ARXIV:2603.13182 · MEDICAL AI · SUBMITTED 02 APR · 02:30 UTC · FRESHNESS STALE

VerifiedSource: PDF linkedPartialPaperPack: 3 of 4 citation fields filledMissingMissing fields: authorsPartialProof: unverified proof status

Diffusion-Based Feature Denoising and Using NNMF for Robust Brain Tumor Classification

arXiv

A robust framework for brain tumor classification using NNMF and lightweight CNNs to enhance adversarial robustness.

Blocked on Code›Score3.0Evidence unverified

Opportunity summary

Pain A robust framework for brain tumor classification using NNMF and lightweight CNNs to enhance adversarial robustness.

Evidence 0 refs | 0 sources | 17% coverage

Blocker Evidence unverified

Open Build Read PDF Signal Canvas Track

PROBLEM

A robust framework for brain tumor classification using NNMF and lightweight CNNs to enhance adversarial robustness. In recent years, deep learning models have achieved high classification accuracy.

METHOD

Full abstract

Brain tumor classification from magnetic resonance imaging, which is also known as MRI, plays a sensitive role in computer-assisted diagnosis systems. In recent years, deep learning models have achieved high classification accuracy. However, their sensitivity to adversarial perturbations has become an important reliability concern in medical applications. This study suggests a robust brain tumor classification framework that combines Non-Negative Matrix Factorization (NNMF or NMF), lightweight convolutional neural networks (CNNs), and diffusion-based feature purification. Initially, MRI images are preprocessed and converted into a non-negative data matrix, from which compact and interpretable NNMF feature representations are extracted. Statistical metrics, including AUC, Cohen's d, and p-values, are used to rank and choose the most discriminative components. Then, a lightweight CNN classifier is trained directly on the selected feature groups. To improve adversarial robustness, a diffusion-based feature-space purification module is introduced. A forward noise method followed by a learned denoiser network is used before classification. System performance is estimated using both clean accuracy and robust accuracy under powerful adversarial attacks created by AutoAttack. The experimental results show that the proposed framework achieves competitive classification performance while significantly enhancing robustness against adversarial perturbations.The findings presuppose that combining interpretable NNMF-based representations with a lightweight deep approach and diffusion-based defense technique supplies an effective and reliable solution for medical image classification under adversarial conditions.

RESULT

ScienceToStartup currently rates this 3.0/10 on the public viability pass. To improve adversarial robustness, a diffusion-based feature-space purification module is introduced.

WHY NOW

Medical AI moved forward this cycle; last verified April 2026. Public score 3.0/10.

Continue into Read for claims, analysis, references, and neighboring papers.

Opportunity summary

Score3.0

PainA robust framework for brain tumor classification using NNMF and lightweight CNNs to enhance adversarial robustness.

Evidence0 refs | 0 sources | 17% coverage

Blockermissing authors

Analysis summary

A robust framework for brain tumor classification using NNMF and lightweight CNNs to enhance adversarial robustness.

VerifiedSource: PDF linkedPartialPaperPack: 3 of 4 citation fields filledMissingMissing fields: authorsPartialProof: unverified proof status

Competitive landscape

A robust framework for brain tumor classification using NNMF and lightweight CNNs to enhance adversarial robustness.

Segment

Medical AI

Adoption evidence

No public code link in the paper record yet

Commercial read

3.0/10 public viability

Direct

not classified

Adjacent

not classified

Substitute

not classified

Unknown

not classified

References(15)

The Past, Present and Future of the ELKH Cloud

2022Mihály Héder, Ernő Rigó et al.

Effect of data leakage in brain MRI classification using 2D convolutional neural networks

2021Ekin Yagis, S. Atnafu et al.

Development of partial least squares regression with discriminant analysis for software bug prediction

2020R. Ferenc, István Siket et al.

Feature Extraction Based on the Non-Negative Matrix Factorization of Convolutional Neural Networks for Monitoring Domestic Activity With Acoustic Signals

2020Seokjin Lee, Hee-Suk Pang

Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attacks

2020Francesco Croce, Matthias Hein

Non-Negative Matrix Factorization-Convolutional Neural Network (NMF-CNN) For Sound Event Detection

2020Teck Kai Chan, C. Chin et al.

Semi-NMF network for image classification

2019Haonan Huang, Zuyuan Yang et al.

Brain Tumor Detection Using Convolutional Neural Network

2019Tonmoy Hossain, Fairuz Shadmani Shishir et al.

Nonnegative Matrix Factorization

2019Ke-Lin Du, M. Swamy

A survey on deep learning in medical image analysis

2017G. Litjens, Thijs Kooi et al.

Intriguing properties of neural networks

2013Christian Szegedy, Wojciech Zaremba et al.

Semantic Segmentation with

2012Second-Order Pooling, Rui Caseiro et al.

An Introduction to Variable and Feature Selection

2003Isabelle M Guyon, A. Elisseeff

Algorithms for Non-negative Matrix Factorization

2000Daniel D. Lee, H. Sebastian et al.

