Spatio-temporal patterns of vocal fold vibrations

From Lab UCLA 2003 04/16: These were the flow meters used when I first arrived in the lab. I don't have photos about the rest of the flow supply path to the larynx. Here is a short description: flow meters; heated humidifier (built for patients); about 2m long flexible tube; metal larynx mounting tube screwed to the experimenting table. The subglottal acoustic environment was not well designed (after all this was just a verbatim copy of the excised larynx setup in Iowa): sometimes the vocal fold vibrations would excite vibrations of the little flow meter ball in the Dwyer flow meters UPSTREAM from the preparation, even past the humidifier (a small expansion chamber). Furthermore, the frequency of oscillations seemed to be in a narrow range, maybe dictated by the subglottal acoustics (this stimulated my interest in the influence of subglottal acoustics on vocal fold vibration). Then, at the end of a long day the larynx tissue would dry out, despite the nominal presence of the humidifier. But this heater/humidifier was never validated to work for flow rates as high as 500 to 1500 ml/s. It could never keep up! Certainly, the dried tissue would show all kinds of nonlinear phenomena and bifurcations. Years later I would redesign this setup. With a large expansion chamber and moisture reservoir right upstream from the larynx, it was then very rare to observe nonlinear phenomena. The larynx would be in the same state even after hours of vibration.
From Lab UCLA 2003 04/16: First excised larynx experiment with David's setup: flimsy and small prism and even more flimsy and shaky mini-traverses (difficult to get flush to glass plate); one suture row (How can I study spatio-temporal patterns with one row?); calibration recticle to be traversed; calibration at end of day, otherwise wasting the whole day, especially if one changes the position of the camera during a 6-hour experimenting day. For the calibration the excised larynx preparation has to be cut off while not changing the position of the glass plate: VERY UNRELIABLE and ERROR-PRONE! And don't even dare to bump the tripod with the camera on it during the experiments. Then ALL data collected before that will be invalid. This procedure also can not be used on an in vivo canine preparation: there is no space to traverse a calibration target (recticle); the preparation can't be fixed enough.
From Lab UCLA 2003 04/30: Second excised larynx experiment with major improvements: bigger prism; camera at an angle to view superior surface suture points; experimenting with 60 degree prism; most importantly: 5 rows of sutures, extending across the superior edge onto the superior surface -- Now I can analyze spatio-temporal patterns of vocal fold vibration!!!
From Lab UCLA 2003 06/11: I tried to tape the two scales on top of each other so that the connection of corresponding vertices would be perpendicular to the glass surface plane.
From Lab UCLA 2003 06/11: First attempt to create a calibration target which would be always visible during the high speed recordings. I taped a vertical and horizontal scale on the front and back of the glass plate.
From Lab UCLA 2003 07/02: During the next excised larynx experiment we used a small brass cube made in the machine shop. This cube was glued on the back surface of the glass plate, superior to the superior surface, within the field of view of the high speed camera.
From Lab UCLA 2003 07/02: View of the brass cube and the previously used recticle. The recticle needed to be traversed perpendicular to the glass plate plane in order to get calibration information in the out-of-plane direction. The advantage of the cube is that this error-prone tranversing is not needed.
From Lab UCLA 2006 03/04: Unfortunately, some people in the lab preferred to publish too many papers on the cube method (without my name on any of the papers), although for me this calibration method was only a simplified way of how eventually a commercial turn-key system would do it: 3-d two-level calibration target, pinhole camera model, random pattern on vocal fold surface (either tattoo ink or graphite powder) to use image cross-correlation analysis for surface deformation analysis (DaVis, made by LaVision).

