How PET Detectors Turn Gamma Rays Into an Image

How PET scanners detect pairs of annihilation photons, determine where radioactive events occurred inside the patient, and build a metabolic image from millions of individual detections

PET imaging is very different from conventional X-ray imaging.

Back to Biomed Basics

What This Page Explains

This page covers:

The Simple Version

PET imaging begins inside the patient.

A simplified chain looks like:

Radiotracer Decays

↓

Positron Is Emitted

↓

Positron Meets Electron

↓

Annihilation Produces Two 511 keV Photons

↓

Photons Travel in Nearly Opposite Directions

↓

PET Detector Ring Detects Both

↓

Scanner Treats Them as One Coincidence Event

↓

Software Draws a Possible Line Between the Detectors

↓

Millions of Events Are Reconstructed Into an Image

PET does not usually know the exact location of one decay event from one coincidence.

Instead, each event provides a possible line.

Millions of those lines statistically build the image.

Start With the Radiotracer

PET requires a radioactive tracer containing a positron-emitting radionuclide.

A familiar example is:

Fluorine-18

used in FDG imaging.

Other tracers use different radionuclides and target different physiological processes.

The Tracer Has Biological Behavior

The scanner is not merely imaging:

Radiation.

It is imaging where a particular radiopharmaceutical accumulates.

That distribution reflects biology.

FDG

FDG is related to glucose metabolism.

Tissues with high glucose utilization may accumulate more tracer.

That is one reason PET can show metabolic activity.

Radioactive Decay

The radionuclide is unstable.

During decay, it may emit a:

Positron.

What Is a Positron?

A positron is the antimatter counterpart of the electron.

It has:

The Positron Does Not Travel Very Far

After emission, the positron travels through tissue, losing energy through interactions.

Eventually it slows enough to interact with an electron.

Annihilation

When the positron and electron meet:

They annihilate.

Their mass is converted into electromagnetic energy.

The event typically produces two gamma-ray photons.

Photon Energy

Each photon has approximately:

511 keV

of energy.

Why 511 keV?

That corresponds to the rest-mass energy of an electron or positron.

The two particles annihilate and produce two photons carrying the energy.

Opposite Directions

To conserve momentum, the two photons travel in nearly opposite directions.

Approximately:

180 degrees apart.

There are small real-world deviations from perfect 180-degree geometry, but the basic principle is opposite travel.

This Is the Core PET Trick

If two detectors on opposite sides of the scanner detect 511 keV photons at nearly the same time:

The scanner assumes they may have come from the same annihilation event.

Coincidence Detection

This paired detection is called:

Coincidence detection.

Timing Window

The scanner only considers two detections to be related if they occur within a very short timing window.

Why Timing Matters

The scanner is constantly detecting radiation.

If any two events were paired regardless of time, unrelated photons would constantly be mistaken for real annihilation pairs.

True Coincidence

A true coincidence occurs when both detected photons came from the same annihilation event and reached the detectors without being redirected in a way that invalidates the assumed line.

Line of Response

When two detectors register a valid coincidence, the scanner knows the annihilation occurred somewhere along the line connecting those detector locations.

That line is called a:

Line of Response, or LOR.

One Event Does Not Give Exact Position

This is important.

Without additional timing information, the scanner does not know exactly where along that line the event occurred.

It only knows:

Somewhere along here.

Millions of Lines

As the scan continues, enormous numbers of coincidence events generate many lines of response.

Where many statistically consistent lines intersect:

Activity is likely concentrated.

Reconstruction algorithms use that information to estimate the tracer distribution.

Detector Ring

PET scanners arrange radiation detectors in a ring around the patient.

Modern systems may contain many rings along the length of the scanner.

This creates a cylindrical detector geometry.

Why a Ring?

The annihilation photons travel in opposite directions.

Surrounding the patient makes it likely both photons can be detected.

Axial Coverage

More detector rings provide greater coverage along the patient's head-to-foot direction.

Modern systems may have substantially wider axial fields of view than older PET systems.

Detector Crystals

Gamma rays cannot be detected directly by ordinary visible-light sensors.

PET systems often use:

Scintillation crystals.

Scintillator

When a 511 keV gamma photon interacts with the crystal, the crystal produces a flash of visible light.

PET Conversion Chain

511 keV Gamma Photon

↓

Scintillator

↓

Visible-Light Flash

↓

Photodetector

↓

Electrical Pulse

↓

Timing + Energy Measurement

Scintillation Materials

PET detector crystals may use materials such as:

and other scintillators depending on scanner design.

