{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,27]],"date-time":"2026-08-27T04:20:21Z","timestamp":1787804421388,"version":"build-2784847793"},"reference-count":41,"publisher":"MDPI AG","issue":"9","license":[{"start":{"date-parts":[[2026,8,25]],"date-time":"2026-08-25T00:00:00Z","timestamp":1787616000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100000780","name":"European Union","doi-asserted-by":"crossref","id":[{"id":"10.13039\/501100000780","id-type":"DOI","asserted-by":"crossref"}]},{"id":[{"id":"https:\/\/ror.org\/019w4f821","id-type":"ROR","asserted-by":"publisher"}]},{"name":"FINSERVICE S.P.A."}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Bioengineering"],"abstract":"<jats:p>Convolutional neural networks (CNNs) are widely used for biomedical image classification, yet it remains unclear under which conditions training on reduced subsets of available data can provide reliable guidance during model development, how much training data is required to achieve stable and comparable performance across CNN architectures, and whether increasing the training set size always leads to improved generalization or can sometimes result in degraded performance. We study this question across four biomedical datasets (breast mammography, pediatric chest X-ray, brain tumor MRI, and skin lesion dermoscopy) using the full grid of 39 CNN architectures (1\u20133 convolutional layers, 16\/32\/64 filters) from our companion architectural study, training each configuration from scratch on seven proportions of the training data (5%, 10%, 20%, 40%, 60%, 80%, and 100%) over 5 independent runs per configuration, with the test set held at a fixed size across all sample-size conditions to ensure a like-for-like comparison of generalization performance. The analysis investigates three complementary aspects of sample-size sensitivity: the stabilization of the training\u2013test generalization gap as training-set size increases, the reliability of architecture rankings obtained from reduced training fractions as a proxy for the full-dataset ranking, and the monotonicity of test performance with respect to training-set size. Taken together, the results point to a rough four-band pattern of reliability across the sampled fractions\u2014unstable below 20% of the training set, of uncertain overfitting status between 20% and 60%, comparatively stable between 60% and 80%, and potentially counterproductive beyond 80%\u2014while showing that this pattern is itself dataset-dependent and offers no guarantee on architecture ranking, arguing against reduced-fraction screening as a reliable shortcut for CNN architecture selection in biomedical imaging. All code and datasets are publicly released for reproducibility.<\/jats:p>","DOI":"10.3390\/bioengineering13090971","type":"journal-article","created":{"date-parts":[[2026,8,25]],"date-time":"2026-08-25T09:04:30Z","timestamp":1787648670000},"page":"971","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["CNN Sample-Size Effects Across Biomedical Datasets: A Reliability Pattern in Overfitting, Ranking, and Monotonicity"],"prefix":"10.3390","volume":"13","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-4189-0674","authenticated-orcid":false,"given":"Giacinto Angelo","family":"Sgarro","sequence":"first","affiliation":[{"name":"Department of Economics, University of Foggia, 71121 Foggia, Italy"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Melle","family":"Mendikowski","sequence":"additional","affiliation":[{"name":"German Research Center for Artificial Intelligence (DFKI), 23562 L\u00fcbeck, Germany"},{"name":"Computer Science Department, University of Hamburg, 22527 Hamburg, Germany"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9505-0673","authenticated-orcid":false,"given":"Domenico","family":"Santoro","sequence":"additional","affiliation":[{"name":"Department of Economics, Statistics and Business, Faculty of Technological and Innovation Sciences, Universitas Mercatorum, 00186 