Automated Subregional Hippocampus Segmentation Using 3D CNNs: A Computational Framework for Brain Aging Biomarker Analysis
Abstract
1. Introduction
2. Literature Review
2.1. Hippocampal Atrophy in Stress, Depression, and Aging
2.2. Hippocampal Volume as a Neurodegenerative Biomarker
2.3. Traditional Approaches to Hippocampal Segmentation
2.4. Emergence of Deep Learning in Neuroanatomical Segmentation
2.5. Toward Brain-Age Modeling via Hippocampal Volume
3. Dataset and Preprocessing
3.1. Dataset Description
3.2. Motivation and Rationale for MSD Conversion
- Standardization: Harmonizes directory structure and naming conventions across all subject volumes and masks.
- Pipeline Automation: Enables nnU-Net to auto-detect modalities, voxel spacing, and channel configurations.
- Reproducibility: Aligns with best practices in open medical image science by adopting an interoperable, benchmarked dataset format.
3.3. Preprocessing Workflow
3.3.1. Image Normalization and Resampling
3.3.2. Spatial Cropping and ROI Localization
3.3.3. Denoising and Artifact Reduction
4. Materials and Methods
4.1. Setting up the Internal Pipeline
- Spatial dimensions of each image volume, voxel spacing across all three anatomical axes, and the presence or absence of anisotropy;
- The ratio of foreground to background voxels, assessing class distribution balance across the segmentation labels;
- Statistical properties of intensity values—mean, variance, and intensity range—to inform the choice of normalization and augmentation strategies.
4.2. Mask Overlay and Region Localization
- Validation of Ground Truth: Ensures alignment between anatomical landmarks and annotated hippocampal structures.
- Quality Assurance: Detects misregistrations, label inconsistencies, and outliers before model training.
4.3. NnU-Net Model
4.4. Customization of the Training Pipeline
- The training duration was set to 100 epochs based on preliminary experiments and convergence patterns observed during model development.
- Empirical monitoring of the validation loss and Dice coefficient across training iterations revealed performance stabilization before the 100-epoch mark.
- We activated 5-fold cross-validation, ensuring robustness across variable anatomical morphologies.
4.5. Prediction Phase and Postprocessing
4.6. Evaluation Measurement
- It balances false positives and false negatives.
- It emphasizes spatial overlap between predicted and ground truth masks.
- It is more stable than accuracy in class-imbalanced tasks.
5. Results and Discussion
5.1. Quantitative Evaluation Metrics
5.2. Visual Assessment and Mask Overlay Analysis
5.3. Volumetric Analysis
5.4. Analysis of Brain-Age Discrepancy Through Hippocampal Biomarkers
5.5. Comparison with Other State-of-the-Art Segmentation Methods
5.5.1. Atlas-Based Segmentation
5.5.2. 2D and 3D U-Net
5.5.3. UGCapsNet
5.5.4. nnU-Net
6. Limitations
7. Conclusions
Author Contributions
Funding
Conflicts of Interest
References
- McEwen, B.S. Possible mechanisms for atrophy of the human hippocampus. Mol. Psychiatry 1997, 2, 255–262. [Google Scholar] [CrossRef] [PubMed]
- Sapolsky, R.M. Why stress is bad for your brain. Science 1996, 273, 749–750. [Google Scholar] [CrossRef] [PubMed]
