Weekly AI digest
Radiology & medical imaging AI · week of 28 June to 4 July 2026
8 peer-reviewed papers · 3 industry and regulatory items · conference and KOL highlights. Adapted from my weekly intelligence report. Full report: FR (PDF) · EN (PDF).
Three things that mattered
Gadolinium-free liver MRI matches full contrast protocol
Zhang Y et al. (Radiology: Artificial Intelligence) validated Li-DiffNet across 1,008 patients at 4 centres: the deep learning single breath-hold abbreviated protocol achieves sensitivity 77.9-88.7% and specificity 91.6-93.1%, non-inferior to conventional complete MRI, eliminating gadolinium risk for cirrhotic patients with renal impairment.Skeleton foundation model beats radiologists on bone metastases
Zhao H et al. (Nature Biomedical Engineering) pretrained BoneCoT on 29.3 million CT images from 30,267 patients across 12 skeletal sites, then fine-tuned it over 26 clinical tasks with clinician-derived chain-of-thought reasoning; across 10-hospital multicentre cohorts it outperformed state-of-the-art methods by 20% AUROC and improved primary versus metastatic lesion discrimination by 40% AUROC, surpassing experienced radiologists.UK AI adoption 75% but workload unchanged
The RCR 2025 Workforce Census found 75% of UK radiology departments using AI clinically (+6% vs 2024) yet overall radiologist workload has not decreased, with report drafting delivering the greatest impact (24% of departments reporting reduced workload) while only 13% use AI for administrative tasks like staff scheduling.
Peer-reviewed
A multimodal feature disentanglement model for lymphadenopathy diagnosis based on BUS and CDFI ultrasound videos: a retrospective, prospective, multicenter study
European Radiology, 1 July 2026Comparison of clinical-radiological and radiomics features for predicting pulmonary nodule malignancy in a multicenter study of mixed clinical and surveillance populations
European Radiology, 1 July 2026Artificial intelligence in the imaging diagnosis of gallbladder and bile duct stones: a systematic review
Abdominal Radiology, 29 June 2026BoneCoT: multicentre validation of a whole-body skeleton foundation model for bone metastases guided by clinician-derived chain of thought
Nature Biomedical Engineering, 2 July 2026Development and validation of a deep learning-enabled single breath-hold abbreviated MRI protocol for hepatocellular carcinoma diagnosis
Radiology: Artificial Intelligence, 1 July 2026A confidence-based artificial intelligence pathology model for diagnosis of intrahepatic cholangiocarcinoma
Annals of Oncology, 1 July 2026Real-time automatic guidance during shoulder ultrasound scanning with artificial intelligence: reducing operator dependency in rotator cuff assessment
Academic Radiology, 1 July 2026Large language model and Gd-EOB-DTPA-enhanced MRI-based risk stratification system for postoperative hepatocellular carcinoma: a multicenter study
European Radiology, 1 July 2026
Industry & regulation
Deepnoid M4CXR receives Korea first Class III clearance for generative AI radiology reporting
Korea MFDS granted M4CXR the highest-risk Class III designation for free-text report generation across 41 chest disease categories, trained on more than 10 million image and report pairs, with FDA submission in preparation via US academic partnerships.RADIN Health expands cloud-native AI radiology platform to eight UHS Nevada hospitals
RADIN Health expanded its PACS, Dictation AI, and RADIN Select platform to four additional UHS Nevada sites on serverless AWS, bringing total Nevada presence to eight hospitals and reflecting accelerating cloud-PACS plus AI bundle adoption over legacy systems.Lumitron receives USD 50 million from Hancock Prospecting for HyperVIEW cancer imaging platform
Hancock Prospecting invested USD 50 million with an option to USD 100 million in Lumitron to support commercial scale-up of HyperVIEW for advanced oncology imaging and VHEE FLASH radiotherapy, with three systems destined for Australia as a world-first deployment.
Conferences & voices
The Royal College of Radiologists held its 2nd Annual Global AI Conference (QEII Centre, London, 29-30 June), themed on safe and practical implementation, with a keynote by Prof. Curtis Langlotz (Stanford AIMI, RSNA President) presenting a task-based quantitative projection of AI's effects on the radiology workforce and asking whether AI alone can surpass the AI-plus-radiologist combination. The simultaneously released RCR 2025 Workforce Census found 75% of UK departments using AI clinically yet no reduction in overall workload, and recommended prioritising administrative AI. On the vendor side, HOPPR launched its Presto Agent (24 June), bringing AI draft reporting into existing PowerScribe 360 and PowerScribe One workflows without platform migration, with HOPPR CSO and CMIO Woojin Kim framing integration rather than raw accuracy as the real adoption bottleneck. Meanwhile the week's major imaging-vendor finance story, Nano-X Imaging's going-concern warning (Radiology Business, 29 June), drew scrutiny from vendor-critical voices such as Amine Korchi.
Source articles are indexed in PubMed with verified DOIs. Manuscript-stage work is excluded. The full weekly report is produced in French.
Get the weekly digest
One email a week on what matters in radiology and imaging AI. Researched by a radiologist and imaging informatics professional. Free, no vendor sponsorship.