Radiomics and Machine Learning in Medical Imaging
New papers in radiomics and machine learning in medical imaging, the code and data they release, and the open problems their authors admit. Rebuilt every day from OpenAlex and arXiv.
New papers
Influence of body composition on survival in breast cancer patients with bone metastasis: a clinicopathologic-radiomic web-based nomogram development and validation
Background Bone metastasis (BM) reduces survival in breast cancer (BC) patients.
TRANSPERINEAL MRI–ULTRASOUND FUSION-GUIDED PROSTATE BIOPSY USING THE KOELIS TRINITY SYSTEM: DIAGNOSTIC OUTCOMES AND INITIAL EXPERIENCE IN A SIX-PATIENT CASE SERIES
Background: Transperineal magnetic resonance imaging–ultrasound (MRI–US) fusion-guided prostate biopsy has emerged as an important diagnostic approach for patients with suspected prostate cancer.
What specific MRI features affect management discordance in prostate cancer diagnosis?
Specific MRI features such as variability in diffusion-weighted, dynamic contrast-enhanced imaging, and inconsistencies in protocol standardization lead to management discordance in prostate cancer diagnosis.
Can Graph Reasoning Improve Contextual Understanding in Histopathology?
Objectives: In this work, we rigorously evaluate whether augmenting a ResNet-18 backbone with a Graph Convolutional Network (GCN) that performs message passing over a feature-space k-nearest-neighbor graph, built from patch feature similarity rather than true spatial coordinates, improves…
HCT-Mammo: A Compact Hybrid CNN-Transformer for Benign–Malignant Mammogram Classification with Uncertainty and Cross-Dataset Evaluation
The difficulty in comparing public-dataset results for benign–malignant mammogram classification is due to patient overlap, lesion crops, and a lack of independent testing.
Development of a Deep Learning Model Integrating Computed Tomography Images and Blood Data for the Diagnosis of Acute Cholecystitis
Abstract Background In this study, we aimed to develop and evaluate an artificial intelligence-based diagnostic model for the diagnosis of acute cholecystitis (AC) using non-contrast CT images and clinical data.
What causes discordance between PI-RADS/Likert score 4-5 MRI findings and histopathology results in prostate cancer diagnosis, and which MRI features are associated with higher discordance rates?
Discordance between PI-RADS/Likert score 4-5 MRI findings and histopathology results can be attributed to interobserver variability, biopsy sampling errors, and imaging quality.
Development and external validation of a predictive model for epithelial ovarian tumors based on MRI features and serum tumor markers
Ovarian cancer remains a major threat to women’s health, and early-stage disease is often asymptomatic, making timely diagnosis difficult.
How does PSA density affect discordance between MRI and biopsy results in the diagnosis of prostate cancer?
PSA density is potentially useful in guiding biopsy decisions and could indirectly reduce the discordance between MRI and biopsy findings in prostate cancer diagnosis by optimizing patient selection for biopsy.
A CT Lesion Foundation Model for Non‐Invasive Assessment of Hepatocellular Carcinoma Aggressiveness and Prediction of Recurrence Risk
Hepatocellular carcinoma (HCC) remains a leading cause of cancer-related mortality worldwide, with high recurrence rates following curative resection.
A Multi-Criteria Benchmarking Framework for Deep Learning Models in Medical Image Classification Toward Reliable Cancer Diagnosis
يُعدّ الكشف المبكر والدقيق عن الأورام السرطانية أمرًا بالغ الأهمية لتحسين معدلات البقاء على قيد الحياة والحدّ من المضاعفات المرتبطة بالعلاج، لا سيما في ظلّ التعقيد البصري العالي للصور الطبية، حيث تكون الفروقات بين الأورام الحميدة والخبيثة دقيقة ومتداخلة في كثير من الأحيان. في هذا السياق، برزت…
The role of radiomics and deep learning in glioma MRI: advances, challenges, and LMIC-tailored opportunities for precision medicine
Glioma is the most common primary malignant central nervous system tumour in adults, with limited access to advanced molecular diagnostics in low- and middle-income countries, like Pakistan.
Deep learning integration of ultrasound perfusion and stiffness maps predicts tumor response to treatment
The complex tumor microenvironment and tumor heterogeneity make treatment responses highly variable, driving the shift toward biomarker-based therapy prediction.
What is the frequency of discordance between histopathology results and PI-RADS/Likert score 4-5 MRI findings in prostate cancer patients?
The frequency of discordance between PI-RADS/Likert score 4-5 MRI findings and histopathology results in prostate cancer patients is indirectly suggested to be around 11%, with some studies indicating broader discordance in biopsy results.
Artificial Intelligence in Breast Cancer Diagnosis, Treatment, and Perioperative Management: A Systematic Review of Current Applications and Future Perspectives.
Background: Breast cancer is the most frequently diagnosed malignancy in women globally, accounting for approximately 2.30 million new cases and 666,103 deaths in 2022.
Non‐Invasive Assessment of Perineural Invasion Density: A Novel Marker for Prognosis and Treatment Stratification in Prostate Cancer
We established and validated perineural invasion density (PNID) as a superior prognostic marker and developed an MRI-based AI model for its preoperative prediction.
Digital pathological classification of peripheral neuroblastic tumors across multi-continent cohorts
Peripheral neuroblastic tumors (pNTs) are embryonal tumors of neural crest origin with highly variable clinical behaviors, ranging from spontaneous regression to aggressive and fatal disease.
Multimodal ultrasound radiomics of carotid plaques is associated with ischemic heart disease: additional discriminative information from vector flow imaging and shear wave elastography
Ischemic heart disease (IHD) is associated with coronary atherosclerosis, but related atherosclerotic phenotypes may also be reflected in carotid plaques.
