Life & medical sciences · OECD FoS 3.2 · China MoE 100207 影像医学与核医学
Medical imaging & pathology
医学影像与病理
L3Systemic transformation → Steady Score 2.80 · Last reviewed 2026-10-21 · Next review 2026-12-20
BIGGEST BREAKTHROUGH 2026-01 Field leapAI · major
MASAI final results: AI-supported screening non-inferior on interval cancers, sensitivity 80.5% vs 73.8% (The Lancet) ↗
RCT of 105,934 women shows non-inferiority on the hard endpoint of interval cancers
Why this level
Imaging is medicine's most AI-penetrated branch: radiology dominates FDA's AI device list; the MASAI RCT showed AI-supported mammography safely cut reading workload by ~44% (Lancet Oncology 2023).
S = 0.30·method + 0.35·discovery + 0.25·adoption + 0.10·education (education scores 0 when there is no evidence; if empty, the other weights are renormalised)
Sources
- fda.gov/medical-devices/artificial-intelligence-enabled-medical-devices/list-artificial-intelligence-enabled-medical-devices
- thelancet.com/journals/lanonc/article/PIIS1470-2045(23)00298-X/fulltext
Composite 2.5–3.49 (or ≥3.5 with any of method/discovery/adoption below 3): AI is standard infrastructure with multiple independent breakthroughs. All levels →
Research directions
Screening & triagePathology foundation modelsReport generation
Papers
- Manual, digital, and AI tumour-infiltrating lymphocyte scoring: a secondary analysis of the APHINITY randomised trial
AI: AI pathology scoring benchmarked against manual scoring in a large RCT cohort; Field: trial-level evidence for standardising TIL biomarkers
2026-09 · The Lancet Oncology
-
Field leap
Interval cancer, sensitivity, and specificity comparing AI-supported mammography screening with standard double reading without AI in the MASAI study: a randomised, controlled, non-inferiority, single-blinded, population-based, screening-accuracy trialRCT of 105,934 women shows non-inferiority on the hard endpoint of interval cancers
2026-01 · The Lancet
- Towards a general-purpose foundation model for computational pathology
Pathology foundation model beats prior methods on 30+ tasks
2024-03-19 · Nature Medicine
- International evaluation of an AI system for breast cancer screening
Fewer false positives and negatives than radiologists across two countries
2020-01-01 · Nature
- Dermatologist-level classification of skin cancer with deep neural networks
Landmark paper that launched the medical-imaging AI wave
2017-01-25 · Nature
- Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs
Among the first to show specialist-level AI reading in a top medical journal
2016-12-13 · JAMA
- Improved three component IVIM fitting using physics-informed neural network with multimodal spatial prior
A physics-informed network using anatomical priors and cross-attention mitigates parameter degeneracy in three-component IVIM fitting.
2026-10-20 · arXiv · Nhat Hoang, Abrar Faiyaz, Emmanuel A. Mensah, Md Nasir Uddin
- CAPS: Clinically-Verifiable Anatomy-Prioritized Synthesis for Missing MRI Imputation in Brain Tumor Segmentation
Anatomy-prioritised synthesis of missing MRI modalities for brain-tumour segmentation (mean Dice 82.38% on BraTS).
2026-10-20 · arXiv · Cecilia Diana-Albelda, Arthur Longuefosse, Mario Rubén Paz Campos, Álvaro García-Martín
- Scaling Vision Foundation Models for Toxicologic Pathology
TOX-FM, trained on 1M+ toxicology whole-slide images across six species, evaluated on 27 tasks.
2026-10-19 · arXiv · Jessica Loo, Theodoros Pissas, Byunghak Kang, Vincent A. Ella
- Evaluating the Interplay and Impact of Quality and Technical Variations on Performance of AI Models in Multi-Institutional Prostate MRI Datasets
568 prostate MRIs from 7 sites show visually high-quality scans do not guarantee consistent AI performance.
