Diagnosis of Alzheimer’s Disease Based on Accelerated Mirror Descent Optimization and a Three-Dimensional Aggregated Residual Network
Abstract
:1. Introduction
- •
- We propose SAMD, an unbiased subgradient accelerated mirror descent optimization algorithm, designed to expedite the training of diagnostic networks. By optimizing and improving the mirror descent algorithm [13], SAMD avoids the optimization problem falling into local optimum under the non-Euclidean distance metric. SAMD achieves accelerated convergence by introducing the step factor and the deviation correction factor to the mirror descent algorithm. Furthermore, the incorporation of a subgradient unbiased estimation mechanism effectively mitigates the issue of vanishing gradient.
- •
- We propose a 3D aggregated residual network (ARCNN) for feature extraction from craniocerebral MRI scans to diagnose AD. The ARCNN leverages our proposed aggregation residual blocks (ARBs) to capture global information within the input data, enhancing its capability to recognize complex patterns and features while improving model stability. Experimental results on the ADNI dataset highlight the superior performance of the diagnostic model trained using the SAMD algorithm in terms of accuracy and efficiency compared to other state-of-the-art methods.
- •
- Our proposed SAMD algorithm achieves a convergence rate that is a square order faster than gradient descent algorithms such as Adam and SGD. Experimental results further validate the efficiency of the SAMD algorithm, showcasing an average time-saving of approximately 19% compared to gradient descent algorithms in the training phase. The SAMD algorithm accelerates AD diagnosis model training, overcoming real-world healthcare computational constraints, ensuring local data security, enabling continuous updates, and enhancing diagnosis efficiency while complying with privacy regulations and evolving medical knowledge.
2. Related Work
3. Materials and Methods
3.1. Datasets
3.2. Proposed Method
3.2.1. Data Preprocessing
3.2.2. Three-Dimensional Aggregated Residual Diagnosis Network
3.2.3. Accelerated Mirror Descent Optimization
Preliminary
SAMD Optimization Algorithm
Algorithm 1 A stochastic unbiased subgradient accelerated mirror descent algorithm |
Input: Loss function |
Output: The optimization result |
1: Initial: , obtain by grid search |
2: while do |
3: Calculate |
4: Calculate |
5: Calculate |
6: Set by |
7: Calculate , and |
8: Set by |
9: if then |
10: Let |
11: break |
12: end if |
13: end while |
14: return . |
Convergence of SAMD
4. Experiments and Discussion
4.1. Experimental Settings
4.2. Compared with Other Optimizers
4.3. Performance of Diagnostic Network
4.4. Performance of Training Time
4.5. Comparison with Previous Work
4.6. Future Research Directions
5. Conclusions
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
References
- Holtzman, D.M.; Morris, J.C.; Goate, A.M. Alzheimer’s disease: The challenge of the second century. Sci. Transl. Med. 2011, 3, 77sr1. [Google Scholar] [CrossRef]
- Alzheimer’s Disease International. World Alzheimer Report 2022. Available online: https://www.alzint.org/resource/world-alzheimer-report-2022/ (accessed on 10 July 2023).
