{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,3]],"date-time":"2026-07-03T02:20:21Z","timestamp":1783045221544,"version":"3.54.6"},"reference-count":13,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2026,5,11]],"date-time":"2026-05-11T00:00:00Z","timestamp":1778457600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,7,3]],"date-time":"2026-07-03T00:00:00Z","timestamp":1783036800000},"content-version":"vor","delay-in-days":53,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["BMC Res Notes"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Objective<\/jats:title>\n                    <jats:p>To benchmark a very small convolutional neural network trained with and without knowledge distillation for pneumonia detection on PneumoniaMNIST, with emphasis on discrimination, calibration, efficiency, and external generalization.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Using the predefined MedMNIST v2 splits and 5 random seeds, a ResNet-18 teacher achieved an area under the receiver operating characteristic curve of 0.9735 (95% confidence interval 0.9687\u22120.9782). A 60,642-parameter TinyCNN retained about 90% of teacher discrimination (0.8818, 0.8634\u22120.9001) while reducing parameters by 184-fold, multiply-accumulate operations by 10.9-fold, and batch-1 CPU inference time by 3.7-fold. Knowledge distillation did not provide consistent discrimination gains over vanilla training on this benchmark. The teacher was less well calibrated before temperature scaling, which improved its negative log-likelihood, Brier score, and expected calibration error. External zero-shot evaluation on Kermany and a balanced RSNA subset showed domain-shift degradation for all models, with partial recovery after brief fine-tuning but persistent teacher\u2013student gaps.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1186\/s13104-026-07793-8","type":"journal-article","created":{"date-parts":[[2026,5,11]],"date-time":"2026-05-11T09:35:24Z","timestamp":1778492124000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Benchmarking knowledge distillation for lightweight pneumonia detection: a multi-seed calibration study on PneumoniaMNIST"],"prefix":"10.1186","volume":"19","author":[{"given":"Ryan Yi Sheng","family":"Neo","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,5,11]]},"reference":[{"key":"7793_CR1","unstructured":"Hinton G, Vinyals O, Dean J. Distilling the knowledge in a neural network. In NIPS Deep Learning Workshop, 2015."},{"key":"7793_CR2","unstructured":"Romero A, Ballas N, Kahou SE, Chassang A, Gatta C, Bengio Y. Fitnets: hints for thin deep nets. In International Conference on Learning Representations (ICLR), 2015."},{"key":"7793_CR3","unstructured":"Guo C, Pleiss G, Sun Y, Weinberger KQ. On calibration of modern neural networks. In International Conference on Machine Learning (ICML), 2017."},{"issue":"15","key":"7793_CR4","doi-asserted-by":"publisher","first-page":"14885","DOI":"10.1007\/s11042-024-19638-2","volume":"84","author":"MA Asham","year":"2024","unstructured":"Asham MA, Al-Shargabi AA, Al-Sabri R, Meftah I. A lightweight deep learning model with knowledge distillation for pulmonary diseases detection in chest x-rays. Multimed Tools Appl. 2024;84(15):14885\u2013913.","journal-title":"Multimed Tools Appl"},{"key":"7793_CR5","doi-asserted-by":"crossref","unstructured":"EL-Assiouti OS, Hamed G, Khattab D, Ebied HM. HDKD: Hybrid data-efficient knowledge distillation network for medical image classification. Eng Appl Artific Intell. 2024.","DOI":"10.1016\/j.engappai.2024.109430"},{"key":"7793_CR6","doi-asserted-by":"crossref","unstructured":"Termritthikun C, Kanjana R, Chinrungrueng S. Explainable knowledge distillation for on-device chest X-ray multi-label classification. IEEE\/ACM Trans Comput Biol Bioinform. 2024.","DOI":"10.1109\/TCBB.2023.3272333"},{"key":"7793_CR7","doi-asserted-by":"crossref","unstructured":"Li X, Xu Y, Liu J, et al. Knowledge distillation and teacher\u2013student learning in medical imaging: comprehensive overview, pivotal role, and future directions. Med Image Anal. 2025.","DOI":"10.1016\/j.media.2025.103819"},{"issue":"11","key":"7793_CR8","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pmed.1002683","volume":"15","author":"JR Zech","year":"2018","unstructured":"Zech JR, Badgeley MA, Liu M, Costa AB, Titano JJ, Oermann EK. Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: a cross-sectional study. PLoS Med. 2018;15(11):e1002683.","journal-title":"PLoS Med"},{"key":"7793_CR9","doi-asserted-by":"crossref","unstructured":"Yang J, Shi R, Wei D, Liu Z, Zhao L, Ke B, Pfister H, Ni B. Medmnist v2: a large-scale lightweight benchmark for 2d and 3d biomedical image classification. Scientific Data, 2023;10.","DOI":"10.1038\/s41597-022-01721-8"},{"key":"7793_CR10","doi-asserted-by":"crossref","unstructured":"Tong R, Liu J,Wang T, Hu X, Liu S, Wang L, Xu J. Does bigger mean better? Comparative analysis of CNNs and biomedical vision-language models in medical diagnosis. arXiv:2510.00411, 2025.","DOI":"10.1109\/ACDSA67686.2026.11467941"},{"key":"7793_CR11","doi-asserted-by":"crossref","unstructured":"Singh G, Jin H, Merz Jr KM. Benchmarking MedMNIST dataset on real quantum hardware. arXiv:2502.13056, 2025.","DOI":"10.1038\/s41598-026-35605-3"},{"issue":"5","key":"7793_CR12","doi-asserted-by":"publisher","first-page":"1122","DOI":"10.1016\/j.cell.2018.02.010","volume":"172","author":"DS Kermany","year":"2018","unstructured":"Kermany DS, Goldbaum M, Cai W, Valentim CS, Liang H, Baxter S, et al. Identifying medical diagnoses and treatable diseases by image-based deep learning. Cell. 2018;172(5):1122-1131.e9.","journal-title":"Cell"},{"issue":"1","key":"7793_CR13","doi-asserted-by":"publisher","DOI":"10.1148\/ryai.2019180041","volume":"1","author":"G Shih","year":"2019","unstructured":"Shih G, Wu CC, Halabi SS, Kohli MD, Prevedello LM, Cook TD, et al. Augmenting the NIH chest radiograph dataset with expert annotations of possible pneumonia. Radiol Artific Intell. 2019;1(1):e180041.","journal-title":"Radiol Artific Intell"}],"container-title":["BMC Research Notes"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/article\/10.1186\/s13104-026-07793-8","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s13104-026-07793-8.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1186\/s13104-026-07793-8.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,3]],"date-time":"2026-07-03T02:06:09Z","timestamp":1783044369000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1186\/s13104-026-07793-8"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,11]]},"references-count":13,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["7793"],"URL":"https:\/\/doi.org\/10.1186\/s13104-026-07793-8","relation":{},"ISSN":["1756-0500"],"issn-type":[{"value":"1756-0500","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5,11]]},"assertion":[{"value":"31 January 2026","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"26 March 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"11 May 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"Not applicable.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Ethics approval and consent for publication"}},{"value":"The authors declare no conflict of interest.","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing of interests"}}],"article-number":"270"}}