After Varian: No More Debate Over CT-Linac

This is an in-depth long article based on the evolution of radiotherapy technology, analysis of physical principles, and industrial strategic perspectives. It discusses not a single device, but a technology—one of the most valuable innovations in the field of radiotherapy over the past decade.

Just as domestic medical institutions were preparing to adopt United Imaging’s "integrated" CT-Linac, Varian suddenly struck back with the launch of HyperSight. Everyone was shocked, forcing a re-examination of the definition of "CT-Linac."

The emergence of this technology has rendered irrelevant the decade-long industry debate: "Should we integrate CT into the linear accelerator, or the linear accelerator into CT?" It has brought two revolutionary changes to the industry:

1).It is a "course correction" for the form of radiotherapy equipment, proving that in the pursuit of ultimate imaging, there is no need to sacrifice the isocenter—radiotherapy’s most cherished asset. A linear accelerator should inherently be a CT.

2).It is a "bold experiment" for the radiotherapy industry, demonstrating that replacing CT with innovative "hardware + algorithms" can also equip accelerators with diagnostic-grade imaging. In the future, every accelerator will be a CT-Linac.

Why must linear accelerators be equipped with diagnostic-grade CT? With Varian HyperSight, have CBCT images truly reached CT standards? These are questions we need to explore in detail.

Since Varian has not publicly disclosed its technical principles, we can only make inferences based on official revelations. We apologize to Varian if there are any inaccuracies—after all, truth becomes clearer through debate.

1. The "Original Sin" of CBCT

The core contradiction of radiotherapy has always been: "failing to hit the intended target (missed lesions) and over-irradiating unintended areas (collateral damage)."

Thus, the history of radiotherapy is a story of humanity’s quest to "see tumors clearly." From early 2D field verification to today’s CBCT, we have largely solved the problem of "whether the position is correct," but have never fully addressed "whether the delivery is accurate" and "whether the dosage is correct."

This is because CBCT has inherent flaws: slow scanning, severe motion artifacts, and inaccurate CT numbers. Medical physicists cannot perform precise dose calculations directly on CBCT images, and doctors struggle to distinguish soft tissue boundaries. This creates an awkward situation: the images we see on the day of treatment (CBCT) are vastly different from those used for treatment planning (simulation CT).

With the rise of adaptive radiotherapy, there is an urgent clinical need to "online" visualize subtle tumor changes and adjust treatment plans in real time—along with a desire for on-board imaging to meet diagnostic-grade CT standards.

To address this pain point, United Imaging offered a bold solution: if CBCT cannot provide clear images, integrate spiral CT and linear accelerator in series to create an "integrated CT-Linac." Unlike Siemens’ split-type CT-Linac, this innovation caught the industry’s attention.

The integrated CT-Linac solves the problems of "poor visibility" and "inaccurate calculation," but it cannot avoid radiotherapy physics’ most dreaded flaw: isocenter deviation.

2. The Achilles’ Heel of CT-Linac

The isocenter is the unique spatial intersection of the gantry rotation axis, collimator rotation axis, and treatment couch rotation axis. All dose calculations and radiation trajectories revolve around this infinitesimal point—it is the "sacred and inviolable" core of radiotherapy.

However, United Imaging’s CT-Linac adopts a "same couch, different centers" design, where the imaging center and treatment center are physically separated (approximately 2 meters for the uRT-linac 506c and 1 meter for the uLinac HalosTx; United Imaging is gradually reducing this distance). This means that after CT scanning, the patient must be moved from the "imaging position" to the "treatment position" via mechanical motion—this is the crux of the problem: "non-coincident centers" introduce mechanical errors and motion management challenges, posing significant risks to treatment accuracy.

United Imaging prioritized ultimate image quality, using "superior imaging" to offset the "mechanical risks of couch movement." To mitigate non-coincident center issues, it implemented a solution combining a movable couch base, laser ranging, and N-wire correction, striving to address registration inaccuracies caused by settlement.

Yet, in radiotherapy, coincident imaging and treatment centers are paramount. In-situ imaging and in-situ treatment represent the ideal state. The industry craves not the native integration of "CT + Linac," but the concentric, coaxial design of "Linac is CT"—seamlessly switching between diagnostic-grade imaging and treatment at the same isocenter in an instant.

Varian’s answer was not to create another accelerator with an external CT, but to invent a technology: HyperSight. Without altering the accelerator’s basic architecture, it uses a combination of "hardware + algorithms" to evolve the accelerator itself into a CT-capable device.