Rome, Italy"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0931-2054","authenticated-orcid":false,"given":"Luca","family":"Grilli","sequence":"additional","affiliation":[{"name":"Department of Economics, University of Foggia, 71121 Foggia, Italy"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,8,25]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","unstructured":"Thakur, G.K., Thakur, A., Kulkarni, S., Khan, N., and Khan, S. (2024). Deep learning approaches for medical image analysis and diagnosis. Cureus, 16.","DOI":"10.7759\/cureus.59507"},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"5455","DOI":"10.1007\/s10462-020-09825-6","article-title":"A survey of the recent architectures of deep convolutional neural networks","volume":"53","author":"Khan","year":"2020","journal-title":"Artif. Intell. Rev."},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Razzak, M.I., Naz, S., and Zaib, A. (2017). Deep learning for medical image processing: Overview, challenges and the future. Classification in BioApps: Automation of Decision Making, Springer.","DOI":"10.1007\/978-3-319-65981-7_12"},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"312","DOI":"10.1016\/j.job.2022.03.003","article-title":"Deep learning models in medical image analysis","volume":"64","author":"Tsuneki","year":"2022","journal-title":"J. Oral Biosci."},{"key":"ref_5","doi-asserted-by":"crossref","unstructured":"Reza, M.H., Sibly, M.N.B., Rabbani, S.G., Sadi, S.H., Ahamed, M.F., Shafi, F.B., Sarmun, R., and Chowdhury, M.E. (2026). A comprehensive review of convolutional neural networks: Foundations, enhancements and applications. Neural Comput. Appl., 38.","DOI":"10.1007\/s00521-025-11827-w"},{"key":"ref_6","unstructured":"Akhand, A.S. (2026). A Comparative Study of Custom CNNs, Pre-trained Models, and Transfer Learning Across Multiple Visual Datasets. arXiv."},{"key":"ref_7","doi-asserted-by":"crossref","unstructured":"Alzubaidi, L., Zhang, J., Humaidi, A.J., Al-Dujaili, A., Duan, Y., Al-Shamma, O., Santamar\u00eda, J., Fadhel, M.A., Al-Amidie, M., and Farhan, L. (2021). Review of deep learning: Concepts, CNN architectures, challenges, applications, future directions. J. Big Data, 8.","DOI":"10.1186\/s40537-021-00444-8"},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"43035","DOI":"10.1007\/s11042-023-17326-1","article-title":"Recent trend in medical imaging modalities and their applications in disease diagnosis: A review","volume":"83","author":"Abhisheka","year":"2024","journal-title":"Multimed. Tools Appl."},{"key":"ref_9","doi-asserted-by":"crossref","unstructured":"Iqbal, S., Qureshi, A.N., Ullah, A., Li, J., and Mahmood, T. (2022). Improving the robustness and quality of biomedical cnn models through adaptive hyperparameter tuning. Appl. Sci., 12.","DOI":"10.3390\/app122211870"},{"key":"ref_10","unstructured":"Bertrand, H. (2019). Hyper-Parameter Optimization in Deep Learning and Transfer Learning: Applications to Medical Imaging. [Ph.D. Thesis, Universit\u00e9 Paris Saclay (COmUE)]."},{"key":"ref_11","unstructured":"Li, Y., Gu, S., Gool, L.V., and Timofte, R. (November, January 27). Learning filter basis for convolutional neural network compression. Proceedings of the IEEE\/CVF International Conference on Computer Vision, Seoul, Republic of Korea."},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Ben Braiek, H., and Khomh, F. (2023). Testing feedforward neural networks training programs. ACM Trans. Softw. Eng. Methodol., 32.","DOI":"10.1145\/3529318"},{"key":"ref_13","unstructured":"Ding, Y., Huang, Z., Shou, X., Guo, Y., Sun, Y., and Gao, J. (March, January 25). Architecture-aware learning curve extrapolation via graph ordinary differential equation. Proceedings of the AAAI Conference on Artificial Intelligence, Philadelphia, PA, USA."},{"key":"ref_14","unstructured":"Le, M., Nguyen, N., and Luong, N.H. (2023). Efficacy of Neural Prediction-Based Zero-Shot NAS. arXiv."},{"key":"ref_15","unstructured":"Park, M. (2019). Data proxy generation for fast and efficient neural architecture search. arXiv."