- Jia, J.; Wei, C.; Chen, S.; Li, F.; Tang, Y.; Qin, W.; Zhao, L.; Jin, H.; Xu, H.; Wang, F.; et al. The cost of Alzheimer’s disease in China and re-estimation of costs worldwide. Alzheimer’s Dement. 2018, 14, 483–491. [Google Scholar] [CrossRef]
- Smith, J.C.; Nielson, K.A.; Antuono, P.; Lyons, J.A.; Hanson, R.J.; Butts, A.M.; Hantke, N.C.; Verber, M.D. Semantic memory functional MRI and cognitive function after exercise intervention in mild cognitive impairment. J. Alzheimer’s Dis. 2013, 37, 197–215. [Google Scholar] [CrossRef] [PubMed]
- Altaf, F.; Islam, S.M.; Akhtar, N.; Janjua, N.K. Going deep in medical image analysis: Concepts, methods, challenges, and future directions. IEEE Access 2019, 7, 99540–99572. [Google Scholar] [CrossRef]
- Haller, S.; Zanchi, D.; Rodriguez, C.; Giannakopoulos, P. Brain structural imaging in Alzheimer’s disease. In Biomarkers for Preclinical Alzheimer’s Disease; Springer: Berlin/Heidelberg, Germany, 2018; pp. 107–117. [Google Scholar]
- Albert, M.S.; DeKosky, S.T.; Dickson, D.; Dubois, B.; Feldman, H.H.; Fox, N.C.; Gamst, A.; Holtzman, D.M.; Jagust, W.J.; Petersen, R.C.; et al. The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimer’s Dement. 2011, 7, 270–279. [Google Scholar] [CrossRef]
- Poloni, K.M.; de Oliveira, I.A.D.; Tam, R.; Ferrari, R.J. Brain MR image classification for Alzheimer’s disease diagnosis using structural hippocampal asymmetrical attributes from directional 3-D log-Gabor filter responses. Neurocomputing 2021, 419, 126–135. [Google Scholar] [CrossRef]
- Alinsaif, S.; Lang, J.; Alzheimer’s Disease Neuroimaging Initiative. 3D Shearlet-Based Descriptors Combined with Deep Features for the Classification of Alzheimer’s Disease Based on MRI Data. Comput. Biol. Med. 2021, 138, 104879. [Google Scholar] [CrossRef] [PubMed]
- Lian, C.; Liu, M.; Zhang, J.; Shen, D. Hierarchical Fully Convolutional Network for Joint Atrophy Localization and Alzheimer’s Disease Diagnosis Using Structural MRI. IEEE Trans. Pattern Anal. Mach. Intell. 2018, 42, 880–893. [Google Scholar] [CrossRef]
- Hochreiter, S.; Schmidhuber, J. Long short-term memory. Neural Comput. 1997, 9, 1735–1780. [Google Scholar] [CrossRef]
- Tieleman, T.; Hinton, G. Rmsprop: Divide the gradient by a running average of its recent magnitude. coursera: Neural networks for machine learning. COURSERA Neural Netw. Mach. Learn. 2012, 4, 26–31. [Google Scholar]
- Ben-Tal, A.; Margalit, T.; Nemirovski, A. The ordered subsets mirror descent optimization method with applications to tomography. SIAM J. Optim. 2001, 12, 79–108. [Google Scholar] [CrossRef]
- Baron, J.; Chetelat, G.; Desgranges, B.; Perchey, G.; Landeau, B.; De La Sayette, V.; Eustache, F. In vivo mapping of gray matter loss with voxel-based morphometry in mild Alzheimer’s disease. Neuroimage 2001, 14, 298–309. [Google Scholar] [CrossRef]
- Fan, Y.; Shen, D.; Gur, R.C.; Gur, R.E.; Davatzikos, C. COMPARE: Classification of morphological patterns using adaptive regional elements. IEEE Trans. Med. Imaging 2006, 26, 93–105. [Google Scholar] [CrossRef]
- Ju, R.; Hu, C.; Zhou, P.; Li, Q. Early diagnosis of Alzheimer’s disease based on resting-state brain networks and deep learning. IEEE/ACM Trans. Comput. Biol. Bioinform. 2017, 16, 244–257. [Google Scholar] [CrossRef]
- Coupé, P.; Eskildsen, S.F.; Manjón, J.V.; Fonov, V.S.; Pruessner, J.C.; Allard, M.; Collins, D.L.; Alzheimer’s Disease Neuroimaging Initiative. Scoring by nonlocal image patch estimator for early detection of Alzheimer’s disease. NeuroImage Clin. 2012, 1, 141–152. [Google Scholar] [CrossRef]
