Anas M. Zafar

Anas Zafar, Leema Krishna Murali, Ashish Vashist

Recent work shows that text-only reinforcement learning with verifiable rewards (RLVR) can match or outperform image-text RLVR on multimodal medical VQA benchmarks, suggesting current evaluation protocols may fail to measure causal visual dependence. We introduce a counterfactual evaluation framework using real, blank, and shuffled images across four medical VQA benchmarks: PathVQA, PMC-VQA, SLAKE, and VQA-RAD. Beyond accuracy, we measure Visual Reliance Score (VRS), Image Sensitivity (IS), and introduce Hallucinated Visual Reasoning Rate (HVRR) to detect cases where models generate visual claims despite producing image-invariant answers. Our findings reveal that RLVR improves accuracy while degrading visual grounding: text-only RLVR achieves negative VRS on PathVQA (-0.09), performing better with mismatched images, while image-text RLVR reduces image sensitivity to 39.8% overall despite improving accuracy. On VQA-RAD, both variants achieve 63% accuracy through different mechanisms: text-only RLVR retains 81% performance with blank images, while image-text RLVR shows only 29% image sensitivity. Models generate visual claims in 68-74% of responses, yet 38-43% are ungrounded (HVRR). These findings demonstrate that accuracy-only rewards enable shortcut exploitation, and progress requires grounding-aware evaluation protocols and training objectives that explicitly enforce visual dependence.

Syeda Hareem Madani, Noureen Bibi, Adam Rafiq Jeraj et al.

Anatomical brain parcellations dominate rs-fMRI-based Autism Spectrum Disorder (ASD) classification, yet their rigid boundaries may fail to capture the idiosyncratic connectivity patterns that characterise ASD. We present a graph-based deep learning framework comparing anatomical (AAL, 116 ROIs) and functionally-derived (MSDL, 39 ROIs) parcellation strategies on the ABIDE I dataset. Our FSL preprocessing pipeline handles multi-site heterogeneity across 400 balanced subjects, with site-stratified 70/15/15 splits to prevent data leakage. Gaussian noise augmentation within training folds expands samples from 280 to 1,680. A three phase pipeline progresses from a baseline GCN with AAL (73.3% accuracy, AUC=0.74), to an optimised GCN with MSDL (84.0%, AUC=0.84), to a Graph Attention Network ensemble achieving 95.0% accuracy (AUC=0.98), outperforming all recent GNN-based benchmarks on ABIDE I. The 10.7-point gain from atlas substitution alone demonstrates that functional parcellation is the most impactful modelling decision. Gradient-based saliency and GNNExplainer analyses converge on the Posterior Cingulate Cortex and Precuneus as core Default Mode Network hubs, validating that model decisions reflect ASD neuropathology rather than acquisition artefacts. All code and datasets will be publicly released upon acceptance.

Muthu Subash Kavitha, Anas Zafar, Amgad Muneer et al.

Clinical deployment of chest radiograph classifiers requires models that can be updated as new datasets become available without retraining on previously ob- served data or degrading validated performance. We study, for the first time, a task-incremental continual learning setting for chest radiograph classification, in which heterogeneous chest X-ray datasets arrive sequentially and task identifiers are unavailable at inference. We propose a continual adapter-based routing learning strategy for Chest X-rays (CARL-XRay) that maintains a fixed high-capacity backbone and incrementally allocates lightweight task-specific adapters and classifier heads. A latent task selector operates on task-adapted features and leverages both current and historical context preserved through compact prototypes and feature-level experience replay. This design supports stable task identification and adaptation across sequential updates while avoiding raw-image storage. Experiments on large-scale public chest radiograph datasets demonstrate robust performance retention and reliable task-aware inference under continual dataset ingestion. CARL-XRay outperforms joint training under task-unknown deployment, achieving higher routing accuracy (75.0\% vs.\ 62.5\%), while maintaining competitive diagnostic performance with AUROC of 0.74 in the oracle setting with ground-truth task identity and 0.75 under task-unknown inference, using significantly fewer trainable parameters. Finally, the proposed framework provides a practical alternative to joint training and repeated full retraining in continual clinical deployment.