Alicia sees a moving ultrasound image of a heart and calls it a video taken through the chest. Tricia points to the transducer. Kai Kai asks what the machine actually sends into the body and what returns. The image is useful only because a chain of physical assumptions connects the returning signal to the anatomy being displayed.
Echocardiography works by sending ultrasound into the body, detecting returning echoes and Doppler frequency shifts, and reconstructing those signals into images and flow-related measurements. It can estimate chamber size, wall motion, valve motion, blood velocity and derived quantities such as stroke volume and ejection fraction. Every result depends on acquisition geometry, timing, signal quality and the model used to transform sound into a physiological quantity.
This article supports How the Heart Works. How Heart Chambers Work owns chamber geometry; How Heart Valves Work owns valve mechanics. Here the reader job is measurement: how those structures become evidence on an ultrasound system.
This is educational imaging physics and physiology, not a guide to interpreting an individual’s echocardiogram or deciding whether a measurement is normal.
1. Ultrasound begins as mechanical waves, not light
An echocardiography transducer produces high-frequency sound waves. These waves travel through tissue, encounter boundaries and structures with different acoustic properties, and partly reflect back toward the transducer.
The National Institute of Biomedical Imaging and Bioengineering explains the pulse-echo principle used in medical ultrasound. The machine measures returning signals rather than seeing the heart with visible light.
The final grayscale image is therefore a reconstruction. Brightness, position and motion correspond to processed echo information, not a direct optical photograph.
2. Echo timing provides depth information
Ultrasound travels through soft tissue at an approximately known speed. If a pulse is emitted and an echo returns after a measured interval, the machine can estimate how far away the reflecting structure lies.
The pulse travels to the structure and back, so the one-way distance is approximately half the total travel distance. This is why depth is derived from elapsed time rather than from the echo’s brightness.
The calculation depends on assuming an appropriate sound speed in tissue. If the path includes materials with different propagation properties, small positioning errors can arise. Reconstruction is therefore model-based even before the image is displayed.
3. Echo strength helps build the grayscale picture
Different tissue interfaces reflect different fractions of the incident sound. Strong returning echoes are displayed differently from weak echoes, producing the familiar grayscale structure of an ultrasound image.
Blood within a chamber often appears relatively dark compared with tissue boundaries because of the way red blood cells and larger structures scatter sound. The exact appearance depends on frequency, gain and processing.
A darker region should therefore not be treated as literal emptiness. It can represent blood that produces relatively little returning signal under the selected imaging settings.
4. Two-dimensional imaging is a moving slice
Conventional 2D echocardiography displays a plane through the heart. Because the heart is three-dimensional, the image is a slice rather than the complete organ.
Different transducer positions produce different standard views. An apical four-chamber view, for example, is designed to include both atria and both ventricles in one plane. A parasternal long-axis view reveals a different cross-section.
The existence of multiple views is not redundancy. Each view exposes structures and dimensions that another can miss or foreshorten.
5. Foreshortening changes geometry before any calculation begins
If an apical view misses the true apex and cuts through the ventricle at a shorter angle, the chamber appears shorter than it really is. This is foreshortening.
A perfectly drawn endocardial contour on a foreshortened image can still underestimate true chamber volume. The problem occurred during acquisition, not during tracing.
This illustrates a powerful measurement rule: precision applied to the wrong view does not create accuracy. The acquisition geometry must be correct before the calculation can be trusted.
6. M-mode trades spatial width for temporal detail
M-mode samples motion along a selected ultrasound line and displays depth against time. Because the system focuses on one line repeatedly, temporal resolution can be very high.
This makes M-mode useful for examining rapid motion of structures such as valve leaflets or ventricular walls when the sampling line is appropriately positioned.
The trade-off is spatial coverage. A one-dimensional line cannot represent the full three-dimensional chamber, so measurements made from it depend strongly on correct alignment and geometric assumptions.
7. Chamber-volume estimation requires geometry
Volume cannot be measured directly from one length unless the chamber shape is assumed. Echocardiography therefore combines multiple dimensions or contours to estimate three-dimensional volume.