- Carmichael, O.T.; Aizenstein, H.A.; Davis, S.W.; Becker, J.T.; Thompson, P.M.; Meltzer, C.C.; Liu, Y. Atlas-based hippocampus segmentation in Alzheimer’s disease and mild cognitive impairment. NeuroImage 2005, 27, 979–990. [Google Scholar] [CrossRef]
- Vieira, S.; Pinaya, W.H.L.; Mechelli, A. Using deep learning to investigate the neuroimaging correlates of psychiatric and neurological disorders: Methods and applications. Neurosci. Biobehav. Rev. 2017, 74, 58–75. [Google Scholar] [CrossRef] [PubMed]
- McConnell, N.; Ndipenoch, N.; Cao, Y.; Miron, A.; Li, Y. Exploring advanced architectural variations of nnUNet. Neurocomputing 2023, 560, 126837. [Google Scholar] [CrossRef]
- Mattson, M.P.; Rydel, R.E. Development and selective neurodegeneration in cell cultures from different brain regions. Brain Res. 1989, 490, 110–125. [Google Scholar] [CrossRef]
- McEwen, B.S. Plasticity of the hippocampus: Adaptation to chronic stress and allostatic load. Ann. N. Y. Acad. Sci. 2001, 933, 265–277. [Google Scholar] [CrossRef]
- Frodl, T.; Meisenzahl, E.M.; Zetzsche, T.; Born, C.; Groll, C.; Jäger, M.; Leinsinger, G.; Bottlender, R.; Hahn, K.; Möller, H.-J. Hippocampal changes in patients with a first episode of major depression. Am. J. Psychiatry 2002, 159, 1112–1118. [Google Scholar] [CrossRef]
- Salunkhe, S.; Bachute, M.; Gite, S.; Vyas, N.; Khanna, S.; Modi, K.; Katpatal, C.; Kotecha, K. Classification of Alzheimer’s Disease Patients Using Texture Analysis and Machine Learning. Appl. Syst. Innov. 2021, 4, 49. [Google Scholar] [CrossRef]
- Raz, N.; Lindenberger, U.; Rodrigue, K.M.; Kennedy, K.M.; Head, D.; Williamson, A.; Dahle, C.; Gerstorf, D.; Acker, J.D. Regional brain changes in aging healthy adults: General trends, individual differences and modifiers. Cereb. Cortex 2005, 15, 1676–1689. [Google Scholar] [CrossRef]
- Czéh, B.; Lucassen, P.J. What causes the hippocampal volume decrease in depression? Are neurogenesis, glial changes and apoptosis implicated? Eur. Arch. Psychiatry Clin. Neurosci. 2007, 257, 250–260. [Google Scholar] [CrossRef] [PubMed]
- Hassouneh, A. Multimodal Feature Fusion and AI-Driven Framework for Early Detection of Alzheimer’s Disease. Ph.D. Thesis, Western Michigan University, Kalamazoo, MI, USA, 2025. [Google Scholar]
- Moreno-Jiménez, E.P.; Flor-García, M.; Terreros-Roncal, J.; Rábano, A.; Cafini, F.; Pallas-Bazarra, N.; Ávila, J.; Llorens-Martín, M. Adult hippocampal neurogenesis is abundant in neurologically healthy subjects and drops sharply in patients with Alzheimer’s disease. Nat. Med. 2019, 25, 554–560. [Google Scholar] [CrossRef] [PubMed]
- Hammers, A.; Heckemann, R.A.; Koepp, M.J.; Duncan, J.S.; Hajnal, J.V.; Rueckert, D.; Aljabar, P. Automatic detection and quantification of hippocampal atrophy on MRI in temporal lobe epilepsy: A proof-of-principle study. NeuroImage 2007, 36, 38–47. [Google Scholar] [CrossRef] [PubMed]
- Sämann, P.G.; Iglesias, J.E.; Gutman, B.; Grotegerd, D.; Leenings, R.; Flint, C.; Dannlowski, U.; Clarke-Rubright, E.K.; Morey, R.A.; van Erp, T.G.M.; et al. FreeSurfer-based segmentation of hippocampal subfields: A review of methods and applications, with a novel quality control procedure for ENIGMA studies and other collaborative efforts. Hum. Brain Mapp. 2020, 42, 1027–1043. [Google Scholar] [CrossRef]