What Makes a Good PET Scintillator?

Important properties include:

Stopping Power

The crystal should have a good probability of interacting with the high-energy gamma photon rather than letting it pass through undetected.

Fast Scintillation

Timing matters tremendously in PET.

The crystal should produce its light quickly enough to support precise coincidence timing.

Photodetector

The visible-light flash must be converted into an electrical signal.

Older systems often used:

Photomultiplier tubes, or PMTs.

Newer systems may use:

Silicon photomultipliers, or SiPMs.

Photomultiplier Tube

A PMT is extremely sensitive to small amounts of light.

A photon reaching the photocathode releases an electron.

A series of internal stages multiplies that signal.

The final output is a measurable electrical pulse.

PMTs Need High Voltage

Photomultiplier tubes generally require a high-voltage supply.

That is one part of the PET detector electronics that may require monitoring and calibration.

Magnetic Fields and PMTs

Traditional PMTs are sensitive to magnetic fields.

This becomes especially important in combined:

PET/MRI

systems.

Silicon Photomultipliers

SiPMs are solid-state photodetectors.

They contain many tiny avalanche photodiode elements operating in a way that allows single-photon-level sensitivity.

Why SiPMs Are Attractive

They can offer:

which is particularly helpful for modern PET designs.

Detector Block

A PET detector module may contain many scintillator crystal elements coupled to one or more photodetectors.

The system must determine:

Which crystal was hit?

Crystal Identification

Depending on detector design, the pattern of light reaching the photodetectors helps identify which crystal received the gamma interaction.

Energy Measurement

The size of the electrical pulse is related to how much energy was deposited in the scintillator.

The scanner uses this to estimate the detected photon energy.

Why Energy Matters

The desired annihilation photon energy is:

511 keV.

If the measured energy is far from the expected range:

The event may be rejected.

Energy Window

The scanner accepts detected events within a selected energy range around the expected photon energy.

This helps reject some unwanted events.

Scatter

One of the major complications in PET is:

Compton scatter.

A 511 keV photon may interact inside the patient and change direction before reaching the detector.

Why Scatter Is a Problem

The scanner assumes the photon traveled approximately straight from annihilation site to detector.

If it scattered:

The detected line of response is wrong.

Scatter Also Reduces Photon Energy

That is one reason energy measurement helps identify some scattered events.

Not All Scatter Is Rejected Perfectly

PET reconstruction also uses scatter-correction methods.

Random Coincidences

Imagine two unrelated radioactive decay events occur close together in time.

One photon from event A hits one detector.

One photon from event B hits another detector.

If they occur within the coincidence timing window:

The scanner may falsely pair them.

That is called a:

Random coincidence.

Randoms Add Background Error

Systems estimate and correct for random events.

Timing Window Tradeoff

A wider coincidence window catches more true pairs but can also admit more random pairs.

A narrower window reduces randoms but demands better timing performance.

Time-of-Flight PET

Modern PET systems may use:

Time-of-flight, or TOF.

This adds another powerful piece of information.

The Two Photons Travel at the Speed of Light

If the annihilation occurs exactly halfway between two detectors:

The photons arrive at nearly the same time.

If the annihilation occurs closer to one detector:

That photon arrives slightly earlier.

Time Difference Gives Position Information

The timing difference allows the scanner to estimate where along the line of response the annihilation occurred.

Not Exact Position

Timing resolution is not perfect.

TOF narrows the likely location along the line rather than determining one exact point.

Why TOF Helps

It improves the reconstruction's ability to localize activity and can improve image signal-to-noise characteristics.

Timing Resolution

Better detector and electronics timing means:

More precise localization along the line of response.

This is one reason fast scintillators and SiPMs are important in modern PET.

Detector Timing Calibration

All detector channels must agree on time extremely accurately.

If one detector's timing is offset:

Coincidence localization can be affected.

Energy Calibration

Detector channels also need proper energy calibration.

A crystal or channel that reports incorrect energy may:

Gain

Photodetector gain can vary.

The system calibrates detector response so channels behave consistently.

Temperature

Detector electronics and SiPM gain can be temperature-sensitive.

Modern systems monitor and compensate for this.

Why Cooling Matters

Stable detector temperature helps maintain:

Detector Temperature Fault

A scanner may still physically detect events while calibration becomes unreliable.

The system may therefore inhibit acquisition or fail QC rather than produce questionable data.

Normalization

Not every detector pair has exactly the same sensitivity.

PET systems use normalization corrections to compensate for differences.