Glottal flow studies: Spatio-temporal behavior of the glottal jet

From Lab UCLA 2003 07/23: A first crude attempt to do flow visualization with an excised larynx. The problems were that 1) the theatrical fog machine could not pump the fog into the pressurized subglottal inflow tube, and 2) the light setup did not use forward scattering to achieve maximum scattered light intensity into the camera (Mie scattering). Here we didn't even use a fog machine, we used some fog sticks, some chemical reaction that produces white fog (I guess the chemical products were not so healthy!)
From Lab UCLA 2004 03/03: another flow visualization attempt. This time probably with a theatrical fog machine which had the problem mentioned above.
From Lab UCLA 2004 06/15: So what to do when the fog machine can't pump fog into the flow supply line? Well, I just simply put the fog machine INTO the flow supply line! On top of that I use a very short inflow tube. And I can acoustically dampen the expansion chamber which houses the fog machine. Furthermore, to test equipment and setup, I started using a physical model (first just a donation from Scott Thomson, later I started making more myself based on Scott's recipe.). And last, I needed a better light source: a laser with a ligh sheet module attached to it! Now I was ready for good flow visualization, thanks to Reza! At some point I also tried one of his suggestions to seed the flow with some 'dust', in this case I tried talc powder: either I tried throwing it into the glottal jet as it exited the physical model glottis OR I put a small talc filled container on a Bruel and Kjaer shaker inside the expansion chamber together with some computer fans to disperse talc powder there. It was pretty messy but it kind of worked!
From Lab UCLA 2005 09/21: Steve Anderson from LaVision visits to demonstrate how to use our brand new 2D3C PIV system with the DaVis software package. We also got a new flow seeder (mineral oil atomizer) so the expansion chamber remains empty and only serves as acoustic termination of the subglottal inflow tube. This setup was still put together on the old small lab table. Notice that the old subglottal flow setup is visible. Or, rather, the absence of a subglottal flow setup is visible. There is only a heated humidifier and a 2m long flexible hose.
Lab UCLA 2006 10/10: Stereoscopic setup for PIV measurements of glottal jet. The cameras are mounted both in side-scattering mode with respect to the laser. Notice the rotated Scheimpflug adapter with respect to the camera housing to fulfill the Scheimpflug condition.
Lab UCLA 2006 10/19: Stereoscopic PIV setup for confined glottal jet measurements. The cameras are both mounted in side-scattering mode.
Lab UCLA 2006 10/25: Stereo PIV for very narrow confined glottal jet (1 inch tube). The cameras are mounted in side-scattering and forward-scattering mode. This is the only way to obtain optical access through the rectangular tube.
Lab UCLA 2008 10/24: Pilot study for flow visualization and PIV in a hemi model, in this case a hemi physical model connected to a hemi model of the vocal tract (mirror on gimbal mount). We wanted to visualize the flow separation in the glottal channel.

Tissue stiffness measurements: Indentation

Lab UCLA 2007 09/11: Eric Goodyer visits with his 'Linear Skin Rheometer' device. Looks almost like an indentometer, but is not capable of working that way. Sucks (in fact, it uses suction to stay attached to the tissue)! The right vocal fold had to be completely removed to allow access to the left vocal fold with the rather large suction tube. There was no humidification during the measurements.
Lab UCLA 2008 01/18: Last time the 'Berkometer' worked before I took it apart to make our own indentometer. All I needed from it was the force transducer. Here some attempt at indenting a full excised larynx is shown. The long rod served as a lever that allowed to mechanically amplify the indentation force so that it was measurable with the not-so-sensitive force transducer. The new larynx holder can also be seen. There was no humidification during the measurements.
Lab UCLA 2008 10/14: I used the motorized traverses (from the hotwire setup) and the Shimpo force transducer (from the Berkometer) to build our own indentometer, controlled by my Matlab program. The Matlab program samples the amplified (10x, 100x) output directly from the force gauge (NOT the digital display of the Shimpo force transducer). Building this amplifier took the longest time here. Additionally, we use very small indenting cylinders: the cylinder diameter should be small compared to the tissue sample thickness or width. And, most importantly, the indentation depth needs to be small compared to the tissue sample dimensions, especially the thickness. Eventually we did indentation on vocal fold samples and also on an intact hemi-larynx. During measurements the tissue samples were always immersed in saline.
Lab UCLA 2008 10/24: Indentation on physical vocal fold models, both from the older poly-urethane rubber and the newer silicone rubber.

In vivo canine larynx experiments

Lab UCLA 2007 08/01: Hotwire anemometry on an in vivo canine larynx preparation: pilot study for feasability of traversed hotwire anemometer to measure the spatio-temporal behavior of the glottal jet
Lab UCLA 2008 05/28: 2D3C PIV on an in vivo canine larynx preparation: pilot study; also: collagen injection