- Du, G.; Cao, X.; Liang, J.; Chen, X.; Zhan, Y. Medical image segmentation based on U-Net: A review. J. Imaging Sci. Technol. 2020, 64, 1–12. [Google Scholar] [CrossRef]
- Zhang, T.R.; Larosa, A.; Di Raddo, M.-E.; Wong, V.; Wong, A.S.; Wong, T.P. Negative memory engrams in the hippocampus enhance the susceptibility to chronic social defeat stress. J. Neurosci. 2019, 39, 7576–7590. [Google Scholar] [CrossRef]
- Aghili, M.; Shojae, M.; Cabrerizo, M.; Rishe, N.; Wang, Y. Automated hippocampus segmentation and volume estimation using a transformer-based deep learning architecture. Res. Sq. 2024. [Google Scholar] [CrossRef]
- Raz, N.; Rodrigue, K.M.; Head, D.; Kennedy, K.M.; Acker, J.D. Differential aging of the medial temporal lobe: A study of a five-year change. Neurology 2004, 62, 433–438. [Google Scholar] [CrossRef]
- Huang, L.; Miron, A.; Hone, K.; Li, Y. Segmenting Medical Images: From UNet to Res-UNet and nnUNet. In Proceedings of the 2024 IEEE 37th International Symposium on Computer-Based Medical Systems (CBMS), Guadalajara, Mexico, 26–28 June 2024; pp. 483–489. [Google Scholar]
- Hippocampus Segmentation Dataset. Neuroimaging Informatics Tools and Resources Clearinghouse (NITRC). Available online: https://www.nitrc.org/frs/?group_id=377 (accessed on 13 March 2025).
- Kalaiselvi, T.; Anitha, T.; Sriramakrishnan, P. Data preprocessing techniques for MRI brain scans using deep learning models. In Brain Tumor MRI Image Segmentation Using Deep Learning Techniques; Syed, I.S., Khan, M.A., Eds.; Academic Press: Cambridge, MA, USA, 2022; pp. 13–25. [Google Scholar]
- Dolz, J.; Desrosiers, C.; Ben Ayed, I. 3D fully convolutional networks for subcortical segmentation in MRI: A large-scale study. arXiv 2017, arXiv:1612.03925. [Google Scholar] [CrossRef]
- Li, Y.; Ammari, S.; Balleyguier, C.; Lassau, N.; Chouzenoux, E. Impact of preprocessing and harmonization methods on the removal of scanner effects in brain MRI radiomic features. Cancers 2021, 13, 3000. [Google Scholar] [CrossRef]
- De Raad, K.B.; Van Garderen, K.A.; Smits, M.; Van der Voort, S.R.; Incekara, F.; Oei, E.H.G.; Hirvasniemi, J.; Klein, S.; Starmans, M.P.A. The effect of preprocessing on convolutional neural networks for medical image segmentation. In Proceedings of the 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI), Nice, France, 13–16 April 2021; pp. 655–658. [Google Scholar]
- Viswan, V.; Shaffi, N.; Subramanian, K.; Hajamohideen, F. Optimizing Medical Imaging Quality: An In-Depth Examination of Preprocessing Methods for Brain MRIs. In Applied Intelligence and Informatics; Mahmud, M., Ben-Abdallah, H., Kaiser, M.S., Ahmed, M.R., Zhong, N., Eds.; Communications in Computer and Information Science; AII 2023; Springer: Cham, Switzerland, 2024; Volume 2065. [Google Scholar]
- SSuhas, C.R. Venugopal, MRI image preprocessing and noise removal technique using linear and nonlinear filters. In Proceedings of the 2017 International Conference on Electrical, Electronics, Communication, Computer, and Optimization Techniques (ICEECCOT), Mysuru, India, 15–16 December 2017; pp. 1–4. [Google Scholar]
- Hazarika, R.A.; Maji, A.K.; Syiem, R.; Samarendra, N.S.; Debdatta, K. Hippocampus Segmentation Using U-Net Convolutional Network from Brain Magnetic Resonance Imaging (MRI). J. Digit. Imaging 2022, 35, 893–909. [Google Scholar] [PubMed]