},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Nakkiran, P., Kaplun, G., Bansal, Y., Yang, T., Barak, B., and Sutskever, I. (2021). Deep double descent: Where bigger models and more data hurt. J. Stat. Mech. Theory Exp., 2021.","DOI":"10.1088\/1742-5468\/ac3a74"},{"key":"ref_17","unstructured":"Nakkiran, P. (2019). More data can hurt for linear regression: Sample-wise double descent. arXiv."},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Stengel-Eskin, E., Platanios, E.A., Pauls, A., Thomson, S., Fang, H., Van Durme, B., Eisner, J., and Su, Y. (2022, January 7\u201311). When more data hurts: A troubling quirk in developing broad-coverage natural language understanding systems. Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, Abu Dhabi, United Arab Emirates.","DOI":"10.18653\/v1\/2022.emnlp-main.789"},{"key":"ref_19","unstructured":"Min, Y., Chen, L., and Karbasi, A. (2021). The curious case of adversarially robust models: More data can help, double descend, or hurt generalization. Proceedings of the Uncertainty in Artificial Intelligence, PMLR."},{"key":"ref_20","unstructured":"Cho, J., Lee, K., Shin, E., Choy, G., and Do, S. (2015). How much data is needed to train a medical image deep learning system to achieve necessary high accuracy?. arXiv."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Rokem, A., Wu, Y., and Lee, A. (2017). Assessment of the need for separate test set and number of medical images necessary for deep learning: A sub-sampling study. BioRxiv.","DOI":"10.1101\/196659"},{"key":"ref_22","doi-asserted-by":"crossref","unstructured":"Liu, S., Zhang, H., and Jin, Y. (2022). A survey on computationally efficient neural architecture search. J. Autom. Intell., 1.","DOI":"10.1016\/j.jai.2022.100002"},{"key":"ref_23","doi-asserted-by":"crossref","unstructured":"Kot, A. (2026). Minimum Fidelity for Reliable Architecture Ranking in Bayesian Nas for Object Detection. Electron. Control Syst., 88.","DOI":"10.18372\/1990-5548.88.20966"},{"key":"ref_24","unstructured":"Yu, Q. (2025). Deep Learning-based Multi-class Classification of Breast Cancer Pathology: A Comparative Study on Full-field Digital Mammograms Using the CBIS-DDSM Dataset. [Bachelor\u2019s Thesis, Satakunta University of Applied Sciences (SAMK)]."},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"G\u00f3mez-Guzm\u00e1n, M.A., Jim\u00e9nez-Berista\u00edn, L., Garc\u00eda-Guerrero, E.E., L\u00f3pez-Bonilla, O.R., Tamayo-Perez, U.J., Esqueda-Elizondo, J.J., Palomino-Vizcaino, K., and Inzunza-Gonz\u00e1lez, E. (2023). Classifying brain tumors on magnetic resonance imaging by using convolutional neural networks. Electronics, 12.","DOI":"10.3390\/electronics12040955"},{"key":"ref_26","doi-asserted-by":"crossref","unstructured":"Das, H.S., Das, A., Neog, A., Mallik, S., Bora, K., and Zhao, Z. (2023). Breast cancer detection: Shallow convolutional neural network against deep convolutional neural networks based approach. Front. Genet., 13.","DOI":"10.3389\/fgene.2022.1097207"},{"key":"ref_27","doi-asserted-by":"crossref","unstructured":"Zahoor, S., Shoaib, U., and Lali, I.U. (2022). Breast cancer mammograms classification using deep neural network and entropy-controlled whale optimization algorithm. Diagnostics, 12.","DOI":"10.3390\/diagnostics12020557"},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"114822","DOI":"10.1109\/ACCESS.2020.3003890","article-title":"Skin lesions classification into eight classes for ISIC 2019 using deep convolutional neural network and transfer learning","volume":"8","author":"Kassem","year":"2020","journal-title":"IEEE Access"},{"key":"ref_29","unstructured":"DeVries, T., and Ramachandram, D. (2017). Skin lesion classification using deep multi-scale convolutional neural networks. arXiv."