2026-10-19 · medRxiv · Azamat, S., Singh, A., Midya, A., Zhang, Z.
- LUMINR: Intensity-conditioned Meta-learned INRs for Accelerated Medical Image Registration
Locally conditioned meta-learned INRs accelerate deformable medical image registration.
2026-10-16 · arXiv · Damian Rynczak, Marek Wodzinski
- GliomaINK: A Tumor and Anatomy Conditioned Flow Matching Framework for 3D MRI Sequence Synthesis in Glioma Patients
Anatomy-guided conditional flow matching synthesises missing 3D MRI sequences in glioma while preserving tumour features.
2026-10-16 · arXiv · Moinak Bhattacharya, Annie Singh, Mohammad Chaudhry, Prateek Prasanna
- Implicit Neural Models for Direct Multi-Shell Diffusion MRI Signal Estimation
Implicit neural representation of full diffusion-MRI signals for super-resolution and q-space interpolation.
2026-10-16 · arXiv · Sanna Persson, Fabian Leander Sinzinger, Christoffer Olsson, Rodrigo Moreno
- Multimodal profiling for prediction of primary resistance to anti-PD-(L)1 therapy in advanced non-small-cell lung cancer: the prospective PIONeeR biomarkers study
A prospective multicentre study building multimodal models to predict primary resistance to anti-PD-(L)1 therapy in advanced NSCLC.
2026-10-16 · medRxiv · Barlesi, F., Monville, F., Vely, F., Greillier, L.
- From arXiv, bioRxiv, medRxiv and OpenAlex (last ~30 days); summaries written by AIMap from the abstract.
Milestones 9· 1 leaps
Big glowing nodes are leaps: AI leap cool · Field leap warm · double ring = both
-
Model
Sophont launches Holon-S1, a pan-cancer pathology model predicting spatial gene expression from H&E (research use)AI: vendor reports gains across 16 organs; Field: no clinical validation yet
PR NewswireVendor claim -
Clinical
Manual, digital and AI TIL scoring: secondary analysis of the APHINITY RCT (Lancet Oncology)AI: AI pathology scoring benchmarked against manual scoring in a large RCT cohort; Field: trial-level evidence for standardising TIL biomarkers
The Lancet OncologyTo be verified -
ClinicalField leap
MASAI final results: AI-supported screening non-inferior on interval cancers, sensitivity 80.5% vs 73.8% (The Lancet)RCT of 105,934 women shows non-inferiority on the hard endpoint of interval cancers
The LancetTo be verified -
Paper
UNI: a general-purpose self-supervised model for computational pathology (Nature Medicine)Pathology foundation model beats prior methods on 30+ tasks
Nature Medicine -
Clinical
MASAI RCT: AI-supported mammography screening (Lancet Oncology)Interim RCT: 44% less reading workload, no drop in detection
Lancet Oncology -
Paper
International evaluation of an AI system for breast cancer screening (Nature)Fewer false positives and negatives than radiologists across two countries
Nature -
Paper
Dermatologist-level classification of skin cancer (Nature)Landmark paper that launched the medical-imaging AI wave
Nature -
Paper
Deep learning for diabetic retinopathy detection at ophthalmologist level (JAMA)Among the first to show specialist-level AI reading in a top medical journal
JAMA
No milestones match this filter.
More papers & repos 6
Latest news 12
- What Language Adds, and at What Cost: Four Controlled Probes of Pathology Vision-Language ModelsarXiv q-bioenPaperPreprint
- Radiology department uses AI to identify lung cancer screening patients in EHRRadiology Businessen
- A 3-step buyer’s guide for evaluating radiology AI solutionsRadiology Businessen
- Agilent to Acquire Proscia for Up to $270 Mn to Expand AI-Enabled Digital PathologyDigital Health Newsen
- AI matches radiologists at prostate cancer detection via MRIRadiology Businessen
- Digital Pathology in India: Infrastructure Readiness & the Next Wave of AdoptionDigital Health Newsen