If United Imaging’s approach was "addition," Varian’s was "multiplication"—forcibly elevating the previously underperforming CBCT to diagnostic CT standards, enabling linear accelerators to achieve "in-situ imaging and in-situ treatment."

3. Can CBCT Truly Reach CT Standards?

Some argue that the advent of HyperSight marks a qualitative shift in CBCT technology from "auxiliary positioning" to "computational imaging." This raises a critical question: Given CBCT’s inherent limitations, how can it claim to meet CT standards?

Before answering, we must break down "CT standards" into two key categories—Radiation Therapy (RT) Planning Standards and Radiology Diagnostic Standards—both of which are essential. Based on FDA, NMPA, and CE approvals, as well as relevant evidence, we conduct a detailed technical comparison:

Dosimetric "CT Standards": CT Number Accuracy

This is the "CT standard" most concerned by medical physicists. In radiotherapy, "CT standards" first refer to the accuracy of electron density mapping—the foundation of radiation dose calculation. Traditional CBCT has highly volatile HU values (water’s HU can fluctuate by ±50), so physicists dare not use CBCT images directly for dose calculation. Instead, they must "map" electron density from the simulation CT via deformable registration.

According to multiple academic studies and Varian’s official data, the HU value stability and electron density curve linearity of HyperSight images are highly consistent with simulation CT, with deviations controlled within 1% [References 1-2].

Currently, HyperSight’s "use for dose calculation" has been certified by the FDA, CE, and NMPA, indicating it meets simulation CT standards. It is no longer a mere "reference image" but a "computational image"—HyperSight’s first revolutionary breakthrough.

Comparison of HU values between HyperSight, GE, and Siemens CT

Radiological "CT Standards": Soft Tissue Contrast

This is the "CT standard" most concerned by radiologists. Traditional CBCT excels at visualizing bones—even surpassing CT in spatial resolution—but struggles with soft tissues, resembling "viewing flowers through fog." Can the boundary between the prostate and rectum be distinguished? Can the pancreas and duodenum be differentiated? In the past, the answer was no.

From a radiologist’s perspective, high-end spiral CT theoretically maintains a signal-to-noise ratio (SNR) advantage when detecting 2mm low-contrast nodules in the liver. While HyperSight still has physical limitations under extreme conditions, these gaps have narrowed to a clinically imperceptible level. For soft tissue delineation required in radiotherapy (e.g., prostate-rectum boundaries, pancreatic contours, lymph node margins), its image quality is fully sufficient—qualifying as "diagnostic-grade."

Thus, HyperSight’s "use for clinical diagnosis" has also received FDA certification, complying with American College of Radiology (ACR) diagnostic standards [References 3-4]. This means it not only meets diagnostic-grade CT standards but also delivers radiation doses more than 10 times lower than conventional CT, laying the foundation for CT-Linac adaptive radiotherapy—HyperSight’s second revolutionary breakthrough.

HyperSight images on Halcyon (source: official website)

4. Evaluations from the Clinical Frontlines

After Varian launched HyperSight and it was implemented clinically, how did opinions from the frontlines change?

If you ask a radiologist: "Can I use HyperSight to replace the department’s 256-slice CT for hospital-wide cardiac coronary angiography screening?"

Answer: No need. Physical structure dictates high-end spiral CT’s dominance in ultimate SNR. However, in the context of tumor diagnosis, this gap is clinically imperceptible.

If you ask a radiotherapy specialist: "Can I use HyperSight to replace simulation CT and directly modify online adaptive treatment plans for patients?"

Answer: Absolutely. Through HyperSight, Varian told the industry: As long as it can accurately calculate doses, clearly delineate target volumes, and eliminate motion artifacts—it is a CT [Reference 5]. Whether it uses cone-beam CT or fan-beam CT is irrelevant. What matters is that all this is accomplished at the isocenter.

Is Varian HyperSight truly flawless?

Objectively, it still faces physical limits in ultra-low contrast resolution and handling complex metal artifacts.

Yet, in radiotherapy scenarios, HyperSight is clearly a "once-used, never-returned" technology: For doctors, it transforms "blind treatment" into "targeted treatment"; for technologists, it turns "patient discomfort" into "fast, smooth procedures"; for physicists, it changes "estimation" into "precision calculation."

Clinically, HyperSight’s greatest contribution is not parameter improvements, but eliminating the insecurity of "poor visibility" in radiotherapy workflows—truly realizing the concept of CT-Linac (integrated imaging-guided treatment) in real-world practice.