The American Society of Echocardiography’s chamber-quantification guidance recommends the biplane method of disks for routine two-dimensional left-ventricular volume measurement. The ventricle is divided conceptually into many small disks whose volumes are summed.
The method reduces reliance on one simple global shape assumption, but it still depends on obtaining suitable apical views and tracing the endocardial border correctly.
8. The biplane method is a numerical reconstruction
Imagine the ventricle cut into a stack of thin disks. If the area of each disk can be estimated from two orthogonal views and multiplied by its thickness, summing the disks approximates total volume.
The image does not contain literal disks. The disks are a mathematical device for converting contours into volume.
This is an example of how measurements are constructed from a model. The better the views represent the true chamber, the more defensible the reconstruction.
9. End-diastole and end-systole must be defined consistently
To calculate stroke volume from chamber volumes, the end-diastolic and end-systolic frames must represent appropriate phases of the same physiological sequence.
ASE guidance describes timing conventions using valve closure and maximal or minimal chamber dimensions. The important principle is consistency: the two volumes must belong to compatible cardiac cycles and clearly defined phases.
Subtracting precisely measured volumes from mismatched beats can produce an arithmetically correct but physiologically misleading stroke volume.
10. Ejection fraction is derived from two volume estimates
Left-ventricular ejection fraction is calculated as end-diastolic volume minus end-systolic volume, divided by end-diastolic volume.
If a model ventricle measures 140 mL at end-diastole and 60 at end-systole, stroke volume is 80 mL and ejection fraction is 80/140, about 57%.
The fraction inherits uncertainty from both volume measurements. A tracing or foreshortening error can therefore propagate into the final ratio. The clean percentage display should not hide the measurement chain behind it.
11. A change in ejection fraction need not equal a change in forward output
Ejection fraction is a ratio of cavity-volume change to starting volume. Cardiac output also depends on heart rate, and useful forward flow can depend on valve competence.
Two ventricles can have different ejection fractions while ejecting similar absolute stroke volumes if their end-diastolic volumes differ.
This is why imaging metrics must remain attached to their definitions. A ratio can be informative without summarising every aspect of pump performance.
12. Doppler measures motion along the ultrasound beam
When ultrasound reflects from moving blood cells, the returning frequency differs slightly from the transmitted frequency. The Doppler shift can be used to estimate velocity along the beam direction.
If blood moves directly toward or away from the transducer, the measured component is large. If flow is nearly perpendicular to the beam, the measured component becomes small even when true speed is high.
Angle therefore matters. Doppler does not magically measure the full velocity vector independent of beam alignment.
13. Pulsed-wave and continuous-wave Doppler solve different sampling problems
Pulsed-wave Doppler sends and receives bursts so velocity can be assigned to a selected depth range. This provides spatial localisation but has limits on the highest unambiguous velocity it can measure.
Continuous-wave Doppler transmits and receives continuously. It can measure very high velocities but does not isolate one precise depth along the beam.
The methods therefore trade spatial localisation against velocity range. Choosing one depends on the question being asked.
14. Aliasing is a sampling problem, not impossible blood speed
With pulsed Doppler, velocity is sampled at discrete intervals. If the Doppler shift exceeds the Nyquist limit for the selected pulse-repetition frequency, the display wraps around. This is aliasing.
The blood has not reversed direction merely because the colour or spectral trace appears to wrap. The measurement system has exceeded its unambiguous sampling range.
This is a good example of instrument behaviour imitating physiology. Recognising the sampling limit prevents a display artifact from being mistaken for a biological event.
15. Colour Doppler is a coded velocity display, not an oxygen map
Colour Doppler assigns colours to estimated flow direction and velocity relative to the transducer. The exact colour convention depends on the map selected.
Red therefore does not mean oxygen-rich blood and blue does not mean oxygen-poor blood. That colour convention belongs to many anatomy diagrams, not Doppler physics.
A single colour image can also hide velocity distributions within a pixel region. It is a compressed visual summary of measured Doppler information.
16. Doppler velocity can support flow calculations only when geometry is known
Volume flow depends on velocity and cross-sectional area. In a suitable model, stroke volume can be estimated by multiplying a flow-area estimate by the velocity-time integral through that area.
Suppose a model left-ventricular outflow tract area is 3.0 cm² and its velocity-time integral is 22 cm. Their product is 66 cm³, or 66 mL per beat.
The calculation is only as good as the matching of area and velocity measurements. Using an area from one location and velocity from another breaks the physical boundary of the model.
17. Diameter error becomes larger when area is calculated
When cross-sectional area is calculated from a circular diameter, area is proportional to diameter squared. Small diameter errors therefore become larger area errors.
If diameter is overestimated by 10%, calculated area becomes 1.1 squared, or 1.21 times the true area: a 21% overestimate before velocity uncertainty is considered.
This is not a flaw unique to echocardiography. It is a mathematical consequence of deriving area from a squared length.
18. Doppler can help estimate pressure differences under defined assumptions
High blood velocity through a narrowed or accelerated region can be related to pressure differences using simplified Bernoulli reasoning. In clinical echocardiography, a commonly used approximation expresses pressure difference in terms of velocity squared.
The usefulness depends on conditions and assumptions about upstream velocity, viscous losses and flow geometry. A derived pressure gradient is not the same thing as a directly inserted pressure sensor.
The distinction between direct and derived measurement matters because different errors enter at different stages of the inference chain.
19. Valve timing can be observed in several echo modes
2D imaging shows leaflet motion. M-mode can give high temporal detail along a selected line. Doppler reveals the onset and cessation of blood movement through the valve.
These measurements do not all show exactly the same event. A leaflet beginning to move, a pressure crossover and measurable forward flow occur within one transition but are not identical timestamps.
Combining modes can therefore refine the mechanical interpretation while also exposing why a simplified open/closed animation is an approximation.
20. Wall motion can be measured regionally, not only globally
Different ventricular regions can be examined for timing and magnitude of motion. A global volume or ejection fraction compresses the chamber into one summary, whereas regional imaging retains spatial information.
Two hearts could have a similar global ejection fraction while distributing motion differently across the wall. The global value would not reveal that spatial pattern.
This is another instance of aggregation deleting structure. Averages are useful when the question is global; they are incomplete when the question is regional.
21. Strain describes deformation, not force directly
Myocardial strain describes relative deformation of tissue. Longitudinal strain, for example, reflects the fractional change in myocardial length along the long axis.
Speckle-tracking echocardiography estimates tissue motion by following natural acoustic patterns through sequential frames. The resulting deformation measure can reveal information not contained in ejection fraction alone.
Strain is not direct muscle force. Load, geometry and image quality influence the measured deformation. Keep the physical definition attached to the result.
22. Three-dimensional echocardiography changes the geometry problem
3D echocardiography acquires volumetric datasets rather than reconstructing volume only from selected 2D planes. This can reduce some geometric assumptions and foreshortening problems when image quality is good.
It does not eliminate every limitation. Temporal resolution, stitching, boundary detection and acoustic windows still matter.
The technology therefore changes the uncertainty structure rather than removing uncertainty altogether.
23. Image quality depends on the acoustic window
Ultrasound must travel from the transducer to the heart and back. Bone and air strongly affect transmission, so echocardiographic views are obtained through anatomical windows where sound can reach the heart more effectively.
Body habitus, lung position and other anatomical factors can change the available window. Poor image quality is therefore sometimes a physical-access problem rather than operator carelessness or machine failure.
Measurement uncertainty begins before any contour is drawn. The pathway carrying the signal matters.
24. Artifacts can create structures that are not really there
Ultrasound reconstruction assumes particular paths and interactions for sound. Multiple reflections, refraction, side lobes and other physical effects can generate image features that do not correspond straightforwardly to anatomy.
The American Society of Echocardiography published dedicated guidance in 2026 on identifying and mitigating cardiac ultrasound artifacts, emphasising that artifacts occur across 2D, spectral Doppler, colour Doppler and 3D imaging.
The important scientific habit is to ask whether a feature behaves consistently across views and modes before treating it as a real structure or flow event.
25. Gain and processing can alter appearance without altering anatomy
Receiver gain changes how strongly returning signals are displayed. Too little gain can make boundaries disappear; too much can fill dark regions with noise and obscure interfaces.
