Imaging Nerd

GI Bleed (CTA)

Key Points
  • CT angiography (CTA) for a GI bleed is a stakeout: you're trying to catch contrast leaking out of a vessel and into the bowel in the act.
  • The money finding is a blush or jet of contrast in the bowel lumen on the arterial phase that grows or spreads on the later (venous/delayed) phase.
  • It needs active, fairly brisk bleeding at the moment of the scan — a stopped or trickling bleed can hide.
  • The protocol is non-negotiable: a non-contrast scan first, then arterial, then a delayed phase. Skip the non-contrast and you'll be fooled by things that were already bright.
  • CTA tells you where and roughly what, which lets the next team — angio or endoscopy — go straight to the source.

Someone is bleeding somewhere into about thirty feet of bowel, and the question lands on your scanner: where? That's the whole job of a GI bleed CTA. It's less a diagnosis and more a manhunt — and like any good stakeout, success depends entirely on whether your suspect does something incriminating while the camera is rolling.

The core idea: catch contrast where it doesn't belong

You inject IV contrast, it lights up the arteries, and if a vessel has a hole in it, contrast spills out the hole and pools in the bowel lumen — a place blood and contrast are not supposed to be hanging out. That little puff of brightness sitting in the gut is the bleed. The radiologists call it a contrast blush or, when it's brisk, active extravasation. In English: a leak caught on camera.

The catch — and it's a big one — is timing. The contrast is only in the arteries for a few seconds. If the bleeding has paused (and GI bleeds love to pause, like a faucet with a mind of its own), there's nothing to leak, and the scan looks clean even though the patient is genuinely bleeding. CTA is a snapshot, not a movie.

Note

CTA detects bleeding at a rate that's surprisingly small in everyday terms — roughly the trickle that fills a contact-lens case over many minutes, not a gusher. But it has to be happening during the scan. Intermittent bleeders are the bane of this study.

Why the protocol has three passes

This is the part people skip and regret. A proper GI bleed CTA is three scans of the same belly:

PhaseWhenWhat it's for
Non-contrastBefore any IV contrastBaseline. Flags things already bright — surgical clips, retained pills, prior contrast, hyperdense clot — so you don't mistake them for a leak.
ArterialContrast in the arteriesThe main event. A fresh blush appearing here that wasn't on the non-contrast is your bleed.
Delayed/venousA bit laterConfirmation. True extravasation pools, spreads, and changes shape as more contrast leaks out.

The non-contrast pass is the unsung hero. Without it, you can't tell "new bright spot" (leak) from "always-bright spot" (a clip from last year's surgery). The delayed pass is the lie detector: a real leak grows and migrates downstream with the bowel's contour, while an impostor — say, a hypervascular bit of mucosa — lights up and then politely fades like normal tissue.

Figure · CT
Three-panel CT angiography of the abdomen: (1) non-contrast image showing no luminal hyperdensity, (2) arterial phase showing a focal high-density blush within a small-bowel loop, and (3) delayed phase showing the contrast has increased and spread along the dependent lumen — the classic signature of active GI bleeding.

Reading it without fooling yourself

Once you spot a bright spot in the lumen, ask three things: Was it there before contrast? Does it grow on the delayed phase? Can you trace it back along a feeding artery? A yes-no-yes pattern is a bleed. A yes anywhere in the first question usually means you've been catfished.

Pitfall

The all-time classic trap: a hyperdense object that was bright before contrast. A retained pill, a calcified node, a surgical clip, oral-contrast residue, or old hemorrhage can all sit in or near the bowel looking exactly like a blush. The non-contrast scan is the only thing that saves you — which is precisely why it's first. No baseline, no verdict.

CTA is also a generous study because it shows you the neighborhood, not just the leak. It can point at the likely culprit — diverticula, a tumor, inflamed bowel, or abnormal tangled vessels (angiodysplasia) — and that context helps the next team aim. Diverticula are a leading source of lower GI bleeds (the bleeding kind are usually not the inflamed kind — diverticulitis is its own, separate story).

Where CTA sits among the options

Think of it as the fast, widely available scout. It needs no prep, takes minutes, and localizes the bleed so the cavalry can be specific.

  • If CTA is positive, the patient often goes straight to catheter angiography and embolization (plugging the bleeding vessel) — the roadmap is already drawn.
  • If you suspect a slow, sneaky, intermittent bleed, a nuclear medicine tagged-red-cell scan is more patient — it watches for a long stretch and catches lower bleeding rates than CTA.
  • For an upper bleed in a stable patient, endoscopy often goes first, since it can look and treat.
Clinical Pearl

A negative CTA does not mean the patient isn't bleeding — it means they weren't bleeding fast enough during those few seconds. In a brisk, active bleeder, CTA is excellent at saying where. In an intermittent one, a clean scan just means "try again, or switch tools."

The one thing to remember

A GI bleed CTA is a three-phase stakeout for contrast escaping into the bowel: nothing on non-contrast, a fresh blush on arterial, and that blush growing on delayed. Get those three frames and you've turned "bleeding somewhere in thirty feet of gut" into an X on a map — which is exactly what the person about to fix it needs.