- Campbell Arnold
- Apr 15
- 3 min read
Welcome to RadAccess: Impressions—your quick-read companion to the main RadAccess newsletter. Like a radiology report's impressions section, we only deliver the essential information to respect your time. For more details, you can always turn to the full RadAccess newsletter.
Bitcoin launches portable X-ray to the moon! Well, space at least…
The FRAM2 mission, a privately funded polar-orbit human spaceflight led by crypto billionaire Chun Wang and operated by SpaceX, made history by capturing the first human X-ray in space. The SpaceXray project was enabled by portable technologies from KA Imaging and MinXray, using the Reveal 35C detector and IMPACT X-ray system to successfully produce diagnostic-quality images in microgravity. The first interstellar x-ray image—a hand with a ring—echoed Wilhelm Roentgen’s original 1895 X-ray, marking both a tribute and a technological milestone. This breakthrough not only supports astronaut health monitoring for longer missions, but also highlights the potential of portable imaging to bring life-saving radiology to remote, underserved, or disaster-stricken regions on Earth.
MedSAM2 slashes medical image segmentation costs by 85%
Recently, a team led by Jun Ma, Zongxin Yang, and Bo Wang released MedSAM2, a new foundation model for 3D medical image and video segmentation. This latest article builds on their prior work with MedSAM by fine-tuning Meta AI's latest segmentation algorithm, Segment Anything Model 2.1. MedSAM2 was fine-tuned on over half a million 3D image–mask pairs spanning modalities like CT, MRI, PET, ultrasound, and endoscopy. This broad dataset enables state-of-the-art performance across 10+ segmentation benchmarks and strong generalization to unseen organs and pathologies. Notably, the team integrated the model into a human-in-the-loop annotation system—slashing annotation costs by over 85% while generating massive new datasets, including 5,000 CT lesions and 250K+ echo frames. MedSAM2 also offers out-of-the-box deployment via 3D Slicer and Google Colab, and the code and model weights are open access.
India Scheduled to Deploy Homegrown MRI in October
India is preparing to launch its first indigenous MRI scanner, marking a major step toward more affordable and accessible imaging. Developed by the government-run R&D institute SAMEER, the 1.5T Indigenous Magnetic Resonance Imaging (iMRI) system is projected to cost around $300K, which is significantly less than current import options. The first prototype is slated for installation and testing at AIIMS Delhi in October. The effort is part of a broader initiative to reduce reliance on foreign medical equipment, which also includes the development of a low-cost Linear Accelerator (LINAC) for cancer treatment. If successful, these technologies could greatly expand access to radiology and oncology services across India, and offer affordable solutions for other countries facing similar healthcare challenges.
EMVision launches clinical trial on portable brain scanner
Australian company EMVision has launched a clinical trial for their portable brain imaging device, emu, which uses electromagnetic imaging (EMI) to provide rapid, non-invasive assessments of brain tissue. Unlike MRI, EMI employs low-power radiofrequency waves and requires minimal infrastructure, making it especially well-suited for stroke detection in emergency or resource-limited settings. The trial, which aims to support an FDA De Novo clearance, has begun at The Royal Melbourne Hospital in Australia and the University of Texas Health Science Center in Houston. EMVision plans to enroll 300 suspected stroke patients across six AU and US sites and aims to demonstrate hemorrhage detection accuracy above 80%—a milestone that could introduce a novel point-of-care tool for emergency medicine.
MedSAM Public Imaging Datasets List
Looking for open-source segmentation data? The MedSAM team has a curated list of over 300K public medical image-segmentation pairs, covering ~20 modalities and 30 anatomical regions. The spreadsheet includes key details like dataset names, modalities, anatomy, segmentation targets, class numbers, image counts, licensing info, and links to the datasets and related papers. It's an excellent resource for anyone working on medical image segmentation.
Feedback
We’re eager to hear your thoughts as we continue to refine and improve RadAccess. Is there an article you expected to see but didn’t? Have suggestions for making the newsletter even better? Let us know! Reach out via email, LinkedIn, or X—we’d love to hear from you.
References
W. Roninson-Smith. Meet the Fram2 crew: A cryptocurrency entrepreneur, a cinematographer, a robotics engineer and an Arctic explorer. Space Flight Now. 2025.
F Fraga. KA Imaging to Support the First Medical X-ray in Space. KA Imaging Press Release. 2025.
Ma, Jun, et al. "MedSAM2: Segment Anything in 3D Medical Images and Videos." arXiv arXiv:2504.03600 (2025).
AIIMS to install first indigenous MRI scanner for clinical evaluation. ETHealthWorld. 2025.
Public Medical Image Datasets for Segmentation Foundation Models. Github.
Disclaimer: There are no paid sponsors of this content. The opinions expressed are solely those of the newsletter authors, and do not necessarily reflect those of referenced works or companies.