AI-HistoPS: decoding proteomic heterogeneity with subtype resolution from histopathology in liver metastases
Liver metastasis is a lethal manifestation of diverse malignancies, but its molecular heterogeneity is not readily captured by routine histopathology.
Toponet: a topology-guided dual-branch framework for medical image classification
To address this limitation, we propose TopoNet, a topology-guided dual-branch framework that integrates multiscale topological structures into volumetric representation learning.
Released code and data
Toponet: a topology-guided dual-branch framework for medical image classification
The code is available at https://github.com/hbqdcq/TopoNet.
MOSAIC-Flow: Orthogonal-Axis and Status-Conditioned Fusion for Missing-Sequence Completion in Brain Tumor MRI
Code is available at https://github.com/CUHK-AIM-Group/MOSAIC-FLOW.
DINOv2 for PET: A Pre-Training Domain Benchmark Across Classification, Segmentation, and Prognosis
Code is made available at https://github.com/nuclivision/pet-dinov2-2d/.
Segment Anything Model Based Context Guided Mamba Unet for Medical Image Segmentation
Our code will be released upon acceptance.
RepConTUR: A Report-Conditioned Volumetric Multimodal Model for Abdominal Tumor Segmentation
Code is available at https://github.com/Oncosoft-Research/RepConTUR.
Radiology Guideline-derived Decision Graph for Efficient Clinical Decision Support
The source code will be available.
Development and Validation of an Automated Image-Based Method for Prognostic Stratification in Lung Adenocarcinoma using Immunohistochemical Markers
The complete pipeline was developed as an open-source Fiji/ImageJ tool and is publicly available.
From Slices to Volumes: Comparing 2D, 2.5D, and 3D Residual Networks for Pulmonary Nodule Malignancy-Risk Classification
Code and experimental results are available at https://github.com/CT-DeepLearning-Team/CT-2D-2.5D-3D-DeepLearning.
Meta-CLIP: Metadata-Conditioned Contrastive Learning for Zero-Shot Disease Inference
Code will be made publicly available upon publication.
Diagnostic Loss: Asymmetric Supervision for Radiology Report Generation
The code for DLoss is available at https://github.com/gmum/DLoss.
Open problems authors admit
Automated methods for assessing structural lung disease in patients with cystic fibrosis and primary ciliary dyskinesia: a scoping review
We synthesized findings from 50 peer-reviewed articles to assess trends in automated CT scoring techniques for assessing structural lung changes in patients with CF and PCD, identify the existing challenge to develop clinical grade systems, and propose future research directions to advance the existing research towards mature technology.
Research Progress of MRI-based Radiomics and Deep Learning in Lymphovascular Space Invasion of Endometrial Carcinoma
This article systematically reviews the research progress of MRI-based radiomics and deep learning in the field of LVSI prediction in endometrial cancer, elaborates on the standard radiomics workflow, traditional radiomics models, deep learning models, and the diagnostic performance of combined clinical-imaging prediction models, summarizes key issues in current research such as limited sample size, cross-center domain shift, insufficient reproducibility, and lack of prospective validation, and provides an outlook on future directions including clinical translation, multi-omics integration, and explainable artificial intelligence.
Identification of a novel lncRNA prognostic signature for lung adenocarcinoma via ML-based feature selection and artificial neural networks
This study introduces a novel lncRNA signature with potential applications in personalized prognosis and therapeutic development for lung adenocarcinoma, supporting future research on discovering lncRNA biomarkers.
Development and External Validation of an Updated Nomogram Including Both MRI and Local PSMA PET/CT Information for the Prediction of Side-specific Extraprostatic Extension in Patients Treated with Radical Prostatectomy
Limitations include a relatively small validation cohort.
Cancer Detection Using AI-based Techniques
This chapter addresses challenges and future directions, including ethical and regulatory concerns and AI integration with metaverse and exascale computation to enhance cancer detection.
A multimodal attention-augmented deep learning framework for oral cancer diagnosis, prognosis and personalised treatment recommendation
The framework is developed through a six-phase methodology; Phase 6 (prospective clinical validation) is defined as future work and has not been carried out, and all results reported here are retrospective.
The Role of Artificial Intelligence in Cancer Diagnostic Imaging Across Six Domains: An Extensive Umbrella Review (Preprint)
However, the evidence base is dispersed across multiple cancer types and imaging modalities, necessitating an integrated synthesis to assess AI’s true clinical readiness and the implementation challenges it faces OBJECTIVE This umbrella review aimed to comprehensively synthesize evidence from published systematic reviews and meta-analyses on the role of AI in cancer diagnostic imaging, identifying demonstrated benefits, limitations, and future research priorities METHODS A comprehensive search of PubMed, Embase, Scopus, Web of Science, and Cochrane Library identified systematic reviews and meta-analyses published between January 2020 and September 2025.
MS-RGT-LF-XAI: Multi-Scale Radiomic–Graph–Transformer Late-Fusion Network with Explainable Artificial Intelligence for Breast Cancer Histopathology Classification
Future work will therefore include prospective multicentre studies with pathologist reader evaluations.
Recent advances in multimodal deep learning for breast cancer diagnosis using data fusion vision language models and foundation models
Future research should focus on robust multimodal foundation models, explainable and trustworthy artificial intelligence, federated and privacy-preserving learning, standardized external validation, and prospective clinical evaluation.
Radiomics-GenAI: a novel model for laryngeal lesion classification
Future work will focus on external validation and conditional generative models tailored to lesion subtypes and imaging acquisition variability.