Why Detector Sensitivity Varies

Possible reasons include:

Normalization Calibration

A known source or calibration procedure helps determine relative detector sensitivity.

The reconstruction then corrects those differences.

Bad Detector Element

A failing crystal or photodetector channel may produce:

One Failed Channel Can Affect Many Lines of Response

Because each detector participates in coincidence pairs with many detectors around the ring.

That means a localized detector problem can create characteristic patterns in the data.

Sinogram

PET coincidence data can be represented in forms such as a:

Sinogram.

This is not the final clinical image.

It organizes projection/coincidence data in a way useful for reconstruction and diagnostics.

Detector Gaps Can Be Visible in Raw Data

A bad detector block may create:

in sinogram or QC data.

Reconstruction

PET reconstruction estimates the spatial distribution of radioactive activity.

Modern systems commonly use iterative reconstruction methods.

The Final Image Is Heavily Corrected

The scanner does not simply count coincidence events and display them directly.

Corrections may include:

depending on system design.

Attenuation

Gamma photons can be absorbed or scattered inside the body before reaching the detectors.

Therefore activity deep inside the body may be undercounted unless attenuation is corrected.

PET/CT

Most modern clinical PET systems are combined with CT.

The CT portion provides:

Attenuation Correction

CT data can be used to estimate how strongly different tissues attenuate the PET photons.

The PET reconstruction then compensates for expected losses.

PET and CT Must Be Spatially Aligned

This is crucial.

If PET and CT images do not line up correctly:

Attenuation correction can be applied to the wrong locations.

That can create artifacts or inaccurate uptake appearance.

Patient Motion Between PET and CT

Because the CT and PET portions are acquired at different times, patient motion can cause misregistration.

Breathing Motion

Chest and diaphragm position can differ between:

This may produce apparent mismatch.

Not Every PET/CT Misregistration Is Hardware Failure

Patient motion and respiratory differences are important causes.

CT Is Not Only There for Pretty Anatomy

It can directly influence corrected PET data.

This is important for troubleshooting.

A CT calibration or alignment problem can affect PET attenuation correction.

PET Without CT

Some systems may use other attenuation-correction approaches.

But PET/CT is very common clinically.

Count Rate

The scanner records many events per second.

If activity is very high:

Detector/electronic processing can become saturated or experience:

Dead time.

Dead Time

After processing one event, detector electronics may need a very short recovery period before accurately processing another.

At very high event rates:

Some events may be lost.

Dead-Time Correction

The system estimates and corrects for this behavior within supported ranges.

Radioactive Decay

Tracer activity decreases over time according to its physical half-life.

The scanner can correct measured activity for radioactive decay when appropriate.

Quantitative PET

PET can provide quantitative measures such as:

SUV, standardized uptake value.

Accurate quantitation depends on more than the detector.

It can involve:

Dose Calibrator Relationship

PET scanner calibration and the dose-calibration system need consistency for accurate quantitative measurements.

Cross-Calibration

Quality programs may compare:

to ensure quantitative agreement.

Why This Matters to Biomeds

A PET image can look visually reasonable while quantitative values are wrong.

That is similar to many other medical devices:

Function is not the same as accuracy.

Daily QC

PET systems commonly perform regular quality-control procedures.

These may evaluate:

depending on system.

QC Trend Matters

A gradual decline can reveal detector issues before obvious clinical image problems appear.

Calibration Source

Systems may use radioactive calibration sources or other built-in methods depending on model.

Radioactive-source handling has specific safety requirements.

PET Detector Ring Does Not Physically Rotate Like CT

This is a useful distinction.

The PET detector ring generally surrounds the patient and remains stationary.

The events come from all around the patient.

CT Portion May Rotate

In a PET/CT scanner:

CT has a rotating X-ray acquisition system.

PET has a surrounding detector ring.

Two very different imaging systems share one platform.

PET/CT Gantry Contains Two Imaging Modalities

This means a complaint like:

Scanner image problem

needs to be narrowed.

Is it:

Table Position Matters

PET and CT need accurate spatial relationship.

If table position information is wrong:

Image fusion and registration can be affected.

Bed Deflection

Patient table loading can also influence alignment.

Manufacturers account for expected table behavior, but mechanical problems can matter.

Detector Electronics

A PET detector module may contain:

Module Communication Failure

If a detector block cannot communicate with the acquisition system:

Events from that region may be missing.

Does the Scanner Keep Scanning?

Depending on severity and design, the scanner may:

High Voltage on PMT Systems

Older PMT-based systems require stable high voltage.

