Respiratory motion compensation with tracked internal and external sensors during CT guided procedures

J. Borgert, Sascha Krüger, Holger Timinger, Jochen Krücker, Neil Glossop, A. Durrani, Bradford J. WoodInternational Congress Series 2005 · DOI

TL;DR

This paper presents a novel method for compensating for respiratory motion during CT-guided interventions by using an external sensor on the patient's sternum to predict the motion of an internal sensor, like a biopsy needle. This approach significantly reduces residual motion, improving the accuracy and safety of minimally invasive procedures.

What Problem Does This Paper Address?

Image-guided minimally invasive procedures, such as biopsies and ablations, are increasingly used for treating tumors in organs like the liver. A major challenge in these procedures is the involuntary movement of organs due to the patient's breathing. This respiratory motion can cause a significant discrepancy between the pre-procedural imaging (e.g., CT scans) and the actual position of the target lesion during the procedure. This discrepancy can lead to inaccurate targeting, increased risk of complications, and reduced procedure effectiveness. While various methods exist to address this, such as breath-holding or external monitoring devices, they often have limitations like patient discomfort or limited accuracy. This paper aims to address this by developing a more robust and integrated system for respiratory motion compensation using electromagnetic tracking.

Methods

The core of this research lies in using an electromagnetic tracking system (MTS) to simultaneously track two sensors: one placed externally on the patient's sternum and another, the "internal sensor," incorporated into a biopsy needle. The sternum sensor acts as an external reference for respiratory motion, while the needle sensor tracks the actual movement within the organ.

The key methodological steps are:

  1. Data Acquisition: Position data from both sensors were collected during a liver biopsy procedure.
  2. Correlation Analysis: The researchers analyzed the correlation between the motion of the external sternum sensor and the internal needle sensor. This was done using a correlation matrix analysis, which quantifies the linear relationship between the coordinate movements of the two sensors. They specifically looked at the percentage correlation, defined as the sum of the absolute values of the correlation matrix entries.
  3. Affine Motion Model Derivation: Based on the high correlation observed, an affine motion model was derived. This model mathematically describes how the displacement of the sternum sensor can be transformed to predict the displacement of the liver sensor. An affine transformation is a linear transformation followed by a translation, which is suitable for representing rigid body motion with some scaling and rotation. The model was derived by minimizing the difference between the predicted and actual liver sensor displacements.
  4. Motion Compensation: The derived affine motion model was then applied to the acquired data to compensate for the respiratory motion of the internal sensor. The compensated position of the internal sensor was calculated by subtracting the predicted respiratory displacement from its measured position.
  5. Evaluation: The effectiveness of the compensation was evaluated by measuring the residual displacement of the internal sensor after applying the motion model, comparing it to the uncompensated displacement.

Datasets

The study utilized data collected during a single tracked biopsy of a 15 mm hepatocellular carcinoma in a sedated, unventilated 61-year-old male patient. The data acquisition involved tracking the needle and sternum sensors at a rate of 32.5 samples per second, totaling 720 seconds of tracking data. For detailed analysis, two specific ranges of data with steady respiration, labeled "datal" and "data2," were selected. These datasets were used for both deriving the affine motion model and evaluating its performance. No new benchmarks were introduced; the focus was on analyzing data from a real clinical procedure.

Key Insights and Ideas

  • External Sensor as a Reliable Proxy: The most significant insight is that an externally placed sensor on the sternum can serve as a highly reliable proxy for predicting the complex respiratory motion of internal organs, particularly the liver. This bypasses the need for more invasive or complex internal motion tracking methods.
  • High Correlation Under Steady Respiration: The study demonstrated a remarkably high correlation (up to 94%) between the sternum and liver sensor movements during periods of steady respiration. This strong correlation is the foundation upon which the motion compensation model is built.
  • Affine Model Sufficiency: For compensating respiratory motion, a relatively simple affine motion model proved to be effective. This suggests that for many interventional scenarios, complex biomechanical models might not be necessary, simplifying implementation.
  • Detecting Needle Manipulation: The compensation method not only corrected for respiratory motion but also allowed for the qualitative and quantitative detection of needle manipulation. This is because needle movements introduce displacements that deviate from the predictable respiratory pattern, which the model can highlight.

Key Results

  • Correlation: The overall correlation between the sternum and liver sensors across all data was approximately 77.51%. For selected ranges of steady respiration ("datal" and "data2"), this correlation increased significantly to 90.54% and 94.61%, respectively.
  • Residual Displacement Reduction: Without motion compensation, the mean residual displacement of the liver sensor was around 3.90-4.18 mm. After applying the affine motion model, this residual displacement was reduced by approximately a factor of four, to about 0.99 ± 0.66 mm for "datal" and 0.94 ± 0.46 mm for the initial part of "data2."
  • Comparison to Free-Form Deformation: The results of the affine motion compensation were comparable to, and in some cases better than, a more complex free-form deformation model, achieving a residual displacement of 1.04 ± 0.91 mm for "datal" and 1.99 ± 1.09 mm for "data2."
  • Needle Manipulation Impact: During needle manipulation in "data2," the residual displacement increased to 3.18 ± 0.48 mm, but subtracting this extra displacement showed results comparable to the non-manipulation cases, demonstrating the model's ability to isolate respiratory motion.

Limitations and Future Work

  • Single Patient Study: The study was conducted on a single patient, which limits the generalizability of the findings. Further validation with a larger patient cohort is necessary.
  • Sedated, Unventilated Patient: The patient was sedated but not ventilated. The behavior of respiratory motion might differ in ventilated patients or those with different breathing patterns.
  • Limited Needle Manipulation Analysis: While needle manipulation was detected, the analysis of its impact on compensation was somewhat limited to specific segments of the data.
  • Future Work: The authors suggest that future research should focus on proving the general applicability of this finding in further studies. They also propose exploring more sophisticated models for motion compensation, such as elastic deformable dynamic models, to account for tissue warping that might occur with applied torque during needle manipulations. The integration of such compensation models into clinical practice for electromagnetic tracking of medical devices is highlighted as a crucial next step for effective and conservative cancer treatment.

Summary updated

BibTeX
@inproceedings{borgert2005respiratory,
  author = {J. Borgert and Sascha Krüger and Holger Timinger and Jochen Krücker and Neil Glossop and A. Durrani and Bradford J. Wood},
  title = {{Respiratory motion compensation with tracked internal and external sensors during CT guided procedures}},
  booktitle = {International Congress Series},
  year = {2005},
  doi = {10.1016/j.ics.2005.03.125},
}
Figure