From handheld to hands-free: Rise of robotic cardiac ultrasound

During a structural heart procedure, the operating room leaves little room for error, and every moment is important. For decades, interventional cardiologists have depended on transesophageal echocardiography (TEE), an imaging technique in which a specialised ultrasound probe is guided through a patient's esophagus to capture live images of the heart. These images allow physicians to monitor the heart's valves and chambers in real time while navigating catheters through delicate tissue.

However, performing TEE can be physically demanding for clinicians. Keeping the images clear often involves hours of precise probe handling and repeated adjustments, sometimes in uncomfortable positions. The procedure requires considerable skill and can put sustained physical strain on the clinicians responsible for obtaining the images.

A growing effort in Silicon Valley aims to shift much of that physical burden from human hands to robotic assistance. One of the companies advancing this approach is Laza Medical, a Campbell, California medical device company established within Shifamed LLC, the medical technology innovation hub founded by serial entrepreneur Amr Salahieh. Laza is developing an AI-driven robotic ultrasound platform designed to help automate aspects of cardiac imaging navigation. The company was backed by a Series A financing round that closed in December 2023, with participation from GE HealthCare and The Capital Partnership.

Bringing automation into the cardiac suite, however, is not as simple as teaching software to recognize what it sees. Modern AI can be excellent at interpreting images, but medicine demands something more than a likely answer. When a system is guiding physical motion close to living tissue, it must operate within strict limits every time.

“Safety in robotics cannot simply be treated as a statistical accuracy score,” said Puja Chaudhury, a robotics software engineer at Laza Medical. Chaudhury joined the company to help connect advanced imaging intelligence with the kind of real-time control required for dependable physical movement. In her view, AI only becomes clinically meaningful when it is paired with careful constraints, constant monitoring, and predictable safeguards.

In September 2025, Laza reached a milestone that reflected this step-by-step approach. The company completed its first human clinical cases evaluating a real-time software model of the heart. Conducted alongside live cardiac procedures, these evaluations used live imaging data to dynamically represent anatomy in software, allowing the team to validate the system’s ability to interpret what it was seeing before fully coupling those insights to robotic manipulation.

At Laza’s Campbell headquarters, turning imaging into controlled movement is meticulous work. Chaudhury focuses on the software that governs how the robot moves, ensuring it can translate requested targets into smooth, precise actions. The system must also respond quickly to change. If the probe meets unexpected resistance, if communication is interrupted, or if anything behaves outside expected boundaries, the platform needs to detect the issue immediately and stop in a controlled way.

Because reliability is non-negotiable, Chaudhury and her team built a robust simulation environment to test the system long before it is run on physical equipment. Working within a distributed ROS 2 framework that includes more than sixty software packages, the platform can replay high-fidelity datasets recorded during real procedures. By running recorded ultrasound images and clinician movements through simulated models, engineers can validate tracking, motion planning, and recovery behavior in a controlled setting. This helps ensure that software updates do not introduce unintended changes in how the robotic system behaves.

Ultrasound itself adds another layer of difficulty. The images can be noisy and full of artifacts influenced by breathing, fluid movement, and changes in tissue density. To maintain a consistent three-dimensional view, the platform combines AI-based reconstruction with feedback from the robot’s own sensors, helping it understand how the probe is positioned as the heart moves throughout each beat.

Making those calculations fast enough for real-time use required careful optimization. Chaudhury supported efforts to streamline the AI models so they could run more efficiently without sacrificing the quality of spatial information the system relies on. By reducing the computational load, the system can continue updating its estimate of position and alignment as the heart’s shape shifts through the cardiac cycle.

Laza’s progress has drawn attention within the cardiovascular community. The company’s framework was selected for the competitive Shark Tank presentation at the Transcatheter Cardiovascular Therapeutics conference, a forum that highlights emerging innovations in the field. Demonstrating that an automated approach can support real-time imaging without slowing procedures or introducing new risks is a critical step toward broader clinical confidence.

Even with these advances, significant work remains before robotic imaging becomes routine in catheterization laboratories. Regulatory review requires extensive documentation of how a system behaves at its limits, and clinicians will expect tools that can handle the unpredictable realities of anatomy without constant recalibration.

For the engineers building these systems, the path forward is clear. The goal is not to treat AI as a replacement for clinical judgment, or to assume an algorithm will always be right. The goal is to build an integrated system where intelligent guidance is continuously checked by strict, built-in safety boundaries. In the high-stakes environment of structural heart intervention, that balance of capability and control may define the next chapter of surgical robotics.