Radiation Biology & Risk
- Ionizing radiation hurts cells mainly by damaging DNA — directly, or indirectly by making reactive molecules out of water.
- Effects come in two flavors: deterministic (above a threshold, dose-dependent severity, like skin burns) and stochastic (no threshold, dose-dependent probability, like cancer).
- At the doses used in diagnostic imaging, the worry is the stochastic kind — a small added cancer risk, not burns.
- We measure the biologically relevant dose in sieverts (Sv), which weights different radiation and tissues by how much harm they do.
- The honest summary: individual risk from a single diagnostic scan is tiny, but it isn't zero, so we don't hand out scans like candy.
Radiation is one of those topics where people either shrug it off completely or assume every chest X-ray is a tiny Chernobyl. The truth lives in the boring middle, and the boring middle is genuinely interesting once you see how it works. So let me walk you through what the radiation actually does in there, and how scared (or not) to be about it.
How radiation actually breaks things
Picture your DNA as a long zipper. It works fine until something snags a tooth. Ionizing radiation is energetic enough to knock electrons clean off atoms — that's what "ionizing" means — and when it barrels through a cell, it can snag that zipper in two ways.
The direct route: a photon or particle scores a hit right on the DNA strand. Bullseye, broken tooth.
The indirect route is sneakier and, in our tissues, actually the bigger player. Your cells are mostly water. Radiation splits water molecules into twitchy, reactive fragments called free radicals, which then go careening around like a toddler with a marker, scribbling on whatever they touch — including DNA. So a lot of the damage isn't the radiation hitting your genes directly; it's the radiation poisoning the water around your genes.
Most of these breaks get repaired flawlessly. The problem is the occasional repair that goes wrong, because that's where lasting trouble starts.
A double-strand break — both sides of the zipper cut at the same spot — is the dangerous one. Single-strand breaks have an intact template to copy from, so the cell usually fixes them cleanly.
Two kinds of harm: thresholds vs. dice
Here's the distinction that organizes the whole field.
| Effect type | Threshold? | What dose changes | Examples |
|---|---|---|---|
| Deterministic | Yes — below it, nothing | The severity of the effect | Skin erythema, hair loss, cataracts |
| Stochastic | No (assumed) | The probability of the effect | Cancer, heritable mutations |
Deterministic effects are like a sunburn. Below a certain exposure, your skin shrugs it off. Cross the threshold and you get a burn, and the more you cross it, the worse the burn. These need a lot of dose — you essentially only see them after long fluoroscopically-guided procedures or radiation therapy, not after a CT.
Stochastic effects are like buying lottery tickets, except it's a lottery you desperately want to lose. There's no safe number of tickets where the chance is exactly zero; every ticket nudges your odds up a hair. One bad cell that survives a botched repair and keeps dividing is, in principle, all it takes. This is the model we assume for diagnostic doses — no threshold, risk rising in proportion to dose. It's a deliberately cautious assumption, and there's real scientific debate about whether risk at very low doses is truly linear, but for safety planning we behave as if it is.
Severity scales with dose for deterministic effects; probability scales with dose for stochastic effects. Burns get worse; cancer just gets likelier.
Why we measure dose in sieverts
Not all radiation energy is equal, and not all tissues are equally fragile. So we don't just measure raw energy deposited — we weight it.
Start with the energy absorbed per kilogram of tissue: that's the gray (Gy). Then weight it for the type of radiation (some particles are nastier per unit energy) and for which tissues got hit (bone marrow and gonads are touchier than, say, skin). The result is the effective dose, measured in sieverts (Sv) — the closest single number we have to "how much biological harm did this exposure risk?" Diagnostic doses are small, so you'll usually see millisieverts (mSv).
That weighting is exactly why dose to a fetus or a child gets its own careful treatment — see pregnancy and pediatric dose — and why the raw scanner output numbers get translated into CT dose metrics before anyone talks about risk.
So how worried should you be?
Honestly? For any single diagnostic study, barely at all. Everyone walks around in a sea of natural background radiation — cosmic rays, radon, the potassium in your own bananas-and-bones — and a typical diagnostic scan adds a modest slice on top of that baseline. The estimated cancer risk from one study is small enough that, for a patient who needs the information, the benefit almost always dwarfs it.
The catch is that "small per scan" and "we image millions of people" multiply into something a population-level public-health person genuinely cares about. That's the whole reason we don't scan reflexively.
Don't tell a sick patient that a needed CT is "dangerous." The risk of missing a diagnosis is usually far larger than the radiation risk. The right move is justified imaging, not avoided imaging — fear of dose shouldn't talk anyone out of a scan they clinically need.
The one thing to carry out the door
Radiation harms cells by damaging DNA, the damage we worry about in imaging is the no-threshold, probability-raising stochastic kind, and we keep it sensible by giving every patient only the dose the question actually requires. That last principle has a name and its own page: ALARA and protection principles — as low as reasonably achievable. Biology tells you why we bother; ALARA tells you what to do about it.