Dynamic range, filtering and post-processing also alter image appearance. These settings do not change the underlying anatomy, but they can change how easily the anatomy can be identified.
This is why visual clarity and physical truth are not identical. A prettier image is useful only if the processing preserves the features needed for the measurement.
26. Repetition reduces some uncertainty but not systematic bias
If repeated chamber tracings vary randomly around the true boundary, averaging can reduce some uncertainty. If every acquisition is consistently foreshortened, repeating the same view many times preserves the bias.
Precision and accuracy therefore answer different questions. A tightly repeatable wrong geometry can be precise and inaccurate.
Quality improvement requires locating the error source: acquisition, signal processing, tracing, timing, geometric model or physiological variability.
27. Beat selection matters when the rhythm varies
When filling and stroke volume vary from beat to beat, measurements from one cycle may not represent the average circulation. Beat selection and averaging strategy become part of the method.
Combining end-diastolic volume from one unusually full beat with end-systolic volume from another unusually small beat would create a fictional stroke volume that never occurred.
Temporal matching is therefore as important as spatial tracing. Measurement belongs to a particular beat, not an abstract permanent heart.
28. Echo measures consequences as well as causes
An echocardiogram can show a chamber becoming smaller during systole. That is a mechanical consequence of contraction, not direct observation of intracellular calcium or molecular cross-bridge cycling.
A high flow velocity can reveal a haemodynamic condition without showing the exact microscopic tissue property that produced it. A pressure-gradient estimate can constrain explanations without proving one cause.
Good interpretation therefore distinguishes the measured phenomenon from the mechanism inferred to explain it.
29. Worked problem: a diameter error propagates into stroke volume
A model outflow diameter is truly 2.0 cm, but it is measured as 2.2 cm. Because area depends on diameter squared, the calculated area is 1.21 times too large.
If the measured velocity-time integral is otherwise perfect, the derived stroke volume is also overestimated by about 21% in this simplified model.
The arithmetic shows why one apparently small linear measurement can dominate the uncertainty of a derived flow calculation.
30. Worked problem: same ejection fraction, different output
Model A has end-diastolic volume 120 mL and ejection fraction 60%, so stroke volume is 72 mL. At 60 beats per minute its output is 4.32 L/min.
Model B has the same ejection fraction and end-diastolic volume but beats at 90 per minute. Its calculated output is 6.48 L/min.
Equal ejection fraction does not imply equal flow per minute. The imaging ratio and the rate belong to different parts of the cardiovascular description.
31. Worked problem: a flat image cannot prove a three-dimensional volume
Suppose two different three-dimensional chambers produce the same cross-sectional area in one selected plane. Can their total volumes be concluded equal? No.
They may differ in length or shape outside that plane. A single section deletes information in the dimension perpendicular to it.
This is why multiple views or volumetric imaging are used for chamber quantification. Geometry is not an optional afterthought; it is the bridge from image to volume.
32. The echocardiography mechanism in one chain
The transducer emits ultrasound. Sound travels through tissue and reflects from acoustic interfaces. Return time helps assign depth; return strength helps create grayscale structure. Repeated pulses create moving 2D or 3D images. Doppler shifts estimate motion along the beam. Contours and timing definitions convert images into chamber volumes. Geometry and velocity can be combined into flow estimates. Derived quantities such as ejection fraction or pressure gradient inherit the uncertainty of the measurements beneath them.
Alicia no longer calls the image a camera video. Tricia labels each transformation. Kai Kai circles the assumptions that could fail. The machine has become a measurement chain rather than a black box.
The deeper lesson is that echocardiography is powerful because several independent physical signals can be aligned with the cardiac cycle. Its limits are equally informative: every claim should remain attached to the signal, geometry and model that support it.
Evidence trail and connected reading
For ultrasound physics, begin with NIBIB: Ultrasound. For standard chamber measurements, use the ASE chamber-quantification guidance. For Doppler measurement principles, see ASE Doppler Echo Quantification. For the current artifact problem, see ASE’s 2026 guidance announcement on cardiac ultrasound artifacts.
Return to the parent: How the Heart Works. Continue to How Heart Chambers Work for chamber geometry, How Heart Valves Work for boundary mechanics and How Cardiac Output Works for flow calculations.