m17 · Impedance (current-based) localization: patch fields, nonlinearity, field scaling
Education only. Not for diagnosis, treatment or any clinical decision.
In 60 seconds
Impedance localization turns every catheter electrode into a position sensor. Pairs of skin patches drive small high-frequency currents through the chest, separated by frequency or time, and the system reads the potential each pair makes at each electrode. If the field were uniform, position would be potential divided by field. It is not: the potential of a patch pair is \(k(1/|\mathbf r - \mathbf a| - 1/|\mathbf r - \mathbf b|)\), and along its axis \(2kx/(d^2 - x^2)\), so a uniform-field map stretches positions toward the patches by \(x^2/(d^2 - x^2)\). Real chests add more distortion where conductivity changes, because the field rises in proportion to the conductivity ratio. Field scaling uses electrodes of known spacing: each pair satisfies \(\ell^2 = \Delta\boldsymbol\phi^\top M \Delta\boldsymbol\phi\), which is linear in the local metric \(M\). Precision is voltage noise divided by field strength.
From the field
- Any electrode can be located. Regular catheter electrodes, used as sensors for a high-frequency transthoracic field applied through skin electrodes, gave positions that were highly linear with the true ones (correlation 0.996–0.999). Electrogram recordings were unaffected (Wittkampf, Wever, Derksen, Wilde, et al. 1999).
- Scaling is the weak point. The same work reported localization accuracy better than 2 mm, with an additional scaling error of 8% to 14% (Wittkampf, Wever, Derksen, Wilde, et al. 1999).
- Distances vary from chamber to chamber. A 54-mm electrode span on a decapolar catheter measured 101% ± 15%, 95% ± 10% and 97% ± 8% of its true length in the right atrium, right ventricle and left ventricle (Wittkampf, Wever, Derksen, Ramanna, et al. 1999).
Derivations
Symbols used throughout. \(I\) patch current; \(\sigma\) conductivity; \(k = I/(4\pi\sigma)\); \(\mathbf a\), \(\mathbf b\) patch positions; \(\phi\) potential; \(d\) patch half-separation; \(x\) position along the axis; \(E_0\) field at the center; \(\boldsymbol\phi\) the vector of three pairs’ potentials; \(J\) the local Jacobian \(\partial\boldsymbol\phi/\partial\mathbf x\); \(M\) the local metric; \(\ell\) known electrode spacing; \(\sigma_{\text{ref}}\), \(\sigma_{\text{local}}\) conductivities of a reference and a local region; \(\delta\phi\) voltage noise.
A patch pair
In a homogeneous medium, current \(I\) entering at \(\mathbf a\) and leaving at \(\mathbf b\) makes the potential (m02):
\[ \phi(\mathbf r) = k\left(\frac{1}{\lvert\mathbf r - \mathbf a\rvert} - \frac{1}{\lvert\mathbf r - \mathbf b\rvert}\right), \qquad k = \frac{I}{4\pi\sigma} \tag{1}\]
It is harmonic between the patches and odd about the plane midway between them. Real chests are bounded and inhomogeneous, which changes the numbers but not the shape of the argument.
The nonlinearity
On the axis between patches at \(\pm d\):
\[ \phi(x) = \frac{2kx}{d^2 - x^2}, \qquad E_0 = \frac{2k}{d^2}, \qquad \frac{\phi(x)/E_0 - x}{x} = \frac{x^2}{d^2 - x^2} \approx \frac{x^2}{d^2} \tag{2}\]
A system that assumes a uniform field reports positions too far out, increasingly so toward the patches: with \(d\) = 200 mm, by 1.0% at 20 mm and 6.7% at 50 mm (lab part A).
A linear map
Three orthogonal pairs, separated by frequency, give three potentials per electrode. The simplest map divides each by its central field:
\[ \hat x_i = \frac{\phi_i(\mathbf r)}{E_0}, \qquad i = 1, 2, 3 \tag{3}\]
It is exact only at the center. Over a cube of half-width 0.2\(d\) the RMS error was 1.8 mm, and over 0.4\(d\) it was 13.9 mm (lab part B).
Exact
Assumptions
- Model: point patches in an unbounded homogeneous medium, Equation 2, mapped by Equation 3. Tested against
epml.impedance. - Parameters: patch separation and probe position are illustrative choices.
- What it ignores: the torso’s boundary, lungs and other inhomogeneities, patch size and respiration.
Key takeaways
- A uniform-field map stretches positions toward the patches.
- The error grows as the square of the distance from the center relative to the patch separation.
Field scaling from known spacing
Catheter electrodes have known spacings. Locally \(\Delta\boldsymbol\phi = J\Delta\mathbf x\), so \(\Delta\mathbf x = J^{-1}\Delta\boldsymbol\phi\) and:
\[ \ell^2 = \lvert\Delta\mathbf x\rvert^2 = \Delta\boldsymbol\phi^\top M\,\Delta\boldsymbol\phi, \qquad M = (JJ^\top)^{-1} \tag{4}\]
Each electrode pair of known spacing gives one equation, linear in the six independent entries of the symmetric \(M\). Pairs in at least six independent orientations near one place determine it. In the lab, twelve orientations of a 4-mm bipole fix \(M\), and a 10-mm pair that the uniform-field map reported as 13.0 mm is measured as 10.0 mm (lab part C). This is one way to formalize field scaling; products do not publish their methods in this form.
