Limits...
Fast and accurate localization of multiple RF markers for tracking in MRI-guided interventions.

Galassi F, Brujic D, Rea M, Lambert N, Desouza N, Ristic M - MAGMA (2014)

Bottom Line: Computational complexity was significantly reduced by avoiding cluster analysis, while higher accuracy was achieved by using optimal projections and by applying Gaussian interpolation in peak detection.The computational time for 6 markers was better than 2 ms, an improvement of up to 100 times, compared to the method by Flask et al. (J Magn Reson Imaging 14(5):617-627, 2001).The proposed method is particularly suitable in systems requiring any of the following: high frame rate, tracking of three or more markers, data filtering or interleaving.

View Article: PubMed Central - PubMed

Affiliation: Mechanical Engineering Department, Imperial College London, London, UK, f.galassi09@imperial.ac.uk.

ABSTRACT

Object: A new method for 3D localization of N fiducial markers from 1D projections is presented and analysed. It applies to semi-active markers and active markers using a single receiver channel.

Materials and methods: The novel algorithm computes candidate points using peaks in three optimally selected projections and removes fictitious points by verifying detected peaks in additional projections. Computational complexity was significantly reduced by avoiding cluster analysis, while higher accuracy was achieved by using optimal projections and by applying Gaussian interpolation in peak detection. Computational time, accuracy and robustness were analysed through Monte Carlo simulations and experiments. The method was employed in a prototype MRI guided prostate biopsy system and used in preclinical experiments.

Results: The computational time for 6 markers was better than 2 ms, an improvement of up to 100 times, compared to the method by Flask et al. (J Magn Reson Imaging 14(5):617-627, 2001). Experimental maximum localization error was lower than 0.3 mm; standard deviation was 0.06 mm. Targeting error was about 1 mm. Tracking update rate was about 10 Hz.

Conclusion: The proposed method is particularly suitable in systems requiring any of the following: high frame rate, tracking of three or more markers, data filtering or interleaving.

Show MeSH
A number of events from the known distribution expected in each bin and a number of events observed in each bin
© Copyright Policy - OpenAccess
Related In: Results  -  Collection


getmorefigures.php?uid=PMC4315881&req=5

Fig9: A number of events from the known distribution expected in each bin and a number of events observed in each bin

Mentions: Using the position of the pixel with the highest intensity for peak localization, would lead to a maximum error of 0.5 pixel, or 0.6 mm in this work. By using Gaussian interpolation, the maximum peak localization error was reduced to 0.283 mm (Table 4). The Chi square goodness-of-fit test showed that the deviation of the peak positions over repeated acquisition is normally distributed (Fig. 9). The standard deviation varied between 0.03 and 0.075 mm. In the experiments with a volunteer in the scanner the standard deviation reached 0.08 mm.Fig. 9


Fast and accurate localization of multiple RF markers for tracking in MRI-guided interventions.

Galassi F, Brujic D, Rea M, Lambert N, Desouza N, Ristic M - MAGMA (2014)

A number of events from the known distribution expected in each bin and a number of events observed in each bin
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC4315881&req=5

Fig9: A number of events from the known distribution expected in each bin and a number of events observed in each bin
Mentions: Using the position of the pixel with the highest intensity for peak localization, would lead to a maximum error of 0.5 pixel, or 0.6 mm in this work. By using Gaussian interpolation, the maximum peak localization error was reduced to 0.283 mm (Table 4). The Chi square goodness-of-fit test showed that the deviation of the peak positions over repeated acquisition is normally distributed (Fig. 9). The standard deviation varied between 0.03 and 0.075 mm. In the experiments with a volunteer in the scanner the standard deviation reached 0.08 mm.Fig. 9

Bottom Line: Computational complexity was significantly reduced by avoiding cluster analysis, while higher accuracy was achieved by using optimal projections and by applying Gaussian interpolation in peak detection.The computational time for 6 markers was better than 2 ms, an improvement of up to 100 times, compared to the method by Flask et al. (J Magn Reson Imaging 14(5):617-627, 2001).The proposed method is particularly suitable in systems requiring any of the following: high frame rate, tracking of three or more markers, data filtering or interleaving.

View Article: PubMed Central - PubMed

Affiliation: Mechanical Engineering Department, Imperial College London, London, UK, f.galassi09@imperial.ac.uk.

ABSTRACT

Object: A new method for 3D localization of N fiducial markers from 1D projections is presented and analysed. It applies to semi-active markers and active markers using a single receiver channel.

Materials and methods: The novel algorithm computes candidate points using peaks in three optimally selected projections and removes fictitious points by verifying detected peaks in additional projections. Computational complexity was significantly reduced by avoiding cluster analysis, while higher accuracy was achieved by using optimal projections and by applying Gaussian interpolation in peak detection. Computational time, accuracy and robustness were analysed through Monte Carlo simulations and experiments. The method was employed in a prototype MRI guided prostate biopsy system and used in preclinical experiments.

Results: The computational time for 6 markers was better than 2 ms, an improvement of up to 100 times, compared to the method by Flask et al. (J Magn Reson Imaging 14(5):617-627, 2001). Experimental maximum localization error was lower than 0.3 mm; standard deviation was 0.06 mm. Targeting error was about 1 mm. Tracking update rate was about 10 Hz.

Conclusion: The proposed method is particularly suitable in systems requiring any of the following: high frame rate, tracking of three or more markers, data filtering or interleaving.

Show MeSH