MSR-ABC: Complementary Feature Fusion for Clear Cell Renal Cell Carcinoma Classification

Yuan Liang; Sourav Bhattacharjee; Abraham Campbell

cs.CVNov 6, 2026

MSR-ABC: Complementary Feature Fusion for Clear Cell Renal Cell Carcinoma Classification

Authors: Yuan Liang, Sourav Bhattacharjee, Abraham Campbell

Organizations: 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

Abstract

Clear cell renal cell carcinoma (ccRCC) classification from computed tomography (CT) may benefit from representations that capture different aspects of tumour phenotype. However, simply concatenating reconstruction-derived features, image--text embeddings, and radiomic descriptors may not fully exploit their complementary information. We propose MSR-ABC, a global--local framework integrating MedVAE, MedSigLIP, and radiomics. Spatially matched local image descriptors interact through cross-representation attention, while radiomic auxiliary supervision encourages the pooled local representation to retain quantitative tumour attributes. A gated residual mechanism integrates local information with global summaries. Area-aware pooling and visual and soft-text prompting complete the framework. We evaluate discrimination using a patient-level split of 396 KiTS23 cases and an external cohort of 111 TCGA/TCIA cases, reporting mean AUROC over three training seeds with 95% patient-bootstrap confidence intervals. MSR-ABC achieves mean AUROCs of 0.806 internally and 0.879 externally. Internally, it exceeds global MedVAE--radiomics and three-source concatenation by 3.35 and 3.61 AUROC percentage points, respectively. Removing any source lowers mean AUROC in both cohorts, with the largest decreases following radiomics removal. Individual removal of cross-representation attention, auxiliary supervision, or gated fusion also lowers internal mean AUROC. These findings are consistent with complementary contributions within the integrated architecture and support further investigation of structured representation fusion for non-invasive renal tumour characterisation.

Figures & tables

Figure 1: MSR-ABC architecture. The CT illustration is schematic.

S^{*}

denotes VPT-adapted global S,

j

the global summary,

q

the pooled local image summary, and

p

the mapped

[q,R]

representation. Brackets denote concatenation after preprocessing. The circled plus adds fusion and image–text logits. Solid and dashed arrows denote inference and training-only paths. Dots mark branches. Crossings without dots are unconnected. B backpropagation is omitted.

Model	Test ( $n=60$ )	External ( $n=111$ )
Global M	0.5922 [0.443, 0.737]	0.8335 [0.763, 0.895]
R-LR	0.7513 [0.589, 0.892]	0.8597 [0.781, 0.927]
Global MR	0.7729 [0.623, 0.898]	0.8792 [0.811, 0.937]
Global MSR	0.7703 [0.616, 0.902]	0.8893 [0.822, 0.945]
ABC + area	0.7897 [0.641, 0.914]	0.8691 [0.795, 0.931]
+ SoftText	0.7897 [0.641, 0.915]	0.8701 [0.796, 0.932]

Table 1: Mean AUROC [95% patient-bootstrap CI] over three seeds. R-LR is a deterministic single-model reference. Prompt rows add SoftText, VPT, or both to ABC + area. Area denotes area-aware pooling.

Variant	Test ( $n=60$ )	External ( $n=111$ )
Full	0.8064 [0.658, 0.927]	0.8790 [0.809, 0.938]
w/o M source	0.7897 [0.645, 0.911]	0.8374 [0.756, 0.909]
w/o S source	0.7937 [0.646, 0.915]	0.8564 [0.781, 0.921]
w/o R source	0.7288 [0.568, 0.872]	0.8142 [0.729, 0.890]
w/o A	0.7972 [0.649, 0.921]	0.8735 [0.802, 0.934]
w/o B	0.7994 [0.650, 0.922]	0.8784 [0.809, 0.937]

Table 2: Source and component ablations: three-seed mean AUROC [95% patient-bootstrap CI]. CIs are for AUROC, not differences. Full denotes MSR-ABC, and w/o means without. Source removals include dependent functionality.

Cancer Subtype Classification

See all 17 papers

Sep 22, 2026

Complementary Roles of Radiomics and Foundation Representations in Renal Cell Carcinoma Classification: A Comparative Study of 2D and 3D CT Encodings

Accurate preoperative subtype classification of renal cell carcinoma (RCC) from contrast-enhanced computed tomography remains clinically challenging. Radiomics provides structured tumour descriptors, whereas foundation representations offer transferable image features. However, it remains unclear whether radiomics still adds value beyond pretrained representations, and how 2D and 3D MedVAE encoders compare in this setting. We compared handcrafted radiomics, 2D MedVAE, 3D MedVAE, and their fusion for binary clear-cell RCC versus non-clear-cell RCC classification on KiTS23 under a unified preprocessing pipeline. Concatenation, cross-attention, and gated fusion were evaluated as representative integration strategies, and radiomics feature importance was analysed to support decision-centric interpretability. Fusion consistently improved discrimination over image-only MedVAE branches. The best overall performance was achieved by 3D gated fusion, with an AUC of 82.7%, outperforming the best 2D fusion model (79.6%), the radiomics baseline (74.4%), and the single-modality MedVAE branches. Ablation analysis further showed clear gains of the full fusion model over both image-only and radiomics-only variants, indicating complementary contributions from radiomics and image representations. These findings suggest that radiomics remains relevant for RCC CT classification in the presence of foundation representations, and that its integration with MedVAE is more effective in the 3D setting. More broadly, the study supports a complementary role for radiomics and foundation representations in clinically meaningful imaging decision support.

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

Medical Imaging Foundation Models Oncology Radiomics Computed Tomography Feature Fusion

Ask PDF

Sep 22, 2026

Radiomics--Foundation Fusion for Interpretable RCC Classification: Internal Benchmarking and Exploratory External Transfer

Accurate preoperative subtype classification of renal cell carcinoma (RCC) from contrast-enhanced CT remains clinically challenging because clear cell RCC (ccRCC) and non-clear cell RCC often show overlapping imaging appearances. This study evaluates whether foundation representations reduce reliance on handcrafted radiomics, or whether radiomics remains complementary for interpretable tumour characterisation. We compared radiomics, conventional CNN features, MedicalNet-pretrained features, MedVAE representations, and fusion variants for binary ccRCC classification on KiTS23, reporting area under the receiver operating characteristic curve (AUC) with bootstrap confidence intervals and average precision (AP) as a complementary class-imbalance-sensitive metric. We further assessed branch-removal ablation, TCGA/AIMI external transfer, and interpretability using radiomics permutation importance and gate-level analysis. Internally, 3D MedVAE gated fusion achieved the best performance, with an AUC of 82.7% and AP of 92.2%. On the external TCGA cohort, the same model achieved an AUC of 79.5% and AP of 98.9%, although specificity remains uncertain because only two external non-ccRCC cases were available. Gate analysis showed a radiomics-dominant fusion regime, suggesting that foundation representations acted as case-dependent refinement signals rather than replacements for structured tumour descriptors. These findings support radiomics as a complementary and clinically interpretable component of CT-based RCC characterisation in the foundation-model era.

Yuan Liang, Fangyijie Wang, Kathleen M. Curran +3

Research Ireland Centre for Research Training in Machine Learning · School of Computer Science, University College Dublin, Dublin, Ireland · School of Medicine, University College Dublin, Dublin, Ireland +1

Medical Image Analysis Radiomics Medical Imaging Foundation Models Oncology Computed Tomography

Ask PDF

Sep 22, 2026

Radiomics-Conditioned Modulation of RenalCLIP Features for Clear Cell Renal Cell Carcinoma Classification

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

Medical Imaging Foundation Models Radiomics

Ask PDF

Abstract

Cancer Subtype Classification

Explore similar work

Complementary Roles of Radiomics and Foundation Representations in Renal Cell Carcinoma Classification: A Comparative Study of 2D and 3D CT Encodings

Radiomics--Foundation Fusion for Interpretable RCC Classification: Internal Benchmarking and Exploratory External Transfer

Radiomics-Conditioned Modulation of RenalCLIP Features for Clear Cell Renal Cell Carcinoma Classification