Exact
Assumptions
- Model: two orthogonal point-patch pairs, Equation 1, in an unbounded homogeneous medium; the grid lies in their plane and is mapped by Equation 3. Tested against
epml.impedance. - Parameters: grid size and patch separation are illustrative choices.
- What it ignores: the third pair’s influence off the plane, the torso’s boundary, inhomogeneity and any correction.
Inhomogeneity
Across an interface between regions of different conductivity, normal current density is continuous (m02, eq-m02-interface), so the field, and with it the apparent distance, scales with the conductivity ratio:
\[ \frac{E_{\text{local}}}{E_{\text{ref}}} = \frac{\sigma_{\text{ref}}}{\sigma_{\text{local}}} \tag{5}\]
A 10-mm layer at half the reference conductivity appears 20 mm thick (lab part D). Lungs, the blood pool and the myocardium all differ in conductivity, which is why impedance geometry needs scaling, and why hybrid systems anchor it with magnetic sensors (m18).
Precision
Voltage noise becomes position noise through the field strength:
\[ \delta x = \frac{\delta\phi}{\lvert E\rvert}, \qquad \lvert E_0\rvert = \frac{2k}{d^2} \tag{6}\]
Closer patches or more current give a stronger field and less position noise, but closer patches also make the nonlinearity worse (Equation 2). With \(d\) = 150 mm the stretch at 60 mm is 19%; with 300 mm it is 4.2% (lab part E).
Key takeaways
- Known electrode spacings calibrate the local field.
- Conductivity changes stretch apparent distance.
Clinical correlate
This section describes published findings only. It is not guidance for patient care, and it has not yet been reviewed by an EP physician.
Where this shows up in EP. Impedance localization lets a mapping system show every electrode of every catheter, not just those with magnetic sensors (Wittkampf, Wever, Derksen, Wilde, et al. 1999). Its geometry depends on scaling: early work reported a scaling error of 8% to 14% (Wittkampf, Wever, Derksen, Wilde, et al. 1999), and measured inter-electrode distances varied by chamber (Wittkampf, Wever, Derksen, Ramanna, et al. 1999). Respiration and changes in the chest’s impedance during a procedure can shift positions over time (m19).
What the evidence is. Two clinical validation studies from the same group (Wittkampf, Wever, Derksen, Wilde, et al. 1999; Wittkampf, Wever, Derksen, Ramanna, et al. 1999).
What a reader should ask:
- Were positions scaled, and with what? (Equation 4)
- How far from the center of the field was the measurement? (Equation 2)
- Did the catheter cross regions of different conductivity? (Equation 5)
Visual summary
Schematic
Assumptions
- Schematic: not to scale, no quantitative claim. Labels follow Equation 1, Equation 2 and Equation 4.
How experts think
When judging impedance-based geometry, ask:
- Where is the region of interest relative to the patches? (Equation 2)
- What map turns potentials into positions? (Equation 3)
- How was it scaled, and from which electrode spacings? (Equation 4)
- Which conductivity boundaries lie nearby? (Equation 5)
- What is the position noise? (Equation 6)
One sentence to keep: potential is not position until the field is known, and electrodes of known spacing tell you the field.
Cheat sheet
| Quantity | Result |
|---|---|
| Patch pair (Equation 1) | \(k(1/\lvert\mathbf r - \mathbf a\rvert - 1/\lvert\mathbf r - \mathbf b\rvert)\), \(k = I/(4\pi\sigma)\) |
| Axis (Equation 2) | \(2kx/(d^2 - x^2)\); \(E_0 = 2k/d^2\); stretch \(x^2/(d^2 - x^2)\) |
| Linear map (Equation 3) | \(\hat x_i = \phi_i/E_0\) |
| Field scaling (Equation 4) | \(\ell^2 = \Delta\boldsymbol\phi^\top M\Delta\boldsymbol\phi\), \(M = (JJ^\top)^{-1}\) |
| Interface (Equation 5) | \(E_{\text{local}}/E_{\text{ref}} = \sigma_{\text{ref}}/\sigma_{\text{local}}\) |
| Precision (Equation 6) | \(\delta x = \delta\phi/\lvert E\rvert\) |
Values used in this module
| Value | Source |
|---|---|
| Correlation 0.996–0.999; accuracy better than 2 mm; scaling error 8–14% | Wittkampf, Wever, Derksen, Wilde, et al. (1999) |
| 54-mm span measured at 101% ± 15%, 95% ± 10%, 97% ± 8% | Wittkampf, Wever, Derksen, Ramanna, et al. (1999) |
| Patch separations, grid sizes, conductivity ratios, noise | illustrative |
| Stretch, map errors, field-scaled distances | our own code |
Interview questions
Try each question before you open its model answer. In a real interview, say which assumptions you are making.
