Recording high quality speech during tagged cine‐MRI studies using a fiber optic microphone

Abstract

Purpose:To investigate the feasibility of obtaining high quality speech recordings during cine imaging of tongue movement using a fiber optic microphone. Materials and Methods:A Complementary Spatial Modulation of Magnetization (C-SPAMM) tagged cine sequence triggered by an electrocardiogram (ECG) simulator was used to image a volunteer while speaking the syllable pairs /a/-/u/, /i/-/u/, and the words "golly" and "Tamil" in sync with the imaging sequence. A noise-canceling, optical microphone was fastened approximately 1-2 inches above the mouth of the volunteer. The microphone was attached via optical fiber to a laptop computer, where the speech was sampled at 44.1 kHz. A reference recording of gradient activity with no speech was subtracted from target recordings.Results: Good quality speech was discernible above the background gradient sound using the fiber optic microphone without reference subtraction. The audio waveform of gradient activity was extremely stable and reproducible. Subtraction of the reference gradient recording further reduced gradient noise by roughly 21 dB, resulting in exceptionally high quality speech waveforms. Conclusion:It is possible to obtain high quality speech recordings using an optical microphone even during exceptionally loud cine imaging sequences. This opens up the possibility of more elaborate MRI studies of speech including spectral analysis of the speech signal in all types of MRI. THE USE OF MRI for measurements of the vocal tract during speech is becoming increasingly popular as MRI frame rates become faster (1,2). In ideal circumstances the acoustic wave would be recorded simultaneously with the MRI recording, to allow vocal tract shape data to be aligned with its acoustic output. A difficulty for researchers, however, is that the speech wave cannot be collected simultaneously with the MRI data due to the highly noisy environment of the MRI. In these cases speech is collected in the scanner after the MRI sequence is completed, or at a separate recording session. Humans, however, do not repeat items identically. This is true even when repetitions are sequential. Thus, speech data collected separately cannot be perfectly aligned and compared to the MRI image sequence.A good quality speech wave is the fundamental tool of speech research because the speech wave is what the speaker means to produce and is what the listener hears. Furthermore, in acoustics, it is known what frequencies and what size spectral changes are important. The physiological equivalent is not known. Therefore, comparison of acoustic and physiological features helps validate the importance of the observed physiological changes. As MRI is applied to speech, the speech wave will be used to corroborate physiological events in the vocal tract, or to predict the events using mechanical and physical inverse models of the vocal tract. The ability to collect a high quality speech wave during MRI recording will significantly enhance the use of MRI in modeling and understanding the control of the vocal tract duri...

A Technique to Reduce Motion Artifact for Externally Triggered Cine-MRI(EC-MRI) Based on Detecting the Onset of the Articulated Word with Spectral Analysis

“…Several studies of acoustic noise reduction and control for MR imaging [16][17][18][19] could reduce sound pressure approximately 20 to 40 dB using a subtraction algorithm with a cross-correlation or leastmean-square algorithm with several adaptiveˆl-ters. Most applied the methods to sound data recorded using a multiple microphone system.…”

“…Remarkably, NessAiver and colleagues 16 applied the NR technique to sound recorded during acquisition of MR imaging data to improve the quality of the reconstructed dynamic MR images. To reconstruct tagged dynamic MR images, they used complementary spatial modulation of magnetization (CSPAMM), which requires multiplying together the subtraction of two pairs of images.…”

“…They attributed this to temporal variability, such as that of speech onset relative to external triggering of MRI scan and that from inconsistency in sound duration. 16 Perhaps consistent speech time would have been achieved if they presented the subject with a rhythmic sound to signal the onset of the utterance and the onset of the following syllables, thereby improving image quality of their tagged MR imaging movie.…”

“…We employed a single microphone to record speech sound from the subject during the MR imaging scan, as proposed by NessAiver' s group, 16 i.e. an optical microphone (FOM-OPTICS 1140; Optoacoustics Ltd., Or-Yehuda, Israel) was mounted on the ‰exible coil so as to place it in front of the subject's mouth (Fig.…”

“…Several studies have proposed techniques to reduce such acoustic noise for acquisition of general, 17 dynamic, 16,18 and functional 19 MR imaging data. In this study, we applied a segmental subtraction approach explained below to reduce the acoustic noise in the recorded speech sound to detect the onset time of utterance of the speech material, /aki/.…”