Skip to content
DraconiticResearch
Search Sources API
203,948 studies31 sourcesUpdated 11 min ago
Peer-reviewedCC BY 4.0Scientific Reports

Standard and decoupled knowledge distillation for lightweight radiographic knee osteoarthritis grading and triage

  • Umar Hasan
  • Muhammad Ali Nayeem
Open at sourceDOICite

Abstract

This retrospective study compares standard knowledge distillation (KD) and decoupled knowledge distillation (DKD) fo​r lightweight artificial intelligence-based​ radiographic knee osteoarthritis grading. GhostNetV2 students learned five Kellgren–Lawrence grades from a Swin-Tiny teacher and were evaluated against a cross-entropy baseline using an Osteoarthritis Initiative-derived dataset of 8260 images from 4130 patients with patient-disjoint partitions. Across three​ runs, mean binary triage accuracy was 84.04%, 84.82%, and 84.82% for baseline, standard KD, and DKD. Thresholds targeting at least 90% validation specificity yielded test sensitivities of 72.77%, 76.56%, and 75.50%, with specificities of 92.09%, 90.94%, and 91.12%; these operating points are not specificity-matched. Mean five-class accuracy was 64.63%, 67.63%, and 67.71%. Separate illustrative checkpoints showed reduced KL-1-to-KL-2 overcalls with DKD (28.0 to 18.9%) but lower KL-0 recall (90.6 to 78.7%). Calibration, exploratory decision curves, and patient-cluster uncertainty intervals were evaluated. Each student contains 4.88 million parameters. Distillation improved mean performance internally, but decoupling did not demonstrate an additional sensitivity benefit over standard KD. No external cohort or clinician-supported decision policy was evaluated; clinical utility and population-level stability remain unestablished.

Related

  1. Lightweight anomaly detection via knowledge distillation from ensemble teachers
    Xia Cui, Tong Li
    Scientific ReportsPeer-reviewedCC BY-NC-ND 4.0Full textCrossref
  2. Knowledge distillation-based lightweight generative large language models for standardized quality control of Chinese radiology reports
    Duoning Jiang, Xiao Han, Ke Xu +1
    Scientific ReportsPeer-reviewedCC BY-NC-ND 4.0Full textEpidemiology & public healthCrossref
  3. A Lightweight Universal Machine-Learning Interatomic Potential via Knowledge Distillation for Scalable Atomistic Simulations
    Sangmin Oh, J.H. You, Jaesun Kim +4
    Journal of Chemical Information and ModelingPeer-reviewedLicence not statedFull text9 citedMaterials & polymersOpenAlex
  4. A temperature-adaptive logit-standardized knowledge distillation for lightweight network intrusion detection
    XiaoFang Dong, Kai Yang, JingChao Liu +1
    PLOS OnePeer-reviewedCC BYFull textNeuroscienceStatistical & nonlinear physicsPLOSEurope PMCPMC
  5. A lightweight knowledge distillation framework for few shot unsupervised industrial surface anomaly detection and localization
    Xian Gao
    Scientific ReportsPeer-reviewedCC BY 4.0Full textCrossref

Cite

BibTeXRIS
BibTeX
@article{hasan2026standard,
  title = {{Standard and decoupled knowledge distillation for lightweight radiographic knee osteoarthritis grading and triage}},
  author = {Umar Hasan and Muhammad Ali Nayeem},
  journal = {Scientific Reports},
  year = {2026},
  month = oct,
  doi = {10.1038/s41598-026-78270-2},
  url = {https://www.nature.com/articles/s41598-026-78270-2},
  note = {CC BY 4.0},
}
RIS
TY  - JOUR
TI  - Standard and decoupled knowledge distillation for lightweight radiographic knee osteoarthritis grading and triage
AU  - Umar Hasan
AU  - Muhammad Ali Nayeem
PY  - 2026
DA  - 2026/10/31
JO  - Scientific Reports
DO  - 10.1038/s41598-026-78270-2
UR  - https://www.nature.com/articles/s41598-026-78270-2
AB  - This retrospective study compares standard knowledge distillation (KD) and decoupled knowledge distillation (DKD) fo​r lightweight artificial intelligence-based​ radiographic knee osteoarthritis grading. GhostNetV2 students learned five Kellgren–Lawrence grades from a Swin-Tiny teacher and were evaluated against a cross-entropy baseline using an Osteoarthritis Initiative-derived dataset of 8260 images from 4130 patients with patient-disjoint partitions. Across three​ runs, mean binary triage accuracy was 84.04%, 84.82%, and 84.82% for baseline, standard KD, and DKD. Thresholds targeting at least 90% validation specificity yielded test sensitivities of 72.77%, 76.56%, and 75.50%, with specificities of 92.09%, 90.94%, and 91.12%; these operating points are not specificity-matched. Mean five-class accuracy was 64.63%, 67.63%, and 67.71%. Separate illustrative checkpoints showed reduced KL-1-to-KL-2 overcalls with DKD (28.0 to 18.9%) but lower KL-0 recall (90.6 to 78.7%). Calibration, exploratory decision curves, and patient-cluster uncertainty intervals were evaluated. Each student contains 4.88 million parameters. Distillation improved mean performance internally, but decoupling did not demonstrate an additional sensitivity benefit over standard KD. No external cohort or clinician-supported decision policy was evaluated; clinical utility and population-level stability remain unestablished.
SN  - 2045-2322
ER  -