Learning the parts of objects by non-negative matrix factorization

1999Daniel D. Lee, H. S. Seung

{ "contract_version": "paper-r2", "paper_id": "ee08f204-6a6c-4a13-920d-3295907552e7", "arxiv_id": "2603.13182", "canonical_route": "/paper/diffusion-based-feature-denoising-and-using-nnmf-for-robust-brain-tumor-classification", "active_tab": "synced from current hash by the drawer client", "selected_artifact": "diffusion-based-feature-denoising-and-using-nnmf-for-robust-brain-tumor-classification", "endpoints": { "paper_pack": "/api/v1/paper/diffusion-based-feature-denoising-and-using-nnmf-for-robust-brain-tumor-classification/paper-pack", "build_passport": "/api/v1/paper/diffusion-based-feature-denoising-and-using-nnmf-for-robust-brain-tumor-classification/build-passport", "mcp_resource": "sciencetostartup://surfaces/paper-workspace" } }

{ "surface": "paper", "mode": "paper", "query": "Diffusion-Based Feature Denoising and Using NNMF for Robust Brain Tumor Classification", "normalized_query": "2603.13182", "route": "/paper/diffusion-based-feature-denoising-and-using-nnmf-for-robust-brain-tumor-classification", "paper_ref": "diffusion-based-feature-denoising-and-using-nnmf-for-robust-brain-tumor-classification", "topic_slug": null, "benchmark_ref": null, "dataset_ref": null }

{ "@context": "https://schema.org", "@graph": [ { "@type": "WebPage", "@id": "https://sciencetostartup.com/paper/diffusion-based-feature-denoising-and-using-nnmf-for-robust-brain-tumor-classification#webpage", "url": "https://sciencetostartup.com/paper/diffusion-based-feature-denoising-and-using-nnmf-for-robust-brain-tumor-classification", "name": "Diffusion-Based Feature Denoising and Using NNMF for Robust Brain Tumor Classification", "description": "A robust framework for brain tumor classification using NNMF and lightweight CNNs to enhance adversarial robustness.", "isPartOf": { "@id": "https://sciencetostartup.com/#website" } }, { "@type": "ScholarlyArticle", "@id": "https://sciencetostartup.com/paper/diffusion-based-feature-denoising-and-using-nnmf-for-robust-brain-tumor-classification#scholarlyArticle", "headline": "Diffusion-Based Feature Denoising and Using NNMF for Robust Brain Tumor Classification", "description": "A robust framework for brain tumor classification using NNMF and lightweight CNNs to enhance adversarial robustness.", "url": "https://sciencetostartup.com/paper/diffusion-based-feature-denoising-and-using-nnmf-for-robust-brain-tumor-classification", "sameAs": "https://arxiv.org/abs/2603.13182", "identifier": { "@type": "PropertyValue", "propertyID": "arXiv", "value": "2603.13182" }, "isAccessibleForFree": true, "isPartOf": { "@id": "https://sciencetostartup.com/#website" }, "datePublished": "2026-03-13T17:15:46.000Z", "citation": [ { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "e9f8645b6dffcd8caa1c93830c1f9c22c8898726" }, "url": "https://www.semanticscholar.org/paper/e9f8645b6dffcd8caa1c93830c1f9c22c8898726" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "b9663073f33a5cff52e7d50b6e7a296cdb9d519c" }, "url": "https://www.semanticscholar.org/paper/b9663073f33a5cff52e7d50b6e7a296cdb9d519c" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "6394e6ce505ba1ea8d9afb76306e63235f3b23ce" }, "url": "https://www.semanticscholar.org/paper/6394e6ce505ba1ea8d9afb76306e63235f3b23ce" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "8cd56c46012a1bb005c555c208452af89618fca6" }, "url": "https://www.semanticscholar.org/paper/8cd56c46012a1bb005c555c208452af89618fca6" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "18939eadc9c4460c8385e0591cde214a1ead067b" }, "url": "https://www.semanticscholar.org/paper/18939eadc9c4460c8385e0591cde214a1ead067b" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "041fddf6110a43aa8acc3cd7e480e0f268d80abd" }, "url": "https://www.semanticscholar.org/paper/041fddf6110a43aa8acc3cd7e480e0f268d80abd" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "ac48b215a512be4d0e9d1fbce98695a11baab360" }, "url": "https://www.semanticscholar.org/paper/ac48b215a512be4d0e9d1fbce98695a11baab360" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "f89fe7d194c63a62b2cb76fac08ef525b483b964" }, "url": "https://www.semanticscholar.org/paper/f89fe7d194c63a62b2cb76fac08ef525b483b964" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "6ff909c6fe089fc8ebfc64eca0f0c3cc34ba277f" }, "url": "https://www.semanticscholar.org/paper/6ff909c6fe089fc8ebfc64eca0f0c3cc34ba277f" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "d891dc72cbd40ffaeefdc79f2e7afe1e530a23ad" }, "url": "https://www.semanticscholar.org/paper/d891dc72cbd40ffaeefdc79f2e7afe1e530a23ad" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "29bae9472203546847ec1352a604566d0f602728" }, "url": "https://www.semanticscholar.org/paper/29bae9472203546847ec1352a604566d0f602728" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "7038e1c5dc2ab26619674195976c0b5bd42aba77" }, "url": "https://www.semanticscholar.org/paper/7038e1c5dc2ab26619674195976c0b5bd42aba77" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "72e22d2bd4f0b5a9c51e09ffcab89a6b388b7da9" }, "url": "https://www.semanticscholar.org/paper/72e22d2bd4f0b5a9c51e09ffcab89a6b388b7da9" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "d8384f7ef288d2d5cb267128471c5427fc98b54b" }, "url": "https://www.semanticscholar.org/paper/d8384f7ef288d2d5cb267128471c5427fc98b54b" }, { "@type": "ScholarlyArticle", "identifier": { "@type": "PropertyValue", "propertyID": "SemanticScholar", "value": "6fb07b90b7fd2785ffec0da1069e75c53f7313c2" }, "url": "https://www.semanticscholar.org/paper/6fb07b90b7fd2785ffec0da1069e75c53f7313c2" } ], "additionalProperty": [ { "@type": "PropertyValue", "propertyID": "viabilityScore", "value": 3 }, { "@type": "PropertyValue", "propertyID": "researchDomain", "value": "Medical AI" } ] }, { "@type": "BreadcrumbList", "itemListElement": [ { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://sciencetostartup.com" }, { "@type": "ListItem", "position": 2, "name": "Medical AI", "item": "https://sciencetostartup.com/topics" }, { "@type": "ListItem", "position": 3, "name": "Diffusion-Based Feature Denoising and Using NNMF for Robust ", "item": "https://sciencetostartup.com/paper/diffusion-based-feature-denoising-and-using-nnmf-for-robust-brain-tumor-classification" } ] }, { "@type": "FAQPage", "mainEntity": [ { "@type": "Question", "name": "What products could be built from this research?", "acceptedAnswer": { "@type": "Answer", "text": "Now is the time because healthcare AI adoption is accelerating post-pandemic, but regulatory scrutiny on AI safety is increasing (FDA's AI/ML action plan), creating demand for robust solutions. The rise of adversarial ML research has exposed vulnerabilities in existing medical AI systems that need addressing before widespread clinical deployment." } }, { "@type": "Question", "name": "What are the practical use cases?", "acceptedAnswer": { "@type": "Answer", "text": "Integrating this robust classification framework into a cloud-based MRI analysis platform that hospitals use for preliminary tumor screening, where it automatically flags suspicious cases for radiologist review while maintaining high accuracy even if image data is corrupted or manipulated." } } ] } ] }

