Fig 1.
Breast prosthesis with 12 implanted catheters fixed to the chest of a male test person.
To monitor the breathing motion, three fiducial sensors are fixed to the breast.
Table 1.
Summary of all patient data and treatment plan variables for all measurements conducted in this study.
Fig 2.
Flowchart of the algorithm employed in the artifact removal step of the data processing chain.
Fig 3.
The measured raw breathing signal is colored in black, and in grey the first component of the SSA is depickted.
Fig 4.
The measured raw sensor signal is drawn in black, and a grey line illustrates the signal resulting after outlier removal.
Fig 5.
Shown is the spatial sensor position along one coordinate axis versus discrete time n ⋅ Ts with Ts the sampling interval and n = 1, …, T.
In the left corner of the top row, the original signal is exposed. The other boxes show the IMFs resulting from an EEMD. On the x-axis of each graph, the number of samples, and on the y-axis, the signal amplitude is plotted. The first IMF comes aside of the original signal, and subsequent IMFs follow in each row from left two right.
Fig 6.
Shown is the spatial fiducial sensor position along one coordinate axis versus discrete time.
In the left corner of the top row, the original signal is exposed. The other boxes show the IMFs resulting from an EEMD. On the x-axis of each graph, the number of samples, and on the y-axis, the signal amplitude is plotted. The first IMF comes aside of the original signal, and subsequent IMFs follow in each row from left two right.
Fig 7.
The originally measured EMT signal from one catheter (black), and its reconstructed counterpart (grey) are shown to illustrate the efficiency of the removal of the breathing mode contamination.
Fig 8.
Three dimensional dwell positions from the treatment plan (crosses) and the reconstructed EMT measured dwell positions, colored according to the absolute deviation as given in the colorbar.
Fig 9.
Absolute deviation for each pair of dwell positions (CT-EMT), for four different reconstructed measurements.
The black vertical lines mark the separation between the different catheters.
Fig 10.
Top: Illustration of a perfect reconstruction of the catheter shapes from EMT measurements of a patient immediately after the CT: 3D dwell positions from the treatment plan (crosses), 3D reconstructed EMT dwell positions, colored according to their absolute deviations from the treatment plan. Middle: Illustration of a clear catheter shift revealed by EMT data reconstructed according to the signal processing chain proposed in this work. The data stem from the same patient after it has been moved to the EMT treatment room. Crosses signify 3D dwell positions from the treatment plan, while 3D reconstructed EMT dwell positions are drawn as squares colored according to the absolute deviations of the dwell positions from the treatment plan. Bottom: Colorbar.
Fig 11.
Absolute deviations for each pair of dwell positions (CT-EMT), are shown for four different reconstructed measurements.
The black vertical lines mark the catheters.
Fig 12.
Three dimensional dwell positions from the treatment plan (crosses) and the reconstructed EMT measured dwell positions (colored according to the absolute deviation—See Fig 8 for the color bar).
Reconstructed with the Jensen-Shannon divergence.
Table 2.
Collection of all mean deviations, measured in mm, for all datasets as obtained with four different similarity measures.