KneeFS-Net: A Lightweight Deep Learning Model for Knee Osteoarthritis Severity Assessment with Cross-Domain External Validation
Authors: Shafqat Alam, Galib Anjum Talukder Mahi, Yeasir Rabbani Tanvir, Md. Ibrahim, Md. Shahed Hossain, Shashank Ghimire, Mahmud Hasan Mostofa Kamal, ASM Shahidul Hossain, Taufiq Hasan
Organizations: Department of Biomedical Engineering, Bangladesh University of Engineering and Technology (BUET), Dhaka, Bangladesh · Telerad Bangladesh Ltd., Dhaka, Bangladesh · Department of Radiology and Imaging, DNCC Dedicated COVID 19 Hospital, Dhaka, Bangladesh · Department of Radiology and Imaging, Bangladesh Medical University, Dhaka, Bangladesh · Center for Bioengineering Innovation and Design, Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD, USA
Knee osteoarthritis (OA) grading from radiographs using the Kellgren-Lawrence (KL) scale requires expertise that may be scarce. We present KneeFS-Net, a lightweight ordinal grader (0.8 million parameters; 0.33 GFLOPs) combining spatial, Fourier and state-space features with landmark-guided attention. On the Osteoarthritis Initiative (OAI) test set, KneeFS-Net achieved a quadratic weighted kappa (QWK) of 0.863, mean absolute error (MAE) of 0.325 and 97.2% within-one-grade accuracy, comparable to models with 9-35-fold more parameters. We evaluated cross-population generalization on two Bangladeshi cohorts, a single-center hospital cohort and a multi-center teleradiology cohort. Without further training or adaptation, the OAI-trained model achieved QWK 0.803, MAE 0.339 and 98.2% within-one-grade accuracy on its hospital cohort, declining to QWK 0.656 and MAE 0.522 on multi-center teleradiology images. Self-supervised pretraining on 6,092 unlabeled radiographs followed by joint fine-tuning achieved QWK 0.823, MAE 0.329 and 99.7% within-one-grade accuracy in cross-validation. These findings demonstrate accurate, efficient KL grading and potential adaptation to new populations with limited annotation.
Figures & tables
Figure 1: Grading errors in the three cohorts and calibration on OAI. A OAI test set ( n=826 ). B Reliability diagram on the OAI test set; bars show the number of knees per confidence bin. C BMU Cohort ( n=271 ). D Multi-District Cohort, Experiment F, out-of-fold predictions ( n=1,000 ). Confusion matrices are row-normalized, so the diagonal gives the sensitivity for each grade.
Model size and cost
OAI test set ( n=826 )
Multi-District, single split ( n=200 )
Model
Params ↓
GFLOPs (per img) ↓
Peak Mem. (MB) ↓
QWK ↑
Acc ↑
Sens. (macro) ↑
Spec. (macro) ↑
QWK ↑
Acc ↑
Sens. (macro) ↑
Spec. (macro) ↑
Part A: Established architectures
CNN-based Models
ResNet50
24,363,588
8.26
840.71
0.8493
68.16
72.81
91.39
0.8021
71.50
74.85
90.22
DenseNet121
7,547,268
5.79
434.59
0.8610
70.80
74.10
91.35
0.8112
71.50
75.21
93.27
ConvNeXt-T
28,226,788
8.97
999.52
0.7627
56.42
60.72
88.02
0.7245
67.00
70.64
84.91
Table 1: Performance and computational cost of KneeFS-Net and comparator models
Figure 2: Agreement versus model size. QWK against the number of trainable parameters (log scale). A OAI test set ( n=826 ). B Multi-District Cohort, single 200-knee split. Recreated KOA-specific models are labeled by first author; full results are in Table 1 .
Figure 3: Agreement on the Multi-District Cohort under six training strategies. QWK against the consensus on out-of-fold predictions ( n=1,000 ), with 95% CIs. Orange, no Multi-District labels used; blue, Multi-District labels used. The shaded band shows the mean inter-radiologist QWK (95% CI) for context only, and the dashed line shows KneeFS-Net on the OAI test set. Brackets mark the planned comparisons (paired bootstrap tests).