- Wang, H.; Shen, Y.; Wang, S.; Xiao, T.; Deng, L.; Wang, X.; Zhao, X. Ensemble of 3D densely connected convolutional network for diagnosis of mild cognitive impairment and Alzheimer’s disease. Neurocomputing 2019, 333, 145–156. [Google Scholar] [CrossRef]
- Cui, R.; Liu, M.; Alzheimer’s Disease Neuroimaging Initiative. RNN-based longitudinal analysis for diagnosis of Alzheimer’s disease. Comput. Med. Imaging Graph. 2019, 73, 1–10. [Google Scholar] [CrossRef] [PubMed]
- Feng, C.; Elazab, A.; Yang, P.; Wang, T.; Zhou, F.; Hu, H.; Xiao, X.; Lei, B. Deep learning framework for Alzheimer’s disease diagnosis via 3D-CNN and FSBi-LSTM. IEEE Access 2019, 7, 63605–63618. [Google Scholar] [CrossRef]
- Puente-Castro, A.; Fernandez-Blanco, E.; Pazos, A.; Munteanu, C.R. Automatic assessment of Alzheimer’s disease diagnosis based on deep learning techniques. Comput. Biol. Med. 2020, 120, 103764. [Google Scholar] [CrossRef]
- Allioui, H.; Sadgal, M.; Elfazziki, A. Utilization of a convolutional method for Alzheimer disease diagnosis. Mach. Vis. Appl. 2020, 31, 1–19. [Google Scholar] [CrossRef]
- Pan, Y.; Liu, M.; Lian, C.; Zhou, T.; Xia, Y.; Shen, D. Synthesizing missing PET from MRI with cycle-consistent generative adversarial networks for Alzheimer’s disease diagnosis. In International Conference on Medical Image Computing and Computer-Assisted Intervention, Proceedings of the 21st International Conference, Granada, Spain, 16–20 September 2018; Springer: Cham, Switzerland, 2018; pp. 455–463. [Google Scholar]
- Ji, H.; Liu, Z.; Yan, W.Q.; Klette, R. Early diagnosis of Alzheimer’s disease using deep learning. In Proceedings of the 2nd International Conference on Control and Computer Vision, Jeju Island, Republic of Korea, 15–18 June 2019; pp. 87–91. [Google Scholar]
- Zhang, Y.; Wang, S.; Xia, K.; Jiang, Y.; Qian, P.; Alzheimer’s Disease Neuroimaging Initiative. Alzheimer’s disease multiclass diagnosis via multimodal neuroimaging embedding feature selection and fusion. Inf. Fusion 2021, 66, 170–183. [Google Scholar] [CrossRef]
- Santos-Rodríguez, R.; García-García, D.; Cid-Sueiro, J. Cost-sensitive classification based on Bregman divergences for medical diagnosis. In Proceedings of the 2009 International Conference on Machine Learning and Applications, Miami, FL, USA, 13–15 December 2009; pp. 551–556. [Google Scholar]
- Angulo, H.; Giménez, J.; Moros, A.M.; Nikodem, K. On strongly h-convex functions. Ann. Funct. Anal. 2011, 2, 85–91. [Google Scholar] [CrossRef]
- Hardt, M.; Ma, T.; Recht, B. Gradient Descent Learns Linear Dynamical Systems. J. Mach. Learn. Res. 2018, 19, 1–44. [Google Scholar]
- Nedic, A.; Lee, S. On stochastic subgradient mirror-descent algorithm with weighted averaging. SIAM J. Optim. 2014, 24, 84–107. [Google Scholar] [CrossRef]
- Xu, L.; Yao, Z.; Li, J.; Lv, C.; Zhang, H.; Hu, B. Sparse feature learning with label information for Alzheimer’s disease classification based on magnetic resonance imaging. IEEE Access 2019, 7, 26157–26167. [Google Scholar] [CrossRef]
- Cui, R.; Liu, M. Hippocampus analysis by combination of 3-D DenseNet and shapes for Alzheimer’s disease diagnosis. IEEE J. Biomed. Health Inform. 2019, 23, 2099–2107. [Google Scholar] [CrossRef] [PubMed]
- Simonyan, K.; Zisserman, A. Very deep convolutional networks for large-scale image recognition. arXiv 2014, arXiv:1409.1556. [Google Scholar]