- Isensee, F.; Wald, T.; Ulrich, C.; Baumgartner, M.; Roy, S.; Maier-Hein, K.H.; Jäger, P.F. nnU-Net revisited: A call for rigorous validation in 3D medical image segmentation. In Medical Image Computing and Computer Assisted Intervention—MICCAI 2024; Springer: Berlin/Heidelberg, Germany, 2024; pp. 488–498. [Google Scholar]
- Isensee, F.; Petersen, J.; Klein, A.; Zimmerer, D.; Jaeger, P.F.; Kohl, S.; Maier-Hein, K.H. nnU-Net: Self-adapting framework for U-Net-based medical image segmentation. arXiv 2018, arXiv:1809.10486. [Google Scholar]
- Choi, Y.; Bang, J.; Kim, S.-Y.; Seo, M.; Jang, J. Deep learning–based multimodal segmentation of oropharyngeal squamous cell carcinoma on CT and MRI using self-configuring nnU-Net. Eur. Radiol. 2024, 34, 5389–5400. [Google Scholar] [CrossRef] [PubMed]
- Krause, F.; Heindl, D.; Jebril, H.; Karner, M.; Unterdechler, M. nnU-Net Pre- and Postprocessing Strategies for UW-OCTA Segmentation Tasks in Diabetic Retinopathy Analysis. In Mitosis Domain Generalization and Diabetic Retinopathy Analysis; Sheng, B., Aubreville, M., Eds.; MIDOG DRAC 2022 2022; Lecture Notes in Computer Science; Springer: Cham, Switzerland, 2023; Volume 13597. [Google Scholar]
- Eelbode, T.; Jeroen, B.; Maxim, B.; Dirk, V.; Frederik, M.; Raf, B.; Matthew, B.B. Optimization for Medical Image Segmentation: Theory and Practice When Evaluating with Dice Score or Jaccard Index. In IEEE Transactions on Medical Imaging; IEEE: Troy, NY, USA, 2020; Volume 39, pp. 3679–3690. [Google Scholar] [CrossRef]
- Franke, K.; Gaser, C. Ten Years of BrainAGE as a Neuroimaging Biomarker of Brain Aging: What Insights Have We Gained? Front. Neurol. 2019, 10, 789. [Google Scholar] [CrossRef]
- Hassouneh, A.; Bazuin, B.; Kaku, H.; Abdel-Qader, I. A data fusion framework for mild cognitive impairment classification: Hippocampal volume and GLCM features using machine learning. In Proceedings of the International Conference on Information Technology and Applications, Sydney, Australia, 17–19 October 2024; pp. 435–445. [Google Scholar]
- Ahmad, A.L.; Sanchez-Bornot, J.; Sotero, R.C.; Coyle, D.; Idris, Z.; Faye, I. A machine learning approach for identifying anatomical biomarkers of early mild cognitive impairment. arXiv 2024, arXiv:2407.00040. [Google Scholar] [CrossRef]
- Feng, Q.; Wang, L.; Tang, X.; Ge, X.; Hu, H.; Liao, Z.; Ding, Z. Machine learning classifiers and associations of cognitive performance with hippocampal subfields in amnestic mild cognitive impairment. Front. Aging Neurosci. 2023, 15, 1273658. [Google Scholar] [CrossRef]
- Erickson, K.I.; Prakash, R.S.; Voss, M.W.; Chaddock, L.; Heo, S.; McLaren, M.; Martin, S.A.; Vieira, V.J.; Woods, J.A.; McAuley, E.; et al. Brain-derived neurotrophic factor is associated with age-related decline in hippocampal volume. J. Neurosci. 2010, 30, 5368–5375. [Google Scholar] [CrossRef]
- Lener, M.S.; Iosifescu, D.V. In pursuit of neuroimaging biomarkers to guide treatment selection in major depressive disorder: A review of recent literature. Ann. N. Y. Acad. Sci. 2015, 1344, 50–65. [Google Scholar] [CrossRef]
- Cabezas, M.; Oliver, A.; Lladó, X.; Freixenet, J.; Bach Cuadra, M. A review of atlas-based segmentation for magnetic resonance brain images. Comput. Methods Programs Biomed. 2011, 104, e158–e177. [Google Scholar] [CrossRef]