Setup/ Metric
Exp A
Exp B
Exp C
Exp D
Exp E
Exp F
Exp. Name
Direct Evaluation
Trained on MD
UDA
SSL init. + OAI fine tune
OAI + MD Joint train
SSL init. + Joint train
Train Data
OAI Train set
MD training folds
OAI Train + MD unlabeled ( n =6,092)
MD unlab. ( n =6,092) + OAI Train
OAI Train + MD training folds
OAI Train + MD training folds
QWK
0.6560
0.7067
0.7030
0.7379 ∗∗∗
0.7826
0.8234 ††‡‡‡
(0.6165–0.6942)
(0.6681–0.7389)
(0.6648–0.7394)
(0.7103–0.7639)
(0.7604–0.8034)
(0.8002–0.8451)
Accuracy
53.60
58.50
66.30
61.30 ∗∗∗
61.90
67.40 ††
(50.30–56.80)
(55.40–61.60)
(63.40–69.10)
(58.40–64.30)
(58.90–64.90)
(64.50–70.30)
Table 2: Performance of six training strategies on the Multi-District Cohort (5-fold cross-validation, n=1,000 )
BMU Cohort ( n=271 )
Multi-District Cohort ( n=1,000 )
Metric / Comparison
Estimate (95% CI)
Δ (95% CI)
Estimate (95% CI)
Δ (95% CI)
Overall model performance vs. consensus
Exact accuracy (%)
67.90 (62.36–73.06)
—
67.40 (64.50–70.30)
—
Within ±1 grade (%)
98.15 (96.31–99.63)
Differed >1 grade: 1.85% (5 knees)
99.70 (99.30–100.00)
Differed >1 grade: 0.30% (3 knees)
QWK
0.803 (0.749–0.848)
Primary metric
0.823 (0.800–0.845)
Primary metric
MAE
0.339 (0.284–0.399)
—
0.329 (0.299–0.358)
—
Table 3: Agreement of KneeFS-Net with the radiologist consensus and with individual radiologists (BMU Cohort, direct evaluation; Multi-District Cohort, Experiment F)
BUET-KOA
Characteristic
OAI
BMU Cohort
Multi-District Cohort
Age, years
Male
64.11 ± 9.74
53.04 ± 5.67
39.56 ± 19.65
Female
64.56 ± 9.27
53.01 ± 5.36
45.95 ± 16.51
Unspecified
—
—
40.67 ± 18.32
Overall
64.37 ± 9.47
53.02 ± 5.50
42.86 ± 18.31
Table 4: Demographic and clinical characteristics of the study cohorts
Figure 4: Study cohorts. Top: Selection and partitioning of the OAI (A) and BUET-KOA (B) datasets. Bottom left: Centers and radiographs of the Multi-District Cohort by administrative division. Bottom right: PCA and UMAP projections of image-noise descriptors.
Figure 5: KneeFS-Net architecture. Top: Stem, four-stage FSSE encoder, two-scale region attention and ordinal head. Bottom left: FSSE layer; the state-space branch is used in stages 3 and 4 only. Bottom right: Masked region attention.
Figure 6: Self-supervised pretraining of the FSSE encoder. The student encodes two global and four local views; the teacher, updated as an exponential moving average of the student (dashed arrow), encodes the global views only. Teacher outputs are centered per cohort, and a CORAL penalty aligns student features across cohorts (Eq. ( 3 )). The projection head is discarded after pretraining.
Cross-domain generalization studies how machine-learning models retain reliable performance when data, sensors, populations, or tasks differ from training.
Orthopaedic Department, Faculty of Medicine UIN Syarif Hidayatullah Jakarta, Indonesia · Informatics Engineering, Institut Teknologi Sepuluh Nopember, Indonesia · Information Technology, Universitas Muhammadiyah Yogyakarta, Indonesia +1
Departamento de Ciencia de la Computación e Inteligencia Artificial, Universidad de Córdoba, Córdoba, España · Department of Political Science, Communication, and International Relations, University of Macerata, Macerata, Italy · Department of Innovative Technologies in Medicine & Dentistry, Università degli Studi "G. D’Annunzio" Chieti - Pescara, Chieti, Italy +5