},{"key":"ref_30","doi-asserted-by":"crossref","unstructured":"Katsamenis, I., Protopapadakis, E., Voulodimos, A., Doulamis, A., and Doulamis, N. (2020, January 20\u201322). Transfer learning for COVID-19 pneumonia detection and classification in chest X-ray images. Proceedings of the 24th Pan-Hellenic Conference on Informatics, Athens, Greece.","DOI":"10.1145\/3437120.3437300"},{"key":"ref_31","doi-asserted-by":"crossref","unstructured":"Ahmed, W.S., and Karim, A.A.A. (2020). The impact of filter size and number of filters on classification accuracy in CNN. Proceedings of the 2020 International Conference on Computer Science and Software Engineering (CSASE), IEEE.","DOI":"10.1109\/CSASE48920.2020.9142089"},{"key":"ref_32","first-page":"1","article-title":"Curated breast imaging subset of digital database for screening mammography (CBIS-DDSM)","volume":"4","author":"Gimenez","year":"2017","journal-title":"Sci. Data"},{"key":"ref_33","unstructured":"Cheng, J. (Brain Tumor Dataset, 2017). Brain Tumor Dataset, Version v8."},{"key":"ref_34","doi-asserted-by":"crossref","first-page":"1122","DOI":"10.1016\/j.cell.2018.02.010","article-title":"Identifying medical diagnoses and treatable diseases by image-based deep learning","volume":"172","author":"Kermany","year":"2018","journal-title":"Cell"},{"key":"ref_35","doi-asserted-by":"crossref","unstructured":"Codella, N.C.F., Gutman, D., Celebi, M.E., Helba, B., Marchetti, M.A., Dusza, S., Kalloo, A., Liopyris, K., Mishra, N., and Kittler, H. (2017). Skin Lesion Analysis Toward Melanoma Detection: A Challenge at the 2017 International Symposium on Biomedical Imaging (ISBI), Hosted by the International Skin Imaging Collaboration (ISIC). arXiv.","DOI":"10.1109\/ISBI.2018.8363547"},{"key":"ref_36","doi-asserted-by":"crossref","unstructured":"Serena Low, W.C., Chuah, J.H., Tee, C.A.T., Anis, S., Shoaib, M.A., Faisal, A., Khalil, A., and Lai, K.W. (2021). An Overview of Deep Learning Techniques on Chest X-Ray and CT Scan Identification of COVID-19. Comput. Math. Methods Med., 2021.","DOI":"10.1155\/2021\/5528144"},{"key":"ref_37","doi-asserted-by":"crossref","unstructured":"Vineth Ligi, S., Kundu, S.S., Kumar, R., Narayanamoorthi, R., Lai, K.W., and Dhanalakshmi, S. (2022). Radiological Analysis of COVID-19 Using Computational Intelligence: A Broad Gauge Study. J. Healthc. Eng., 2022.","DOI":"10.1155\/2022\/5998042"},{"key":"ref_38","doi-asserted-by":"crossref","first-page":"4733","DOI":"10.1016\/j.csbj.2022.08.039","article-title":"Accurate brain tumor detection using deep convolutional neural network","volume":"20","author":"Khan","year":"2022","journal-title":"Comput. Struct. Biotechnol. J."},{"key":"ref_39","doi-asserted-by":"crossref","unstructured":"Kaur, R., GholamHosseini, H., Sinha, R., and Lind\u00e9n, M. (2022). Automatic lesion segmentation using atrous convolutional deep neural networks in dermoscopic skin cancer images. BMC Med. Imaging, 22.","DOI":"10.1186\/s12880-022-00829-y"},{"key":"ref_40","first-page":"183","article-title":"Brain tumor classification using convolutional neural network","volume":"Volume 1","author":"Abiwinanda","year":"2018","journal-title":"Proceedings of the World Congress on Medical Physics and Biomedical Engineering, Prague, Czech Republic, 3\u20138 June 2018"},{"key":"ref_41","unstructured":"Kermany, D., Zhang, K., and Goldbaum, M. (2018). Labeled Optical Coherence Tomography (OCT) and Chest X-Ray Images for Classification, Mendeley Data. Version V2."}],"container-title":["Bioengineering"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2306-5354\/13\/9\/971\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,27]],"date-time":"2026-08-27T04:15:48Z","timestamp":1787804148000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2306-5354\/13\/9\/971"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,25]]},"references-count":41,"journal-issue":{"issue":"9","published-online":{"date-parts":[[2026,9]]}},"alternative-id":["bioengineering13090971"],"URL":"https:\/\/doi.org\/10.3390\/bioengineering13090971","relation":{},"ISSN":["2306-5354"],"issn-type":[{"value":"2306-5354","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,25]]}}}