Retrieving Patient-Specific Radiomic Feature Sets for Transparent Knee MRI Assessment
Authors: Yaxi Chen, Simin Ni, Jingjing Zhang, Shaheer U. Saeed, Yipei Wang, Aleksandra Ivanova, Rikin Hargunani, Chaozong Liu, +2 more
Organizations: Mechanical Engineering Department, University College London, London, UK · Hawkes Institute, University College London, London, UK · Institute of Orthopaedic & Musculoskeletal Science, University College London, Royal National Orthopaedic Hospital, Stanmore, UK · School of Engineering and Materials Science, Queen Mary University of London, London, UK · Centre for Bioengineering, Queen Mary University of London, London, UK · Department of Medical Physics and Biomedical Engineering, University College London, London, UK · Royal National Orthopaedic Hospital, Stanmore, UK
Classical radiomic features are designed to quantify image appearance and intensity patterns. Compared with end-to-end deep learning (DL) models trained for disease classification, radiomics pipelines with low-dimensional parametric classifiers offer enhanced transparency and interpretability, yet often underperform because of the reliance on population-level predefined feature sets. Recent work on adaptive radiomics uses DL to predict feature weights over a radiomic pool, then thresholds these weights to retain the top-k features from large radiomic pool F (often ~10^3). However, such marginal ranking can over-admit redundant descriptors and overlook complementary feature interactions. We propose a patient-specific feature-set selection framework that predicts a single compact feature set per subject, targeting complementary and diverse evidence rather than marginal top-k features. To overcome the intractable combinatorial search space of F choose k features, our method utilizes a 2-stage retrieval strategy: randomly sample diverse candidate feature sets, then rank these sets with a learned scoring function to select a high-performing feature set for the specific patient. The system consists of a feature-set scorer, and a classifier that performs the final diagnosis. We empirically show that the proposed two-stage retrieval approximates the original exhaustive all k-feature selection. Validating on tasks including ACL tear detection and KL grading for osteoarthritis, the experimental results achieve diagnostic performance, outperforming the top-k approach with the same k values, and competitive with end-to-end DL models while maintaining high transparency. The model generates auditable feature sets that link clinical outcomes to specific anatomical regions and radiomic families, allowing clinicians to inspect which anatomical structures and quantitative descriptors drive the prediction.
Figures & tables
Figure 1: Motivation and feasibility of feature-set retrieval. Left: Candidate feature-set score distribution (blue histogram) estimated via kernel density estimation (KDE); the red-shaded band marks the top-1 score range, illustrating near-optimal set identification without enumerating. Right: Illustration of top- k vs. proposed feature-set selection approaches, where top- k may select correlated features (limited coverage), whereas feature-set selection provides complementary evidence (broader coverage).
Figure 2: Overview of the proposed feature set selection framework. The red arrow indicates the stage 1 retrieval and the blue arrow indicates the stage 2 retrieval.
Figure 3: Case studies for model interpretation. Examples of (Case 1) partial ACL tear, (Case 2) non-tear, and (Case 3) KL4 knee osteoarthritis. The histogram shows, for each ROI, the number of selected features contributing to the prediction. The symbols and their adjacent arrows indicate whether the corresponding feature values are relatively high or low compared with the rest of the dataset (z-score).
Evaluation Metrics
Method
k
Pstep
Npool
Ens
Task
Acc.
MacroF1
BAcc
AUC
QWK
Pstep ↓
30
5
1000
3
ACL
0.59±0.05
0.36±0.06
0.38±0.07
0.59±0.02
0.18±0.03
OA-b
0.69±0.16
0.65±0.17
0.66±0.17
0.80±0.06
-
OA-5
0.37±0.02
0.26±0.04
0.31±0.03
0.66±0.03
0.30±0.06
k=20
20
10
1000
3
ACL
0.67±0.03
0.48±0.07
0.46±0.06
0.65±0.03
0.34±0.07
OA-b
0.74±0.05
0.69±0.08
0.68±0.07
0.71±0.20
-
Table 1: Comparison of different configurations and methods. k: the feature set size; Pstep: number of probe gradient steps; Npool: size of the candidate pool in stage-2; Ens: the use of ensembling during inference; RS: random sampling; OA-5: 5-class classification; OA-B: binary classification. We bold the top-2 Acc/BAcc scores.