Competitive landscape

A robust framework for brain tumor classification using NNMF and lightweight CNNs to enhance adversarial robustness.

Segment

Medical AI

Adoption evidence

No public code link in the paper record yet

Commercial read

3.0/10 public viability

Direct

not classified

Adjacent

not classified

Substitute

not classified

Unknown

not classified

References(15)

The Past, Present and Future of the ELKH Cloud

2022Mihály Héder, Ernő Rigó et al.

Effect of data leakage in brain MRI classification using 2D convolutional neural networks

2021Ekin Yagis, S. Atnafu et al.

Development of partial least squares regression with discriminant analysis for software bug prediction

2020R. Ferenc, István Siket et al.

Feature Extraction Based on the Non-Negative Matrix Factorization of Convolutional Neural Networks for Monitoring Domestic Activity With Acoustic Signals

2020Seokjin Lee, Hee-Suk Pang

Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attacks

2020Francesco Croce, Matthias Hein

Non-Negative Matrix Factorization-Convolutional Neural Network (NMF-CNN) For Sound Event Detection

2020Teck Kai Chan, C. Chin et al.

Semi-NMF network for image classification

2019Haonan Huang, Zuyuan Yang et al.

Brain Tumor Detection Using Convolutional Neural Network

2019Tonmoy Hossain, Fairuz Shadmani Shishir et al.

Nonnegative Matrix Factorization

2019Ke-Lin Du, M. Swamy

A survey on deep learning in medical image analysis

2017G. Litjens, Thijs Kooi et al.

Intriguing properties of neural networks

2013Christian Szegedy, Wojciech Zaremba et al.

Semantic Segmentation with

2012Second-Order Pooling, Rui Caseiro et al.

An Introduction to Variable and Feature Selection

2003Isabelle M Guyon, A. Elisseeff

Algorithms for Non-negative Matrix Factorization

2000Daniel D. Lee, H. Sebastian et al.

Learning the parts of objects by non-negative matrix factorization

1999Daniel D. Lee, H. S. Seung

Diffusion-Based Feature Denoising and Using NNMF for Robust Brain Tumor Classification

Diffusion-Based Feature Denoising and Using NNMF for Robust Brain Tumor Classification

Claim map

Constellation map

Competitive landscape

Buzz

PDF

References(15)

Related Papers

Related Resources

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline

Claim map

Constellation map

Competitive landscape

Buzz

PDF

References(15)

Related Papers

Related Resources

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline