**4. Discussion of Obtained Results**

For 3D MR scanning of the vocal tract [1,8] during phonation, the frequency range of 25 Hz to 3.5 kHz covering pitch and formant frequencies must be included in the recordings. A preliminary analysis of the properties of the vibration sensors suitable for measurement in the low magnetic field environment confirmed a general presumption of inverse relationship between the diameter of the

used sensor and the minimum frequency of vibration picked up from the measured surface. At the same time, the maximum frequency, as well as the sensitivity, were decreased in the massive aluminum microphone capsule of the phonocardiographic sensor HM692—its frequency response in Figure 1b shows a local maximum at about 50 Hz, which makes it useless for our purpose. The piezo film vibration sensor SDT1 has the best sensitivity but its frequency response shows higher non-linearity in comparison with the sensor SB-1. In addition, the vibration sensor SB-1 has the highest low frequency sensitivity, so it was finally chosen for all next recording and measurement experiments.

The acoustic noise intensity was measured at several distances from the center of the scanning area, beginning with 45 cm because, for shorter distances, the magnetic field inhomogeneity due to interaction with metal parts of the sound level meter causes the display of a warning message on the MRI control console and interruption of further scanning to prevent failure of the MRI device [6]. At distances longer than 90 cm, the measured sound level approached the background noise level of the temperature stabilizer. The results of these measurements for high-resolution SE and GE pulse sequences can be seen in Figure 3. All the following measurements were carried out at the distance of 60 cm.

Results of the first comparison of energetic relations of vibration and noise signals are documented in Figure 6. Small differences were found for all sequence types, but the 3D-CE sequence produced the noise with minimal intensity expressed by the signal RMS parameter and SS-3Dbal calculated using the *En*c0 parameter. Further investigation was aimed at the influence of the choice of slice orientation on the energy of the produced vibration and noise signals. The graphs in Figure 5 show maximum energy in the sagittal plane and minimum energy in the transversal plane. Sagittal orientation was used in the remaining experiments to explore the worst case. In accordance with our previous research [15–17], the current experiments confirm the influence of TR and TE times on the vibration and acoustic noise properties. Prolongation of the TE time was accompanied by a slight lowering of the final signal energy with decreased first dominant frequency *F*V1, as documented in Figure 7. Higher TR parameter determining the fundamental frequency *F*V0 had greater influence on lowering the signal energy, especially for TR = 400 and 500 ms, as can be seen in Figure 8. The obtained vibration signal energy was higher for the water phantom than for the lying person (see Table 2), due to partial attenuation of vibration pulses by higher effective weight pressing on the bottom plastic holder of the gradient coils. At the same time, the noise signal energy had its maximum for the lying male person because of higher noise induced by vibration of the upper gradient coils that are not loaded by the tested object. From a physical point of view, larger volume of a person in the MRI scanning area needs higher electrical currents in the gradient coils, causing higher Lorentz forces and greater vibration of upper gradient coils for the examined human body in comparison with the 140-mm diameter spherical testing phantom.

Preliminary analysis shows positive influence of increased slice thickness on the predicted MR image *Q*<sup>F</sup> for both SE and GE Hi-Res scanning sequences—compare Figure 9. The effect of TR and *N*ACC on *T*DUR and predicted *Q*<sup>F</sup> of the executed scanning sequence is shown in Tables 3 and 4. While the increased repetition time caused only slightly greater overall time duration, the number of accumulations affected the final time duration very much. This applies for both types of Hi-Res sequences: Increasing *N*ACC from 1 to 16 resulted in about 4 times greater *Q*<sup>F</sup> and about 15 times greater *T*DUR. The same trend was observed for increased TR, its change from 60 to 500 ms caused about 7 (1.9) times greater *Q*<sup>F</sup> and about 7 (8) times greater *T*DUR for SE (GE). This course is valid also for two 3D scanning sequences, as seen in Figure 10. Here, the increased time duration was influenced also by the number of 3D phases being an equivalent to the number of slices (selected by the slice thickness) in Hi-Res sequences.