Radiomics provides quantitative descriptions of tumour appearance that may complement disease-specific foundation models in small labelled cohorts. We investigate this complementarity for computed tomography-based classification of clear cell renal cell carcinoma. Our framework uses radiomics to modulate RenalCLIP features through feature-wise linear modulation (FiLM), while retaining a direct radiomics contribution. Internal testing and external validation compare it with conventional fusion strategies and reference classifiers. The FiLM model achieves an area under the receiver operating characteristic curve (AUC) of 0.804 internally and 0.854 externally, with the highest mean AUC among the evaluated RenalCLIP fusion strategies in both cohorts. Pathway ablations examine the contributions of conditional modulation and the direct radiomics residual, while feature permutation highlights the role of tumour texture. These findings support radiomics as a useful complement to RenalCLIP in a small labelled cohort and identify FiLM as an effective approach to integrating their representations for robust renal tumour classification.
Yuan Liang, Sourav Bhattacharjee, Abraham Campbell
Research Ireland Centre for Research Training in Machine Learning · School of Computer Science, University College Dublin, Dublin, Ireland · School of Veterinary Medicine, University College Dublin, Dublin, Ireland
Grading knee osteoarthritis (KOA) on plain radiographs is poorly reproducible across readers. A single-grade disagreement on the Kellgren-Lawrence (KL) scale can alter surgical management or redirect a patient from conservative therapy to intra-articular injection. Meanwhile, deep learning models that outperform human readers often offer no explanation for their decisions. We present Knee-xRAI, a pipeline that decomposes the grading process by mimicking clinical radiological workflows. It independently measures joint space narrowing (JSN), osteophytes, and subchondral sclerosis, then combines these findings into an explainable KL grade. Specifically, a U-Net++ architecture quantifies JSN via contour segmentation, an SE-ResNet-50 multi-task network grades osteophytes per anatomical site on the OARSI scale, and a hybrid texture-CNN detects binary sclerosis. This pipeline yields a 50-dimensional feature vector evaluated via an XGBoost-SHAP classifier (Path A, audit) and a ConvNeXt hybrid predictor (Path B, deployed). On 8,260 OAI-derived radiographs, the JSN module achieved a Dice score of 0.8909 and an mJSW ICC of 0.8674. Path A reached a QWK of 0.6294 and an AUC of 0.8046, confirming the structured feature vector carries substantial diagnostic signal. Path B achieved a QWK of 0.8436 and an AUC of 0.9017. SHAP analysis identifies JSN as the dominant feature, with osteophytes adding a consistent increment and sclerosis contributing marginally. Removing JSN evidence collapses KL3-KL4 recall while early grades remain intact, aligning with the KL diagnostic criteria. Knee-xRAI grounds every prediction in an auditable chain of measured radiographic findings, providing clinical transparency at the point of care.
Azmul A. Irfan, Nur Ahmad Khatim, Alfan Alfian Irfan +3
Orthopaedic Department, Faculty of Medicine UIN Syarif Hidayatullah Jakarta, Indonesia · Informatics Engineering, Institut Teknologi Sepuluh Nopember, Indonesia · Information Technology, Universitas Muhammadiyah Yogyakarta, Indonesia +1
Radiomics enables extraction of quantitative imaging biomarkers from medical images and has become an important tool for computer-aided cancer diagnosis. However, radiomics datasets are typically high-dimensional with limited samples, making feature selection a critical step for building reliable predictive models. This study proposes a Gradient-Loss Recursive Feature Elimination (GL-RFE) framework that integrates gradient sensitivity analysis from a deep neural network to identify the most influential radiomic features for lung cancer stage detection. A total of 106 radiomic features were extracted from chest Computed Tomography (CT) scans using the PyRadiomics extension of the 3D Slicer platform. The proposed method evaluates feature importance by computing gradients of the network loss with respect to input features and recursively eliminates features with minimal contribution. The resulting top-15 radiomic features are used to train a deep neural network classifier for distinguishing early-stage and advanced-stage lung cancer. The proposed framework achieves strong classification performance, with accuracy of 90.22%, precision of 90.10%, recall of 90.24%, and F1-score of 90.16% on the test dataset. Visualization analyses, including correlation heat maps and distribution plots, further confirm reduced feature redundancy and improved class separability. Compared to conventional feature selection techniques, GL-RFE effectively captures nonlinear feature interactions and enhances model generalization. The presented protocol provides a reproducible and interpretable methodology for radiomics-based cancer stage detection and is particularly suitable for high-dimensional, small-sample biomedical datasets, with potential applications in other domains such as genomics and multimodal clinical analysis.
Hina Shakir, Mohammad Mohatram, Javeed Hussain +2
Department of Software Engineering, Bahria University, Karachi, Pakistan · Global College of Engineering and Technology, Muscat, Oman · Software Engineering & Business Incubation Center, Bahria University Karachi, Pakistan