ILD Pattern Pitfalls
- Interstitial lung disease (ILD) is read by pattern and distribution, not by hunting for a single magic finding — and the patterns love to disguise themselves.
- The classic trap is calling honeycombing when you're actually looking at traction bronchiectasis, subpleural cysts, or even paraseptal emphysema.
- "Ground glass" is a description, not a diagnosis — it can be inflammation, fibrosis hiding below the resolution of your eye, or just blood and edema crashing the party.
- Distribution (upper vs. lower, central vs. peripheral) and the company a finding keeps usually break the tie better than the finding itself.
- Bad technique — a shallow breath, expiration, or motion — manufactures fake disease. Always blame the scanner before you blame the lung.
Interstitial lung disease is the part of chest imaging where confident people go to be humbled. The findings are subtle, the names sound like a law firm (UIP, NSIP, HP, COP — honestly, pick four), and the whole game is matching a pattern to a diagnosis rather than spotting one obvious thing. So naturally, the patterns spend their free time impersonating each other. Let me walk you through where I've face-planted, so you can face-plant a little less.
Honeycombing is not "any cluster of holes"
Honeycombing is the money finding for the usual interstitial pneumonia (UIP) pattern, and because it changes prognosis, everyone is itching to call it. That itch is the problem.
True honeycombing is clustered, stacked cystic airspaces with shared, thick walls — think of a honeycomb (shocking, I know), or a stack of bubble wrap that's been compressed against the pleura. The key features: the cysts are usually similar in size, they sit in layers in the subpleural lungs, and they share walls like row houses.
What fools you is anything else round and dark down there:
| Mimic | How it's faking you out | The tell |
|---|---|---|
| Traction bronchiectasis | Dilated airways cut in cross-section look like cysts | Scroll up/down — real airways are tubes; they connect and branch. |
| Paraseptal emphysema | Single row of subpleural lucencies | Walls are thin (or absent); no stacking, often in upper lobes. |
| Subpleural cysts of other ILDs | Scattered cysts that look honeycomby | They don't stack into clusters; check distribution and the rest of the lung. |
The single most common honeycombing error is calling traction bronchiectasis "honeycombing." Both mean fibrosis, but honeycombing carries more diagnostic weight for a confident UIP read. Before you commit, scroll through the stack: airways are continuous tubes that branch, while honeycomb cysts are dead-end pockets stacked in layers.
"Ground glass" is a symptom of your vocabulary, not a diagnosis
Ground-glass opacity (GGO) is hazy increased lung density that you can still see vessels through — like breathing on a cold window. It feels like a finding. It is actually a confession that something is filling or thickening the lung below the resolution your eyeballs can resolve.
The trap is treating ground glass as a single thing. It can be:
- Active inflammation — the kind that might actually get better with treatment.
- Fine fibrosis — fibrosis so subtle it just reads as haze, especially when paired with traction bronchiectasis or architectural distortion. That combination means it's not reversible, and that changes everything.
- Filling, plain and simple — blood, pulmonary edema, pus, or organizing pneumonia crashing the party.
Ground glass with traction bronchiectasis or volume loss is usually fibrosis, not reversible inflammation. Calling it "active, treatable disease" sends the patient down the wrong road. The company GGO keeps matters more than the GGO itself.
Distribution breaks the tie when the findings won't
When two patterns look identical zoomed in, zoom out. Where the disease lives is often the deciding vote.
- Lower-lobe, peripheral, basal → think UIP-flavored fibrosis.
- Upper-lobe predominant → start thinking hypersensitivity pneumonitis or sarcoid in the conversation.
- Relative sparing right at the pleura → a clue that points toward the NSIP (nonspecific interstitial pneumonia) end of things.
None of these are laws — they're weighted dice. But when the finding-level call is a coin flip, distribution is how you stop flipping coins.
The mosaic that isn't disease (and the one that is)
Mosaic attenuation — patchy light and dark lung, like a poorly mixed latte — is a beautiful trap because it can mean air trapping, small-vessel disease, or infiltrative disease, and it can also just be normal lung next to abnormal lung playing tricks in the contrast.
If you suspect air trapping, get (or look for) expiratory images. On expiration, normal lung empties and turns gray, while air-trapped regions stay stubbornly dark — the difference exaggerates and the diagnosis pops. On inspiration alone, that same mosaic can be nearly invisible.
Blame the scanner before the lung
Half the "disease" I've nervously pointed at early on was technique. A patient scanned in expiration has diffuse hazy, denser-looking lung that mimics ground glass everywhere. Dependent atelectasis — the gravity-squished gray at the back of the lungs — fakes early fibrosis until you flip the patient prone and it vanishes. Motion smears vessels into reticulation that doesn't exist.
Before you diagnose a single thing, sanity-check the technique. If the haze is dependent, symmetric, and gone on prone or inspiratory images, it was never disease — it was physics. This is exactly why a proper HRCT protocol, with inspiratory, expiratory, and sometimes prone images, exists.
The honest takeaway: in ILD, no single finding earns a diagnosis on its own. You read the pattern, you read the distribution, and you rule out the scanner before you rule in the disease. Do that, and the law firm of UIP, NSIP, HP & COP becomes a lot less intimidating.