Skip to content
Open access

Uncertainty-Aware Selective Prediction in Brain-Tumour MRI Classification: Selective Risk Reduction and Component-Wise Reliability Findings

Oct 2026 · Bioengineering · Vol 13 · 1 citation · 32 references
Medicine

TL;DR

Confidence- and entropy-based abstention produced a measurable reduction in selective risk at low rejection rates, and it did so because errors concentrated among the cases were withheld by the confidence/entropy rule rather than because predictions were discarded indiscriminately.

Abstract

Background: Automated brain-MRI analysis is usually judged by accuracy at a single operating point, yet a model that is accurate on average may still fail without signalling that it is about to fail. Selective prediction addresses this by allowing a model to abstain, but the resulting trade-off between coverage and risk is rarely quantified in medical imaging. Objective: This study asks whether confidence- and entropy-based selective prediction produces a measurable reduction in selective risk in a biomedical vision–language classifier by concentrating errors among withheld cases, and at what coverage cost. Three further components of a brain-MRI-analysis framework—volumetric segmentation, glioma grading and structured report generation—are evaluated as complementary evidence about the reliability profiles of different modules, not as independent contributions. Methods: A LoRA-fine-tuned LLaVA-Med classifier was evaluated on the 1000-image BRISC2025 test partition, with a disjunctive rejection rule based on softmax confidence and predictive entropy. Volumetric segmentation used nnU-Net v2 (3D full-resolution) on BraTS2020 with a three-fold ensemble evaluated on 74 held-out cases; glioma grading (high- versus low-grade) used DenseNet121 on multi-slice representations; report generation used Meta Llama 3.2 3B Instruct fine-tuned with QLoRA on structured findings. All proportions are reported with 95% Wilson score intervals under a single analysis plan documented before the reported analyses were finalised. Results: Withholding 1.80% of predictions (95% CI 1.14–2.83) raised selective accuracy to 97.76% (95% CI 96.63–98.52) and reduced selective risk from 3.40% to 2.24%, a relative reduction of 34.1%. Withheld predictions were strongly enriched for errors (66.7% versus 2.24%; odds ratio 87.3, Fisher exact p = 4.6 × 10−15), capturing 35.3% of all errors while issuing 99.38% of the predictions that were correct. Considered as a detector of its own errors, the rule has a balanced accuracy of 67.34%; this describes the binary rejection decision at the realised threshold and does not identify the AUROC of the underlying continuous score, which the single operating point leaves anywhere in [0.351, 0.996]. Because the per-sample probability vectors were not retained, the full risk–coverage curve, the rule ablations and calibration error could not be computed, and the thresholds (τc = 0.6, τH = 1.0) were selected empirically on an undocumented partition; the selective-prediction finding is therefore reported as a single-operating-point case study and is exploratory. Among the complementary components, three-fold ensembling improved held-out mean Dice from 0.7948 to 0.8150 (+0.0202), reported descriptively because the per-case values needed for a paired test were not preserved. Glioma grading did not achieve better-than-chance discrimination: balanced accuracy 49.91%, Cohen’s κ −0.0023, Matthews correlation coefficient −0.0026, low-grade recall 8.96% (95% CI 5.81–13.57), and accuracy (74.03%) below a constant majority-class classifier (79.46%); the associated McNemar statistic is reported descriptively because slice-level observations are not independent. Report generation produced high automatic similarity scores (BERTScore F1 0.9513) that are not interpretable without baselines. Conclusions: Confidence- and entropy-based abstention produced a measurable reduction in selective risk at low rejection rates, and it did so because errors concentrated among the cases were withheld by the confidence/entropy rule rather than because predictions were discarded indiscriminately. At the realised coverage the rule captured 65.5% (95% bootstrap CI 41.9–87.8) of the reduction achievable by an oracle that rejects only true errors. The component-wise comparison shows that this reliability behaviour cannot be assumed for every module: high aggregate accuracy and useful discrimination came apart in the grading task. These findings are component-wise and do not constitute an end-to-end, externally validated or clinically validated pipeline. The vision–language evaluation is at image level: BRISC2025 carries no patient identifiers, so patient-level independence cannot be established and the reported intervals should be read with that constraint in mind.

Read PDF

Similar papers

Open access 2026

Uncertainty-Aware Brain-Tumor Type Classification From MRI: Segmentation-Guided ROI, Conformal Prediction, and Selective Abstention Under Distribution Shift

An uncertainty-aware classification framework is presented that couples segmentation-guided region-of-interest (ROI) localization with post-hoc calibration, split-conformal prediction, and selective abstention, and a deferral rule that abstains on unreliable inputs.

N. L. Nguyen, Tuan van Huynh, Liet van Dang · 0 citations
Review Oct 2026

Enhancing clinical trust in brain tumor MRI: a meta-analysis of AI classification, segmentation, and explainability

A systematic review and meta-analysis of artificial intelligence models for brain tumor analysis using MRI shows that while AI models already approach expert-level performance, their successful clinical integration depends on advancing explainable, transparent, and reproducible frameworks.

Kashif Mazhar, Pragya Dwivedi · 0 citations
#artificial intelligence Open access Sep 2026

Towards Trustworthy AI for Glioma Diagnosis: A Task-Aware Evaluation of Uncertainty Quantification

Uncertainty Quantification (UQ) is a key requirement for trustworthy AI in high-stakes medical image analysis. In this work, we evaluate UQ within a multi-task Deep Learning (DL) framework for MRI-based glioma diagnosis that performs tumor segmentation and predicts Isocitrate Dehydrogenase (IDH) mutation status, 1p/19q...

Gonzalo E. Mosquera-Rojas, S. R. van der Voort, Carolin M. Pirkl et al. · 0 citations
Review Open access 2026

A COMPARATIVE STUDY OF AI MODELS FOR LUNG CANCER PREDICTION AND CLINICAL DECISION SUPPORT

Artificial-intelligence (AI) research on lung cancer in 2026 shows rapid growth across lightweight convolutional networks, CNN–transformer hybrids, capsule-inspired models, segmentation-driven pipelines, radiomics, multimodal fusion and reader-assistance systems. Model selection is now more than accuracy maximisation:...

A. Sirajdeen, S. Rajakumaran · 0 citations
Review Open access Oct 2026

SYNAPSE-X: a calibration-aware vision transformer framework for selective referral in multi-class brain tumor MRI classification

SYNAPSE-X is presented as a research-stage reliability-assessment framework for calibrated brain tumor MRI classification, supporting uncertainty-aware referral while acknowledging limitations related to public image-level datasets, unavailable patient-level identifiers, and the need for prospective clinical validation...

Sadaf Tabatabaee, M. Khasawneh · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.