Q1. How does impedance localization turn an electrode into a position sensor?
Model answer
Skin patch pairs drive high-frequency currents; the electrode’s potential from each pair, separated by frequency, depends on its position (Equation 1) (Wittkampf, Wever, Derksen, Wilde, et al. 1999).
Q2. Why are the currents at high frequency?
Model answer
To separate them from the electrograms and from each other by frequency, and so the electrograms are unaffected (Wittkampf, Wever, Derksen, Wilde, et al. 1999).
Q3. Derive the nonlinearity along a patch pair’s axis.
Model answer
\(\phi = k(1/(d - x) - 1/(d + x)) = 2kx/(d^2 - x^2)\); dividing by \(E_0 = 2k/d^2\) overshoots by \(x^2/(d^2 - x^2)\) (Equation 2).
Q4. Where is a uniform-field map most accurate?
Model answer
Near the center between the patches (Equation 3).
Q5. How can known electrode spacing calibrate the field?
Model answer
Each pair satisfies \(\ell^2 = \Delta\boldsymbol\phi^\top M\Delta\boldsymbol\phi\), linear in the metric \(M\); pairs in at least six orientations determine it (Equation 4).
Q6. Why does one straight catheter in one orientation not fix the metric?
Model answer
All its pairs point the same way, giving one direction’s scale only; the other five entries of \(M\) need other orientations (Equation 4).
Q7. Why do lungs distort impedance geometry?
Model answer
Normal current is continuous across conductivity boundaries, so the field and apparent distances change with the conductivity ratio (Equation 5).
Q8. What sets position noise?
Model answer
Voltage noise divided by field strength (Equation 6).
Q9. Trade-off in moving patches closer together?
Model answer
A stronger field and less noise, but more nonlinearity across the heart (Equation 2, Equation 6).
Q10. Impedance against magnetic localization: one strength and one weakness each.
Model answer
Impedance locates every electrode but needs scaling and can drift; magnetic localization measures distance directly but needs sensor coils and is disturbed by conductors (m16).
Q11. What did early validation report for scaling?
Model answer
An additional scaling error of 8% to 14% on top of better than 2 mm accuracy (Wittkampf, Wever, Derksen, Wilde, et al. 1999).
Q12. A map’s chamber looks 10% too large. What would you check?
Model answer
Whether field scaling was applied and from which catheter, where the chamber sits relative to the patches, and whether the patches or the chest impedance changed (Equation 2, Equation 4).
Code lab
The lab builds a small impedance localizer.
| Part | What you do | Data |
|---|---|---|
| A | A patch pair and its nonlinearity (Equation 1, Equation 2) | synthetic |
| B | Three pairs and a linear map (Equation 3) | synthetic |
| C | Field scaling from known electrode spacing (Equation 4) | synthetic |
| D | Conductivity changes stretch distance (Equation 5) | synthetic |
| E | Precision (Equation 6) | synthetic |
- Python notebook: labs/python/m17_lab.ipynb. Run it from the course repository after
pip install -r requirements.txt, or open it in Google Colab. - Data: none to download.
Each part ends with checks: lines that pass with the code as given. If you change something and a check fails, that is information, not an error to hide.
Self-check
Short checks you can do in your head. Open each answer after you have tried it.
S1. Patches at ±200 mm. Stretch of a uniform-field map at 40 mm?
Answer
\(1600/(40000 - 1600) \approx\) 4.2% (Equation 2).
S2. Patch separation doubles. Central field?
Answer
One quarter (Equation 2).
S3. How many independent pair orientations to determine the local metric?
Answer
Six (Equation 4).
S4. A layer has one third the reference conductivity. Apparent thickness of a 5 mm layer?
Answer
15 mm (Equation 5).
S5. Voltage noise doubles. Position noise?
Answer
Doubles (Equation 6).
S6. A uniform-field map reports a 10 mm pair as 13 mm. Is the region nearer the patches or the center?
Answer
Nearer the patches, where the field is stronger (Equation 2).
Literature figures
These figures are worth looking at alongside this module. We link to them instead of copying them: some licenses do not allow reuse here, and others we could not confirm. Each entry says what to look for and where this module explains it.
- Measured against true electrode positions. Wittkampf and colleagues’ linearity results in three chambers (Wittkampf, Wever, Derksen, Wilde, et al. 1999). This module: the nonlinearity and field scaling.
- Distances during ablation procedures. The same group’s measured inter-electrode distances by chamber and procedure (Wittkampf, Wever, Derksen, Ramanna, et al. 1999). This module: field scaling.
Further reading
- The technique. Real-time localization of regular electrodes (Wittkampf, Wever, Derksen, Wilde, et al. 1999); accuracy during ablation (Wittkampf, Wever, Derksen, Ramanna, et al. 1999).
- Related modules. m02 (volume conduction), m16 (magnetic localization), m18 (hybrid localization), m19 (motion and drift), m20 (geometry).