ILD: A Pattern-Based Deep Dive
- ILD ("interstitial lung disease") is a big family of diseases that scar and stiffen the delicate scaffolding the lungs use to hold air. The job of the CT is to sort that family by pattern, not by name.
- The high-resolution chest CT is the star here. Two questions drive everything: what does the scarring look like, and where does it live (top vs bottom, center vs edge).
- The pattern radiologists fear most is UIP (usual interstitial pneumonia): bottom-and-edge predominant honeycombing with traction bronchiectasis. It strongly suggests idiopathic pulmonary fibrosis (IPF).
- Diagnosis is a team sport: a radiologist, a pulmonologist, and often a pathologist argue it out together. No single test wins alone.
- Ground glass that's diffuse and upper-zone hints at something inhaled or inflammatory; basal honeycombing hints at established fibrosis.
Interstitial lung disease is where a lot of smart people quietly panic, because it's not one disease — it's a giant, unruly family reunion of conditions all crammed into one acronym. The good news: you don't have to memorize the whole guest list. You have to learn to read the room. On CT, ILD is a problem of patterns, and patterns are learnable.
Start with the scaffolding. Picture the lung as a sponge: not solid, but a fine mesh of struts and walls (the interstitium) holding millions of tiny air pockets open. ILD is what happens when that mesh gets inflamed, thickened, or scarred. The sponge stiffens, the air pockets distort, and the lung gets worse at its one job — being soft and full of air. If you want the gentle on-ramp first, start with the interstitial lung disease patterns page; this one goes a level deeper.
The CT is the main event (and it's a special CT)
Plain film will whisper that something's wrong — small lungs, hazy bases, lines that shouldn't be there — but it's hopeless at telling the fibroses apart. The real tool is the high-resolution CT (HRCT), which is just a chest CT acquired with thin slices and a sharp reconstruction so the fine mesh actually shows up. If you're rusty on how a chest CT is built and read, the approach to chest CT page is the warm-up.
When I read one of these, I'm answering two questions, and that's genuinely most of the battle:
- What is the dominant finding? (Reticulation? Ground glass? Honeycombing? Nodules?)
- Where does it live? (Upper vs lower zones, and central vs the very edge of the lung — the subpleural region right under the ribs.)
A vocabulary of textures
These words sound fancy, but each is just a texture you can picture.
| Finding | What it looks like | Plain-English version |
|---|---|---|
| Reticulation | A fine net or lattice of lines | The mesh got thick enough to see. |
| Ground glass | A soft gray haze, vessels still visible through it | Like breathing on a window — frosted, not blacked out. |
| Honeycombing | Clustered small cysts stacked in rows | End-stage scarring; the lung remodeled into little air pockets. |
| Traction bronchiectasis | Airways pulled wide and irregular | Scar tissue yanking the airways open like taffy. |
Ground glass is the great tease of chest CT. It can mean active, treatable inflammation — or just thin early fibrosis. Context and the company it keeps (does it sit next to honeycombing?) decide which.
The pattern that earns its own paragraph: UIP
The single most important pattern to recognize is usual interstitial pneumonia (UIP). Picture the damage starting at the bottom of the lungs and at the very edges, right up against the chest wall, then creeping inward and upward over years. The classic UIP look is subpleural, basal-predominant honeycombing with traction bronchiectasis, and crucially, not much ground glass.
Why care so much? Because a confident UIP pattern strongly points toward idiopathic pulmonary fibrosis (IPF) — a relentless, scarring disease with a serious prognosis and very different treatment from the inflammatory ILDs. Calling UIP correctly genuinely changes a patient's life.
Don't confuse honeycombing with emphysema or with traction bronchiectasis. Emphysema cysts have no walls and cluster up high near the smoking damage; honeycombing has thin walls, sits at the bases and edges, and stacks in rows. Mistaking one for the other sends the whole diagnosis off a cliff.
When it's NOT UIP
If instead the disease favors the upper lungs, spares the immediate edge, and is full of ground glass and tiny nodules, your mind should swing toward an inhaled or inflammatory cause — think reactions to dusts, molds, or birds, or connective-tissue disease attacking the lung. The distribution map is doing real diagnostic work.
A quick orientation, kept deliberately loose because real lungs are messy:
| Clue | Leans toward |
|---|---|
| Basal + subpleural + honeycombing | UIP / fibrotic disease |
| Upper-zone + ground glass + small nodules | Inhalational / inflammatory ILD |
| Lots of ground glass, changes fast | Active inflammation or, importantly, edema — not fibrosis |
That last row is a reminder: fluid backing up into the lungs (pulmonary edema) makes ground glass too, and it comes and goes in hours. Fibrosis doesn't pack up and leave overnight. If the haze cleared by Tuesday, it was never scar.
Why this is a team sport
ILD is one of the few areas where the official, guideline-endorsed way to reach a diagnosis is to put a radiologist, a pulmonologist, and (when needed) a pathologist in a room and have them argue. This multidisciplinary discussion exists because no single dataset — not the CT, not the biopsy, not the labs — is reliable alone. The CT pattern carries enormous weight, but it's a vote, not a verdict.
So your reading isn't just "fibrosis, present." It's a structured contribution: dominant finding, distribution, your best pattern label (and how confident you are), plus the mimics you considered. Note the airways too — distorted, dilated airways often travel with fibrosis, and the dedicated airways disease page handles those in detail.
If you remember one thing: in ILD, the radiologist's superpower isn't naming the disease outright — it's reading the pattern and its address precisely enough that the team can name it together. Texture and location. Get those right, and you've done the hard part.