- He, K.; Zhang, X.; Ren, S.; Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 27–30 June 2016; pp. 770–778. [Google Scholar]
- Janoušová, E.; Vounou, M.; Wolz, R.; Gray, K.R.; Rueckert, D.; Montana, G. Biomarker discovery for sparse classification of brain images in Alzheimer’s disease. Ann. BMVA 2012, 2012, 1–11. [Google Scholar]
- Liu, S.; Liu, S.; Cai, W.; Che, H.; Pujol, S.; Kikinis, R.; Feng, D.; Fulham, M.J.; ADNI. Multimodal neuroimaging feature learning for multiclass diagnosis of Alzheimer’s disease. IEEE Trans. Biomed. Eng. 2015, 62, 1132–1140. [Google Scholar] [CrossRef]
- Korolev, S.; Safiullin, A.; Belyaev, M.; Dodonova, Y. Residual and plain convolutional neural networks for 3D brain MRI classification. In Proceedings of the 2017 IEEE 14th international symposium on biomedical imaging (ISBI 2017), Melbourne, VIC, Australia, 18–21 April 2017; pp. 835–838. [Google Scholar]
- Karasawa, H.; Liu, C.L.; Ohwada, H. Deep 3D convolutional neural network architectures for alzheimer’s disease diagnosis. In Asian Conference on Intelligent Information and Database Systems, Proceedings of the 10th Asian Conference, ACIIDS 2018, Dong Hoi City, Vietnam, 19–21 March 2018; Springer: Cham, Switzerland, 2018; pp. 287–296. [Google Scholar]
- Khvostikov, A.; Aderghal, K.; Benois-Pineau, J.; Krylov, A.; Catheline, G. 3D CNN-based classification using sMRI and MD-DTI images for Alzheimer disease studies. arXiv 2018, arXiv:1801.05968. [Google Scholar]
- Lin, W.; Tong, T.; Gao, Q.; Guo, D.; Du, X.; Yang, Y.; Guo, G.; Xiao, M.; Du, M.; Qu, X.; et al. Convolutional neural networks-based MRI image analysis for the Alzheimer’s disease prediction from mild cognitive impairment. Front. Neurosci. 2018, 12, 777. [Google Scholar] [CrossRef]
- Zhu, T.; Cao, C.; Wang, Z.; Xu, G.; Qiao, J. Anatomical Landmarks and DAG Network Learning for Alzheimer’s Disease Diagnosis. IEEE Access 2020, 8, 206063–206073. [Google Scholar] [CrossRef]
- Lin, W.; Lin, W.; Chen, G.; Zhang, H.; Gao, Q.; Huang, Y.; Tong, T.; Du, M.; Alzheimer’s Disease Neuroimaging Initiative. Bidirectional mapping of brain MRI and PET with 3D reversible GAN for the diagnosis of Alzheimer’s disease. Front. Neurosci. 2021, 15, 646013. [Google Scholar] [CrossRef]
- Gao, X.; Shi, F.; Shen, D.; Liu, M. Task-induced pyramid and attention GAN for multimodal brain image imputation and classification in alzheimer’s disease. IEEE J. Biomed. Health Inform. 2021, 26, 36–43. [Google Scholar] [CrossRef] [PubMed]
- Li, H.; Tan, Y.; Miao, J.; Liang, P.; Gong, J.; He, H.; Jiao, Y.; Zhang, F.; Xing, Y.; Wu, D. Attention-based and micro designed EfficientNetB2 for diagnosis of Alzheimer’s disease. Biomed. Signal Process. Control 2023, 82, 104571. [Google Scholar] [CrossRef]
Parameters | ADNI |
---|---|
Magnetic field intensity | 1.5T |
Field of view | 160 |
Acquisition matrix | |
Thick of scans | 1.2 mm |
Manufacturer | SIMENS |
TE (Echo time) | 3.5–3.7 ms |
TI (Inversion time) | 1000 ms |
TR (repetition Time) | 3000 ms |
Weight | T1 |
Layer | Stage | Input | Size | Output |
---|---|---|---|---|
1 | conv1 | 256 × 256 × 256 × 1 | 7 × 7 × 7 × 64 | 128 × 128 × 128 × 64 |
2 | max pool | 128 × 128 × 128 × 64 | — | 64 × 64 × 64 × 64 |
3 | conv2 | 64 × 64 × 64 × 64 | 64 × 64 × 64 × 256 | |
4 | conv3 | 64 × 64 × 64 × 256 | 32 × 32 × 32 × 512 | |
5 | conv4 | 32 × 32 × 32 × 512 | 16 × 16 × 16 × 1024 | |
6 | conv5 | 16 × 16 × 16 × 1024 | 8 × 8 × 8 × 2048 | |