If HV changes:

PMT gain changes.

That can alter:

and detector calibration.

SiPM Bias

SiPM systems also require carefully controlled bias voltage.

Small changes can affect gain.

Temperature Compensation

Because SiPM behavior varies with temperature, bias control may compensate based on detector temperature.

Cooling Failure Can Become Detector Failure

Again, cooling is part of the measurement system.

Not merely a comfort feature.

Real-World Example: Detector Block Failure

Daily QC shows a persistent region with reduced counts.

Repeat calibration does not correct it.

Detector diagnostics identify one block with abnormal gain.

Clinical images may show corresponding nonuniformity.

Real-World Example: Timing Calibration

TOF performance degrades.

Energy response appears normal.

Timing calibration identifies channel offset.

The detector still detects gamma photons, but its timing information is no longer accurate enough.

Real-World Example: PET/CT Misregistration

Fused image shows PET uptake shifted relative to CT anatomy.

PET and CT QC are individually acceptable.

Patient movement occurred between acquisitions.

This is not necessarily detector failure.

Real-World Example: Attenuation-Correction Artifact

Metal implant produces unusual CT attenuation values.

Corrected PET image shows corresponding artifact.

Reviewing non-attenuation-corrected PET data can help distinguish the effect.

Real-World Example: Cooling Problem

PET system passes QC when cold.

After extended operation:

Detector temperature rises and gain shifts.

System generates calibration or detector-stability warning.

Cooling, not the scintillator itself, may be the root problem.

Common Mistakes

Thinking PET Detects the Positron Directly

The scanner primarily detects the gamma photons produced after positron-electron annihilation.

Thinking One Coincidence Gives the Exact Decay Location

Without TOF, it provides a line of possible locations.

Even TOF narrows rather than perfectly identifies the point.

Assuming Every PET Image Artifact Is Detector Failure

Attenuation correction, motion, scatter, and CT misregistration can also matter.

Ignoring Temperature

Detector gain and timing can depend on thermal stability.

Treating PET and CT as One Imaging Chain

They are separate modalities that interact.

Ignoring Calibration Because the Scanner Still Produces Images

Quantitative accuracy requires stable calibration.

A Useful PET Detector Framework

Think:

Radiotracer

↓

Positron Emission

↓

Annihilation

↓

Two 511 keV Photons

↓

Scintillator

↓

Photodetector

↓

Energy + Timing Measurement

↓

Coincidence Detection

↓

Line of Response

↓

Corrections

↓

Reconstruction

↓

PET Image

Every step contributes to the final result.

Another Useful Troubleshooting Split

Ask whether the problem involves:

Tracer / Patient

Detection

Energy Measurement

Timing

Coincidence Processing

Attenuation Correction

Reconstruction

PET/CT Registration

That is much more precise than:

PET looks bad.

What Did You Actually Prove?

If every detector module communicates:

You proved:

The acquisition system currently recognizes those modules.

You did not prove:

If daily QC passes:

You have stronger evidence that the detector system meets the tested stability and calibration requirements under those conditions.

If PET and CT images are misaligned:

You proved:

The fused datasets do not currently line up correctly.

You have not yet proven whether the cause is:

Final Thoughts for Biomeds

PET is one of the best examples of how medical imaging turns statistics into anatomy and physiology.

The scanner never sees the positron.

It never directly watches a glucose molecule moving through the body.

Instead it detects pairs of gamma photons.

From those pairs, it builds lines of response.

From millions of those lines, it estimates where radioactive activity exists.

Then it corrects for:

and reconstructs an image that clinicians can use.

That means a PET problem can originate in:

The final image may look like one thing.

But underneath it is an enormous chain of measurements and corrections.

And that is exactly why understanding how the system works helps when something goes wrong.

Instead of saying:

The PET image looks weird,

you can start asking:

Is this detection, timing, correction, registration, or reconstruction?

That is a much better troubleshooting question.

And as always:

What did you actually prove?

— Jake

Important Note

PET detector materials, photodetectors, timing resolution, energy windows, coincidence processing, time-of-flight implementation, attenuation correction, calibration procedures, QC requirements, radioactive-source handling, and service boundaries vary significantly by manufacturer and scanner model. PET systems involve ionizing radiation and radioactive materials. Follow current OEM documentation, radiation-safety procedures, nuclear-medicine requirements, and authorized service scope when troubleshooting PET or PET/CT equipment.

Related Biomed Basics

Back to Biomed Basics