There is a number in interventional pulmonology that refuses to move. The diagnostic yield for biopsying a peripheral lung nodule sits around 70 percent, and it has stayed there through a decade of innovation.

Every year brings a better way to get a tool out to the edge of the lung, and every year the yield holds roughly where it was. That stubbornness is a clue, and reading it correctly matters, because the wrong reading turns into another decade spent improving the wrong step.

The right reading is not that the navigation work was wasted. It was necessary, and it succeeded. The right reading is that sampling a nodule has more than one hard step, and the field solved the first one. I want to lay out the second step, why the obvious fixes for it fell short, and why I treat the remaining opportunity as an open question rather than a solved one.

Three steps, and only one of them is solved

Sampling a peripheral nodule has three steps. Navigate to the lesion. Confirm the tool is actually in it. Acquire the tissue. The field poured a decade of effort into the first step and made extraordinary progress. Electromagnetic systems build a map from the pre-procedure CT and guide the catheter toward the target. Shape-sensing and robotics made the journey steadier and more repeatable.

By the measures of navigation, these tools succeed. The catheter gets to where the map says the lesion is. And then the diagnostic yield says, often enough, that getting there was not the same as being in it.

The map drifts from the body

Here is the quiet problem at the center of it. The map is built from a CT taken before the procedure, with the patient in one state. During the case the patient is in another. They breathe, they are sedated, parts of the lung collapse into atelectasis, and the lesion moves relative to the map.

The field has a name for this, CT-to-body divergence, and it means a navigation system can report success while the tool sits beside the target rather than in it.

That is not a flaw the field can navigate its way out of, because the problem appears after navigation is done.

The confirmation tools, and their tells

The field has not ignored confirmation. It has tried to bolt it on, and each attempt is instructive. Radial ultrasound was added to confirm the lesion at the tip, and it helped less than hoped, partly because collapsed lung tissue can look like a nodule on ultrasound, so the confirmation step itself can produce a false positive.

Tomosynthesis and augmented fluoroscopy sharpen the picture but inherit the limits of projection imaging. Cone-beam CT is the one that genuinely works; it confirms the tool in the lesion in three dimensions and the data show it lifts yield when used. But it drags a second imaging system into the room and adds radiation, cost, time, and workflow complexity to every case.

That operators adopt it anyway is the clearest evidence of how much real confirmation is worth.

The robotics lesson

Robotics is the most visible advance of the era, and its results are the cleanest diagnosis of where the bottleneck is not. The robot navigates beautifully, steadier and more reachable than what came before. And when yield is measured head to head against electromagnetic navigation, it comes out broadly similar. That is not a criticism of robotics, which is excellent at what it does.

It is a precise finding. If making navigation dramatically better does not move the yield, then navigation was not the thing holding the yield down. The field built a faster, smoother way to arrive, and arrived at the same uncertainty about whether it was in the right place.

Confirmation is a seeing problem

Once you accept that navigation is solved and confirmation is not, the shape of the remaining problem gets clearer. You cannot navigate your way to certainty about whether the tool is in the lesion. That is something you would have to see, directly, at the tip, independent of the map that already drifted. In other words, the open part of this procedure is a seeing problem, not a steering one.

Whether a direct, near-field look at what is in front of the tool would close it is precisely the question, and it is an empirical one. The mechanism is straightforward and the gap is well-described, but the result would have to be earned in evidence, not assumed.

The economics of a non-diagnostic biopsy

It is easy to treat 70 percent as a technical statistic. It is not. Walk the path of a single non-diagnostic result and the cost compounds. A patient came in worried about a spot on a scan and leaves without an answer. The case has to be repeated, with its own sedation, pneumothorax risk, room time, and staff. In between there are weeks or months of additional CT surveillance and of not knowing.

Some patients get routed to a more invasive transthoracic biopsy or to surgery for what should have been a bronchoscopic answer. Multiply that by the volume of peripheral nodules worked up every year and the missing thirty percent stops looking like a statistic and starts looking like one of the larger pools of avoidable cost and delay in thoracic medicine.

Why the stakes are real, without overstating them

Lung nodules are not a neutral target. Many are early cancers, and in lung cancer, stage at diagnosis weighs heavily on prognosis. That is the cited backdrop that makes this particular yield number matter more than most. I want to be careful, though, about the leap from a better confirmation step to better outcomes.

It is plausible that converting more first attempts into answers would shorten time to diagnosis, and in a stage-dependent disease that would matter. It is also unproven, and it has to be earned with evidence rather than claimed. The percentage is standing in for something that matters, which is why it deserves attention, not why anyone should overpromise about it.

The confidence trap

There is a subtle reason the confirmation gap has been hard for the field to face. The current tools do not fail loudly. They fail with confidence. A navigation system that has guided the catheter to the mapped location reports success, and the operator reasonably believes it. The failure reveals itself days later, when the pathology comes back non-diagnostic.

The gap between felt certainty in the room and actual position in the body is invisible at the only moment it could be acted on. That is the deepest reason confirmation, rather than more navigation, is where the open question lives.

The alternative is a needle through the chest

There is a reason the field keeps pushing on bronchoscopic yield rather than sending these patients elsewhere. The main alternative for a peripheral nodule is a transthoracic biopsy, a needle through the chest wall under CT guidance. It has higher yield for many peripheral lesions and a materially higher rate of pneumothorax, sometimes requiring a chest tube and admission.

The bronchoscopic route is gentler, reaching the periphery from inside the airway without crossing the pleura. That promise is only fully redeemed if the bronchoscopic yield is high enough to make the gentler route the reliable one, which is the practical reason the stuck yield matters.

Why now, and why I’m watching

The same why-now that applies across interventional imaging applies here. Earlier attempts to see at the tip of a bronchoscope ran into sensors too large for the working channel, too little processing near the tip, and images too marginal to act on. What has changed is what has changed across the board: smaller sensors, more compute near the tip, more mature reconstruction.

The lung is, if anything, the indication where the navigation investment has most clearly hit its ceiling, which is what makes the confirmation question worth following. As with the rest of this space, the payoff is an open, evidence-gated question, and that is exactly the posture I think it deserves.

The honest limits

A confirmation step, however good, would not diagnose the nodule. Tissue is still tissue and pathology has the final word. It would not guarantee a result or make a hostile nodule easy to reach.

What the field is missing is narrower and more specific: a way to know, at the moment of sampling, that the tool is in the lesion rather than beside it. Whether that gap gets closed, and by what, is unsettled. That it is the right gap to be working on is what the stuck number, read honestly, suggests.