Good points Rob.
This is a HUGE topic, but I will throw my two cents in on this as I am very familiar with this issue and have been preaching about this for years.
Intrinsic spectral broadening became a big issue when the world shifted from round mechanical crystals to linear arrays. However certain types of mechanical crystals can exhibit this as well.
One must remember that there are multiple crystals in the active aperture of a linear array, all receiving the reflected Doppler shifts. Every one of these crystals will receive the shift at a slightly different point in time and based on the angles of those interfaces, produce different frequency shifts. This is one of a few reasons that spectral broadening is so much worse on electronic probes than on mechanicals. Now, there is no standard among manufacturers as to how many elements can or should be active in the apertures at any given time. This is because the aperture is probe/frequency dependant and dependant on the manufacturers own specifications,  technologies and performance levels so it is different for all.
Another issue when comparing Doppler results from one system to another is that operators assume that just because the manufacturer lists the Doppler frequency as a 4 or 5 or 6 MHz, or whatever, that they are indeed that frequency. This is simply not true. They could be 4.5, or 5.2, or 6.6. So, not all manufacturers use exactly the same Doppler frequencies on all probes at all times.
However if we use intelligent angle correction and convert to velocities, the detected frequency shift changes would be minimal. Of course when all of these "minimal" effects start adding up, the differences can be startling as your research noted.
Now before everyone gets upset,,,,  all manufacturers Doppler technologies must meet accuracy guidelines to receive FDA approval and to my knowledge, all current systems meet these. It is not a perfect world however and we must accept the fact that there are differences in performance and technology levels, and we must  educate ourselves so we know how to compensate. (Remember, angle correction is a compensation, which we accept and use).
And as you mention, when you factor in all of the different probe types, frequencies, angles and apertures being used, and where in the aperture the Doppler cursor is being generated and the fact that the aperture automatically increases/decreases with image and sample volume depth and the severity of the lesion well, the proverbial worms have exited the can !!
Then,,, factor in all of the different stages/types of vessel lesions and residual lumens, and that no one really knows the exact flow direction, (we are making an assumption of a 3 dimensional area based on a 2D X&Y scan plane), and I think it is amazing that we are as accurate as we are !
Finally, I worked with a company that worked with you on your project and we did implement a S/W program that would compensate for the elements in the active aperture and correct the velocities by using the "end element for angle correction" which most closely approximated the correct angle, thereby eliminating the interference in the equation from the other elements that did not sit at the "displayed" angle. It worked very well in fact too well.
Why did this not become standard ?
Because another manufacturer, threatened to sue, claiming restraint of trade if the word got out to customers that such inaccuracies in Doppler actually existed.
In other words, who would buy the system if the customer knew it could be innacurate ? So, this "fix" died a quiet death. Could it be ressurected ?
Will it be?
Too many lawyers.
Oh, what a tangled web.
I wish you would reference the article you wrote so FLOWNET subscribers who care about this,   could look it up and educate themselves on this issue.
As for what can be done, there are new probe technologies in development that operate differently from the Piezoelectric crystals and arrays in use today.
You can "google" them by looking for capacitive micro machined ultrasound transducers (cmuts).
One of the many hopes of these transducers is more control over not only the aperture itself, but the shape of the aperture, in other words, we could have a multi-element probe that mimics a mechanical aperture for certain applications. Will it reduce intrinsic spectral broadening ?
Don't know yet,, but I am hopeful.
Big topic.
I look forward to discussion on this.

From: UVM Flownet [mailto:[log in to unmask]] On Behalf Of [log in to unmask]
Sent: Thursday, April 13, 2006 10:13 AM
To: [log in to unmask]
Subject: Re: HELP

Terry and Kirk, linear array Dopplers are a mess!
Let me start by saying the more I know about this issue, the more I realized how much I don't know..
The idea that a Doppler constant angle to flow solves all our inconsistency issues is misleading. The Doppler steering angle has perhaps a greater impact on variability than a 10 degree difference in Doppler angle, eg, 50-60 degrees  This is instrument dependent.  On phantom tests a few years back, we found a 5% difference in averaged velocities between 50 and 60 degrees obtained at the same Doppler steering angle. This isn't much. On the other hand, and on a different high-end instrument, we found a 17 % difference in averaged velocity between a precisely aligned 60 degree Doppler angle with the Doppler beam steered versus a 60% angle to flow with Doppler unsteered. 
We also observed and recorded considerable frequency variation on some ultrasound systems with the position of the Doppler beam cursor line along the face of the transducer,  at the same steering angle and the same Doppler angle. (When you more the line closer to the side of the transducer, you change effective Doppler aperture size, change sample volume size and change the amount of intrinsic spectral broadening and change the frequency display.)
A few years back I worked design engineers for a major ultrasound  company that had designed and demonstrated an algorithm that would compensate for the effects of intrinsic spectral broadening caused by the above Doppler steering and angle issues and reduce or eliminate all these variables.  They never  put it on their ultrasound systems, because to do so meant they'd have to: 1) admit that their Doppler frequency/velocity was not accurate (most aren't), 2) explain why linear array Doppler technology is so "messy", 3) market the advantages of a "fix" when 99.9% of their customers don't realize there needs to be a "fixing". So why shoot themselves in the foot?
My passion against the "fixed angle" concept is based on seeing too many exams being rejected by the interpreting physician because velocities were obtained at 57 degree angles rather than 60 degrees. Too many requests for a normal carotid patient to return for a repeat exam because some radiologist went to a lecture in which the speaker said "if you're not at 60 degrees, the exam is worthless". This stuff happens out in the clinical world, I see it today (last week in fact). It's also one of the reasons why we see too many instances of  uninformed users (read morons) setting the Doppler angle to 60 degrees and unrelated to flow direction or wall alignment (yes, it is ugly!) 
Perhaps rather than advocating a well-worn, partial solution (fixed 60 degree angle) to only one of the many variables affecting frequency display and subsequent velocity calculation, we should be pushing manufacturers to introduce solutions so that the Doppler systems don't introduce these variables. It is possible to get accurate velocities at 70 degree angle, it's just that our new tools won't allow it.
Get on it Kirk; talk to those boys and girls at that company in Bothell!
Rob Daigle
To unsubscribe or search other topics on UVM Flownet link to: To unsubscribe or search other topics on UVM Flownet link to: