How probe movement becomes a training image

In a typical simulator architecture, a sensor is mounted in the mock probe and the system obtains its coordinates in a defined reference frame. Software relates those data to an anatomical model or prerecorded volume and displays the section corresponding to the virtual scanning plane.

Polhemus describes this relationship in its UltraSim training case: PATRIOT tracks the mock probe’s position and orientation relative to the manikin, and the simulator uses those data to change the ultrasound image. This is an architectural example, not evidence that every electromagnetic solution delivers the same accuracy.

The working chain is:

  1. a sensor and field generator, or cameras, acquire source measurements;
  2. the tracker calculates position and orientation;
  3. calibration relates tracker coordinates to the manikin and probe;
  4. the application calculates the imaging plane in the anatomical data;
  5. the renderer updates the image;
  6. the training scenario records actions and, where provided, assesses performance.

An error at any stage affects the training result. A fast tracker cannot compensate for an incorrect probe transform, and accurate coordinates cannot remove rendering latency.

Why full 6DoF is needed

Six degrees of freedom comprise X, Y and Z position plus orientation about three axes. Calculating the imaging plane requires more than a contact point: the application must know the scanning plane’s direction and probe rotation.

A system with an incomplete set of observed parameters can leave rotation about one axis unobserved. Whether that is acceptable depends on sensor geometry and the training task. If the scenario must distinguish rotation about the probe’s longitudinal axis, the architecture must measure that rotation or obtain it from another channel.

A useful data contract includes:

  • probe and sensor identifiers;
  • position in agreed units;
  • orientation as a quaternion or matrix with a documented axis convention;
  • timestamp and frame sequence number;
  • measurement-quality status;
  • calibration and coordinate-transform versions.

Without these conventions, two components can produce formally valid numbers in different reference frames.

What to coordinate in probe and manikin design

OEM integration starts with mechanics. Define the physical working plane of the probe, its relationship to the sensor coordinate frame, and the transform that maps the measurement into virtual anatomy.

Probe

Document mounting rigidity, sensor offset from the working plane, mass, balance, cabling and service access. Even a small loose joint changes the imaging-plane orientation regardless of tracker quality. Housing electronics and materials should be part of the test configuration from the outset.

Field generator

Its position determines the working volume and the influence of surrounding structures. Test not only the centre of the manikin but also scanning-zone boundaries, probe tilts and the trainee’s hand position. Generator placement cannot be selected solely for enclosure convenience.

Manikin

Metal parts, electronics, wiring and moving components can affect measurements. Record the exact manikin configuration and equipment layout in the protocol. Replacing one component may require revalidation.

Anatomical data and imaging plane

Define the coordinate origin, volume orientation, scale, permitted scanning regions and behaviour outside them. Separately verify that the image does not continue moving after the probe stops because of jitter or filtering.

Comparing probe-tracking approaches

Approach When it makes sense What deserves particular testing
Electromagnetic 6DoF The sensor must be embedded and direct visibility is difficult Field, working volume, probe and manikin effects, latency and quality statuses
Optical Cameras and markers can be installed and an external coordinate reference is needed Hand and body occlusion, probe markers and calibration to the manikin
Inertial with additional referencing Orientation matters more than independent global position, or a second channel is available Drift, position recovery and channel synchronisation
Hybrid One method cannot meet occlusion or referencing requirements Shared coordinates and time, switching logic and return to measurement

Electromagnetic tracking is often mechanically convenient for a compact training simulator, but no line-of-sight requirement does not eliminate environmental testing. Final architecture should follow measured requirements.

Errors to measure separately

The generic term “accuracy” hides different causes of an incorrect image. Include separate metrics in the protocol.

Stationary jitter. Hold the probe at several working points. Record coordinate and orientation variation as well as movement of the final imaging plane.

Position error. Compare coordinates with reference points at the centre and boundaries of the working volume. State the metric: mean, RMS, median, 95th percentile or maximum.

Orientation error. Test rotations about each axis and combined poses. A plane-direction error can be more visible than a small XYZ error.

Dynamic error. Perform slow scanning and typical working movements. A static test does not reveal tracking-lag errors or filter effects.

End-to-end latency. Measure the interval from probe movement to image update. 6DoF output rate and tracker latency describe only part of the chain.

Recovery. Leave the working volume or create a specified measurement interruption, then return. Check jumps, recovery time and the status exposed to the application.

Long-term stability. Record repeatability, heating and changes after a series of sessions. A short demonstration does not represent a simulator’s working day.

Test programme for the assembled simulator

  1. Freeze the configuration. Record sensor, generator, electronics, firmware, SDK, application, probe and manikin versions.
  2. Check mechanical registration. Measure the sensor and working-plane locations relative to the probe. Repeat after disassembly and reassembly.
  3. Create a reference-point grid. Cover the centre, edges and different heights of the actual scanning zone. Use an appropriate reference and document its uncertainty.
  4. Perform rotations. Test every axis, including longitudinal probe rotation where it affects the training image.
  5. Record slow scanning. Assess small movements, stops and image smoothness without excessive filter delay.
  6. Add real interference. Test standard electronics, cables, metal parts and nearby equipment in their operating state.
  7. Test boundaries and recovery. The application must not present stale data as a current measurement.
  8. Run an extended session. Repeat exercises while monitoring heating, interruptions and recalibration requirements.
  9. Compare with an independent channel. During development, use a parallel reference system with synchronised recording where appropriate.
  10. Document acceptance. Define acceptable values for each zone and mode before reviewing results.

Why results cannot be transferred automatically from one probe to another

A published assessment of an NDI Aurora compact field generator evaluated six different ultrasound probes with a standardised protocol. For two probes, mean distance error increased to 1.5 mm compared with 0.5 mm in the reference measurement; jitter remained below 0.2 mm in all cases, and mean relative orientation error was below 3°. These values apply only to the configuration described by the authors. They are not TAU specifications or universal values for every system. The practical conclusion is that each probe design needs its own assessment.

What an OEM customer should receive with the tracker

  • an agreed coordinate and mechanical-mounting scheme;
  • documented working volume and limitations;
  • a data contract with timestamps and quality statuses;
  • calibration and post-service verification procedures;
  • diagnostics for losses, out-of-volume operation and stale frames;
  • test results for the specific configuration;
  • revalidation requirements after probe, manikin or electronics changes.

This package reduces the risk that a successful tabletop demonstration will not repeat in a production simulator.

The TAU Med development direction

TAU Med is an R&D OEM direction for training ultrasound simulators. The prospective configuration is a compact tracker in which the sensor, field sources, processing and calibration are coordinated with the probe and manikin. Wound receiving sensors are a research branch requiring their own measurement chain and validation.

Comparisons with PATRIOT, Aurora and other solutions should use the same bench: position and orientation error, jitter, end-to-end latency, recovery, heating and integration effort. Numerical superiority is not claimed before those checks are complete.

For a technical discussion, provide the probe and manikin scheme, working volume, required movements and imaging-plane calculation. The general tracking-method selection criteria can help structure the comparison.

Frequently asked questions

Can an orientation-only sensor be used?

If the virtual imaging plane depends on probe position relative to the manikin, orientation alone is insufficient. Full 6DoF or another validated architecture is needed to reconstruct the required plane unambiguously.

Why not compare update rate alone?

Update rate does not show coordinate error, jitter or end-to-end latency. Training depends on how correctly and promptly the image changes during real scanning.

Is calibration needed after probe replacement?

If the mechanical or electromagnetic configuration changes, registration and performance should be rechecked. The required work depends on the product design and service procedure.

Is TAU Med a finished medical device?

No. The direction described here concerns tracking development for training ultrasound simulators. The status and characteristics of a specific delivery must be confirmed by separate project documentation.

Sources and further reading