MediView's XR90 Guided a Needle 10 Centimeters Into a Liver on a Headset Microsoft Stopped Building

MediView XR announced on August 4 that Cleveland Clinic Weston Hospital had completed the first intraprocedural use of its XR90 holographic navigation system in the state of Florida, and the case is worth reading closely for what it shows about where augmented reality has actually landed inside a hospital. The procedure was a percutaneous liver mass biopsy. The target was a lesion roughly 10 centimeters deep, sitting between two branches of the hepatic vein, which is the sort of position that makes an interventional radiologist work slowly and take more passes than anyone wants. Dr. Kevin Stadtlander performed it without X-ray guidance.

Stadtlander's description in Cleveland Clinic's own announcement on August 20 was matter-of-fact: "Extended reality allows us to combine two different imaging modalities, CT and ultrasound, in real time, to do a procedure such as a biopsy." In the company's release he was more expansive, calling it a modality that "can be a game changer because we're creating a holographic 3D image overlaid into the patient," and adding that "we're seeing into the patient's body 3-dimensionally without any X-rays." MediView chief executive Mina Fahim framed the case as proof of the original thesis: "This case represents exactly what we set out to build. Fusing CT and ultrasound into a single 3D augmented reality view."

What the system does

XR90 registers a preoperative CT scan to live, tracked ultrasound and projects the combined result into the operator's field of view through a see-through headset. MediView calls the live ultrasound overlay the Ultrasound Flashlight, because the tracked probe paints its image into the patient's anatomy the way a beam would rather than parking it on a monitor across the room. A feature the company calls the Holographic Light Ray displays the tracked path of the instrument, supporting in-plane and out-of-plane trajectory planning, and a configurable heads-up display carries the rest of the procedural information. Anatomic structures are shown or hidden by voice command and hand gesture.

The value proposition is a head-up one. An interventional radiologist guiding a needle by ultrasound normally works with eyes on a screen and hands on a patient, reconciling a 2D slice with a 3D body by experience. Removing that glance is the entire pitch, and it is the same pitch enterprise XR has been making to procedural teams for a decade. What separates this case from a conference demo is the paperwork behind it.

The clearance is specific and it is on the record

FDA's 510(k) database lists the clearance as K223125, device XR90 (XR90-SYS), applicant Mediview XR, Inc., product code LLZ, with a decision date of July 13, 2023. The company announced it on July 20 that year and described it as the first 510(k) granted to an augmented reality device combining live imaging with 3D XR visualization across both pre-operative and intra-operative indications. The cleared indication itself is narrow and worth quoting rather than paraphrasing: the system is intended to be used adjunctively for minimally invasive ultrasound and CT guided needle-based procedures for soft tissue and bone. Adjunctively is doing real work in that sentence. XR90 is a visualization aid attached to a procedure a physician is already qualified to perform, not a substitute for any part of it.

The headset underneath

Here is the part that should interest anyone signing a purchase order rather than reading a case report. When XR90 cleared in 2023, it ran on Microsoft's HoloLens 2. Microsoft ended HoloLens 2 production in October 2024 and has committed only to critical security and software regression updates through December 31, 2027. MediView's current XR90 product page describes the display as a see-through headset and names no manufacturer at all. The company has not publicly announced a successor platform.

That is not evidence of a problem and should not be read as one. Vendors routinely qualify replacement components without announcing it, and MediView has continued to expand deployments and run a multicenter validation study for soft tissue biopsy while the hardware question sat in the background. But the timeline is not ambiguous. A hospital buying XR90 today is buying a workflow whose display vendor has published an end-of-support date, and FDA requires a new premarket submission when a change could significantly affect a device's safety or effectiveness. Whether swapping the headset crosses that line is a determination the manufacturer makes first and the agency reviews after, which means it is not something a buyer can simply look up.

The pattern this fits

Enterprise XR keeps arriving at the same structural problem. The clinical result is real, the procurement is real, and the hardware layer underneath belongs to a company whose commitment to the category is decided somewhere else entirely. VR.org has covered the retreat of large platform vendors from enterprise XR, Magic Leap's pivot away from finished headsets toward waveguide supply, and a surgical study crediting Vision Pro with a 19 percent time improvement, all within four months. In each case the software argument is stronger than the hardware guarantee.

So the useful question for a 2026 procurement is not what the hologram does in the demo, because every one of these systems demonstrates well. It is what display the system will be running in 2028, whether the vendor has qualified that path, and what the regulatory consequence of the swap turns out to be. A needle placed 10 centimeters into a liver between two hepatic vein branches is a genuinely impressive result. It also has a support calendar attached to it, and the calendar is public.