faheem M , Al‐Khasawneh MA , et al. Secure medical image transmission using deep neural network in e‐health applications. Healthc Technol Lett. 2023;10 (4 ):87‐98.37529409
89 Raza B , Kumar YJ , Malik AK , et al. Performance prediction and adaptation for database management system workload using case‐based reasoning approach. Inf Syst. 2018;76 (6 ):46‐58.
90 Faheem M , Al‐Khasawneh MA . Multilayer cyberattacks identification and classification using machine learning in internet of blockchain (IoBC)‐based energy networks. Data in Brief. 2024;54 (5 ):110461.38774244
91 Faheem M , Butt RA , Raza B , et al. A multiobjective, lion mating optimization inspired routing protocol for wireless body area sensor network based healthcare applications. Sensors. 2019;19 (23 ):5072.31757104