- Huang, C.; Han, H.; Yao, Q.; Zhu, S.; Zhou, S.K. 3D U2-Net: A 3D Universal U-Net for Multi-domain Medical Image Segmentation. In Medical Image Computing and Computer Assisted Intervention—MICCAI 2019; Shen, D., Liu, T., Peters, T.M., Staib, L.H., Essert, C., Zhou, S., Yap, P.-T., Khan, A., Eds.; Lecture Notes in Computer Science; Springer: Cham, Switzerland, 2019; Volume 11765. [Google Scholar]
- Su, R.; Zhang, D.; Liu, J.; Cheng, C. MSU-Net: Multi-Scale U-Net for 2D Medical Image Segmentation. Front. Genet. 2021, 12, 639930. [Google Scholar] [CrossRef]
- Tran, M.; Vo-Ho, V.-K.; Quinn, K.; Nguyen, H.; Luu, K.; Le, N. Chapter 3—CapsNet for medical image segmentation. In Artificial Intelligence in Medicine; Suzuki, A., Ed.; Academic Press: Cambridge, MA, USA, 2024; pp. 45–66. [Google Scholar]
- Pham, W.; Jarema, A.; Rim, D.; Chen, Z.; Khlif, M.S.H.; Macefield, V.G.; Henderson, L.A.; Brodtmann, A. A Comprehensive Framework for Automated Segmentation of Perivascular Spaces in Brain MRI with the nnU-Net. arXiv 2024, arXiv:2411.19564. [Google Scholar] [CrossRef]
- Vossough, A.; Khalili, N.; Familiar, A.M.; Gandhi, D.; Viswanathan, K.; Tu, W.; Haldar, D.; Bagheri, S.; Anderson, H.; Haldar, S.; et al. Automatic segmentation of pediatric brain tumors in MRI with nnU-Net ensemble and post-processing techniques. arXiv 2024, arXiv:2401.08404. [Google Scholar]
Model/Method | Preprocessing | Challenges | Performance |
---|---|---|---|
Atlas-Based Segmentation (Harvard-Oxford) | MNI registration, affine alignment | Dimensional mismatch; image–mask misalignment | Produced only background masks; not suitable for patient-specific precision |
2D/3D U-Net (Manual) | Normalization, resizing, mask standardization | Poor convergence; sensitive to class imbalance and intensity variation | Revealed noisy [0–2] outputs, still anatomically inaccurate |
UGCapsNet | Fully standardized inputs, label encoding | Memory intensive; slow convergence | Detected anterior/posterior; moderate Dice proxy (~0.03); viable but suboptimal |
nnU-Net (AutoML) | Raw MSD-compliant input | Automatically resolves all preprocessing and architectural tuning | High Dice (~0.772 anterior, ~0.711 posterior); robust, reproducible, ready for clinical use |
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. |
© 2025 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
Gogia, E.; Dehzangi, A.; Dehzangi, I. Automated Subregional Hippocampus Segmentation Using 3D CNNs: A Computational Framework for Brain Aging Biomarker Analysis. Algorithms 2025, 18, 509. https://doi.org/10.3390/a18080509
Gogia E, Dehzangi A, Dehzangi I. Automated Subregional Hippocampus Segmentation Using 3D CNNs: A Computational Framework for Brain Aging Biomarker Analysis. Algorithms. 2025; 18(8):509. https://doi.org/10.3390/a18080509
Chicago/Turabian StyleGogia, Eshaa, Arash Dehzangi, and Iman Dehzangi. 2025. "Automated Subregional Hippocampus Segmentation Using 3D CNNs: A Computational Framework for Brain Aging Biomarker Analysis" Algorithms 18, no. 8: 509. https://doi.org/10.3390/a18080509
APA StyleGogia, E., Dehzangi, A., & Dehzangi, I. (2025). Automated Subregional Hippocampus Segmentation Using 3D CNNs: A Computational Framework for Brain Aging Biomarker Analysis. Algorithms, 18(8), 509. https://doi.org/10.3390/a18080509