AI in Medical Imaging, Diagnostics & Pathology

Session Overview

Medical imaging stands at the epicenter of artificial intelligence adoption in clinical medicine. From radiology departments processing thousands of scans daily to pathology laboratories digitizing tissue slides at scale, the integration of AI into image-based diagnostics is no longer a future prospect — it is an accelerating clinical reality. This session brings together researchers, clinicians, computational scientists, and industry innovators to examine the full spectrum of AI-driven advances in medical imaging, diagnostic systems, and computational pathology, spanning foundational algorithmic developments through to real-world clinical deployment.

This session features a keynote lecture, four oral presentations, and a poster presentation segment covering the breadth of AI applications across all major imaging modalities, diagnostic specialties, and translational challenges in bringing imaging AI from research into practice.

Why This Session Matters Now

The convergence of large-scale annotated imaging datasets, transformer-based architectures, and unprecedented computational resources has triggered a fundamental shift in how diagnostic AI systems are designed, validated, and deployed. Models that once required thousands of labeled examples per task are giving way to foundation models capable of generalizing across imaging modalities with minimal supervision. Simultaneously, regulatory agencies across major markets are actively developing frameworks for AI-based medical devices, making clinical validation and responsible deployment more urgent than ever. For researchers, clinicians, and technology developers working in this space, 2027 represents a pivotal moment — the transition from demonstrating that AI can match human diagnostic performance to demonstrating that it reliably improves patient outcomes at scale.

Key Scientific & Technical Themes

Foundation Models, Self-Supervised Learning & the Future of Medical Vision AI

The emergence of large-scale foundation models pretrained on diverse imaging data is redefining the boundaries of diagnostic AI. Unlike task-specific models trained on narrow datasets, foundation models for medical vision demonstrate remarkable generalization across imaging modalities and clinical tasks with substantially reduced dependence on labeled training data. Self-supervised learning strategies are enabling representations that capture clinically meaningful features without exhaustive expert annotation, while multi-modal image fusion frameworks are integrating radiological, pathological, and clinical data streams within unified architectures. The deployment of such models on edge AI devices further extends their reach to resource-limited and point-of-care settings globally.

Deep Learning in Radiology, Cross-Sectional & Cardiac Imaging

Deep learning applied to CT and MRI interpretation is achieving performance levels that challenge specialist radiologists across a growing range of conditions, from pulmonary nodule detection and intracranial hemorrhage classification to abdominal organ segmentation. In cardiac imaging, AI-driven echocardiography analysis is enabling automated measurement of cardiac function with reproducibility that exceeds manual assessment. Breast imaging and mammographic screening represent one of the most clinically impactful applications, with large-scale trials now evaluating AI-assisted reading as an independent diagnostic tool. This theme covers the full pipeline from architecture design and training methodology through to prospective clinical evaluation and workflow integration.

AI in Digital Pathology, Histology & Tumor Analysis

Computational pathology is undergoing a transformation as whole-slide imaging matures and AI models demonstrate diagnostic accuracy across a widening range of histopathological tasks including tumor grading, subtype classification, margin assessment, and identification of molecular alteration signatures. Image segmentation and tumor delineation algorithms enable precise quantification of tumor cellularity, stromal composition, and immune infiltration patterns with direct prognostic significance. Radiomics and imaging biomarker extraction bridge the gap between macroscopic imaging and tissue-level biology, providing noninvasive surrogates for histopathological characteristics. This theme addresses computational methodologies, diagnostic validation standards, and the translational pathway from research algorithms to clinically deployed diagnostic support systems.

AI in Endoscopy, Ophthalmology & Specialty Imaging

Beyond radiology and pathology, AI is demonstrating transformative diagnostic impact across specialty imaging domains that rely on real-time human interpretation under time pressure. In gastrointestinal endoscopy, deep learning models for polyp detection and characterization are now deployed in clinical settings, with prospective trials confirming meaningful reductions in diagnostic miss rates. In ophthalmology, AI systems for retinal disease screening have achieved regulatory approval and are being deployed at population scale across multiple health systems. Dermatology represents a uniquely accessible AI imaging domain, with deep learning classifiers for skin lesion characterization achieving specialist-level sensitivity in controlled evaluations. This theme examines the computational challenges, dataset requirements, and clinical validation standards relevant to AI across these specialty imaging domains.

Clinical Deployment, Validation & Real-World Integration of Imaging AI

Demonstrating algorithmic performance on benchmark datasets is a necessary but insufficient condition for clinical adoption — models must maintain performance across distributional shifts introduced by different imaging protocols, patient populations, and institutional workflows. Prospective clinical validation, including randomized controlled trials and real-world studies, is increasingly demanded by regulators and health technology assessment bodies. Validation frameworks must address not only accuracy metrics but clinical utility — whether AI assistance meaningfully changes clinical decisions and improves patient outcomes. The evolving regulatory landscape introduces new obligations around transparency, post-market surveillance, and algorithmic accountability that every imaging AI developer must navigate. This theme brings together methodological, regulatory, and implementation science perspectives required for the pathway from validated algorithm to sustainably deployed clinical tool.

Research Landscape & Data Trends

Medical imaging AI is one of the most prolific areas of biomedical research, with indexed publications in this domain growing at a rate that substantially exceeds the broader AI in healthcare literature. The field is characterized by a productive tension between methodological innovation driven by computer science communities and clinical validation driven by specialist clinicians with access to real-world patient data. The current research frontier is defined by three converging priorities — the development of foundation models capable of generalizing across imaging tasks, the design of prospective clinical trials providing the evidentiary standard required for regulatory decisions, and the creation of federated learning frameworks enabling model training across institutional boundaries without centralizing sensitive patient data. By 2027, the literature is expected to be dominated by real-world deployment studies, health economic evaluations, and post-market surveillance analyses as first-generation imaging AI products mature and foundation model-based systems enter clinical evaluation.

Who Should Attend

  • Radiologists, neuroradiologists, and interventional radiologists integrating AI tools into clinical reporting workflows
  • Pathologists and computational pathology researchers developing and validating digital diagnostic systems
  • Gastroenterologists, pulmonologists, and endoscopists working with AI-assisted procedural guidance
  • Ophthalmologists and optometrists involved in AI-based screening program design and evaluation
  • Dermatologists and dermatopathologists evaluating AI classification systems for clinical use
  • Medical physicists and imaging scientists developing acquisition and reconstruction algorithms
  • Computer scientists and machine learning researchers working on vision architectures for medical applications
  • Biomedical engineers developing AI-enabled imaging devices and edge deployment systems
  • Regulatory scientists and health technology assessment professionals evaluating AI medical devices
  • Industry researchers and technology developers bringing imaging AI products through clinical validation and commercialization

Session Perspective

Medical imaging AI has moved decisively past the proof-of-concept phase. The questions that now define the field are not whether AI can perform diagnostic tasks with high accuracy, but whether it can do so reliably across the full diversity of clinical environments, whether its deployment genuinely improves outcomes that matter to patients, and whether regulatory and reimbursement infrastructure can keep pace with technological development. This session provides a platform for rigorous, critical, and forward-looking examination of these questions — bringing together the methodological depth of computational research with the clinical authority of specialist practice. Researchers, clinicians, and innovators shaping the next generation of imaging AI are invited to contribute their work to this session.

If your research aligns with this session, we invite you to submit an abstract for consideration.

Submit Your Abstract