7 | GAP | 8 × 8 × 8 × 2048 | — | 1 × 1 × 1 × 2048 |
8 | FC1 | 1 × 2048 | — | 1 × 512 |
9 | FC2 | 1 × 512 | — | 1 × 2 |
Name | Meanings |
---|---|
Bregman divergence of a and b | |
Bregman divergence generating function | |
Strong convex coefficient | |
L | Lipschitz constant |
Cost function | |
Scalar product of x, y | |
Gradient of | |
Subgradient of | |
Step factor | |
Deviation correction factor | |
Dynamically adjusted factor |
Methods | AD vs. NC | sMCI vs. pMCI | ||||||
---|---|---|---|---|---|---|---|---|
ACC | SEN | SPE | AUC | ACC | SEN | SPE | AUC | |
SGD | 0.947 | 0.972 | 0.876 | 0.969 | 0.779 | 0.733 | 0.762 | 0.787 |
ADAM | 0.951 | 0.960 | 0.929 | 0.967 | 0.788 | 0.526 | 0.854 | 0.781 |
SAMD (ours) | 0.954 | 0.901 | 0.952 | 0.970 | 0.799 | 0.737 | 0.833 | 0.789 |
Reference | Subject | AD vs. NC | sMCI vs. pMCI | ||||||
---|---|---|---|---|---|---|---|---|---|
ACC | SEN | SPE | AUC | ACC | SEN | SPE | AUC | ||
Janousova et al. [34] | 231 NC + 63 sMCI + 168 pMCI +198 AD | 0.880 | 0.850 | 0.910 | - | 0.700 | 0.640 | 0.750 | - |
Liu et al. [35] | 77 NC + 85 AD | 0.914 | 0.923 | 0.904 | - | - | - | - | - |
Korolev et al. [36] | 61 NC + 77 sMCI + 43 pMCI + 50 AD | 0.800 | - | - | 0.870 | 0.520 | - | - | 0.520 |
Karasawa et al. [37] | 574 NC + 346 AD | 0.940 | - | - | - | - | - | - | - |
Khvostikov et al. [38] | 58 NC + 48 AD | 0.854 | 0.883 | 0.900 | - | - | - | - | - |
Lin et al. [39] | 229 NC + 188 AD | 0.799 | 0.840 | 0.748 | 0.861 | - | - | - | - |
Xu et al. [30] | 165 NC + 95 sMCI +126 pMCI + 142 AD | 0.904 | 0.924 | 0.887 | 0.954 | 0.637 | 0.786 | 0.454 | 0.679 |
Cui et al. [31] | 223NC + 231 sMCI + 165 pMCI + 192 AD | 0.923 | 0.906 | 0.937 | 0.970 | 0.750 | 0.733 | 0.762 | 0.777 |
Zhu et al. [40] | 419 NC + 345 AD | 0.916 | 0.874 | 0.948 | 0.958 | - | - | - | - |
Poloni et al. [8] | 302 NC + 209 AD | 0.826 | - | - | 0.900 | - | - | - | - |
Alinsaif et al. [9] | 50 sMCI +50 pMCI | - | - | - | - | 0.700 | 0.600 | 0.800 | - |
Lin et al. [41] | 308 NC + 233 sMCI +183 pMCI + 362 AD | 0.923 | 0.904 | 0.944 | 0.928 | 0.741 | 0.750 | 0.731 | 0.766 |
Gao et al. [42] | 427 NC + 342 sMCI +234 pMCI + 352 AD | 0.920 | 0.891 | 0.940 | 0.956 | 0.753 | 0.773 | 0.741 | 0.786 |
Li et al. [43] | 153 NC + 151 AD | 0.930 | 0.924 | 0.948 | - | - | - | - | - |
proposed | 100 NC + 117 sMCI + 53 pMCI + 78 AD | 0.954 | 0.901 | 0.952 | 0.970 | 0.799 | 0.737 | 0.833 | 0.789 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Tu, Y.; Lin, S.; Qiao, J.; Zhang, P.; Hao, K. Diagnosis of Alzheimer’s Disease Based on Accelerated Mirror Descent Optimization and a Three-Dimensional Aggregated Residual Network. Sensors 2023, 23, 8708. https://doi.org/10.3390/s23218708
Tu Y, Lin S, Qiao J, Zhang P, Hao K. Diagnosis of Alzheimer’s Disease Based on Accelerated Mirror Descent Optimization and a Three-Dimensional Aggregated Residual Network. Sensors. 2023; 23(21):8708. https://doi.org/10.3390/s23218708
Chicago/Turabian StyleTu, Yue, Shukuan Lin, Jianzhong Qiao, Peng Zhang, and Kuankuan Hao. 2023. "Diagnosis of Alzheimer’s Disease Based on Accelerated Mirror Descent Optimization and a Three-Dimensional Aggregated Residual Network" Sensors 23, no. 21: 8708. https://doi.org/10.3390/s23218708
APA StyleTu, Y., Lin, S., Qiao, J., Zhang, P., & Hao, K. (2023). Diagnosis of Alzheimer’s Disease Based on Accelerated Mirror Descent Optimization and a Three-Dimensional Aggregated Residual Network. Sensors, 23(21), 8708. https://doi.org/10.3390/s23218708