Pi controlled Ultrasound Imaging machine.

[lilzz]

I have those HY-SRF05 Sonar ranging devices to detect presence of objects. can they be used as Ultrasound imagining equipment they used in the medical areas? I am looking for an off the shell , inexpensive design that transmit and receive well through human membrane, and tissues.
=================================================================


Transducers
In some respects, the transducer is the most critical component in any ultrasonic imaging system. In other words, such is the state of the art in systems such as electronic circuitry and display technology that it is the performance of the transducer which determines how closely the limits imposed by the characteristics of the tissues themselves can be approached.
Nowadays, the transducers which are in clinical use almost exclusively use a piezoelectric material, of which the artificial ferroelectric ceramic, lead zirconate titanate (PZT), is the most common. The ideal transducer for ultrasonic imaging would have a characteristic acoustic impedance perfectly matched to that of the (human) body, have high efficiency as a transmitter and high sensitivity as a receiver, a wide dynamic range and a wide frequency response for pulse operation. PZT has a much higher characteristic impedance than that of water but it can be made to perform quite well by the judicious use of matching layers consisting of materials with intermediate characteristic impedances. Even better performance can be obtained by embedding small particles or shaped structures of PZT in a plastic to form a composite material: this has a lower characteristic impedance than that of PZT alone, although it has similar ferroelectric properties.


Polyvinylidene difluoride (PVDF) is a plastic which can be polarized so that it has piezoelectric properties. The piezoelectric effect can be enhanced by the addition of small quantities of appropriate chemicals. The advantages of this material are that it has a relatively low characteristic impedance and broad frequency bandwidth; it is fairly sensitive as a receiver but rather inefficient as a transmitter.
Piezoelectric transducers are normally operated over a band of frequencies centred at their resonant frequency. The resonant frequency of a transducer occurs when it is half a wavelength in thickness. Typically, a PZT transducer resonant at a frequency of, say, 3 MHz is about 650 μm thick and this means that it is sufficiently mechanically robust for simple, even manual, fabrication techniques to be employed in probe construction. Higher frequency transducers are proportionally thinner and, consequently, more fragile.
The potential of capacitive micromachined ultrasonic transducers (cMUTs) at least partially to replace PZT and PVDF devices in ultrasonic imaging is the subject of current research. A cMUT consists of a micromachined capacitor, typically mounted on a silicon substrate and with a thin electroded membrane as the other plate of the capacitor: this acts as the active surface of the transducer. A dc voltage is applied between the plates of the device; the application of an ac voltage causes the membrane to transmit a corresponding oscillatory force, while a received wave causes a corresponding change in the spacing between the plates, thus generating an electrical signal. cMUTs are adequately sensitive as receivers, but need high voltages to be effective transmitters. Some of the potential advantages of these devices are that they can be fabricated into arrays with integrated electronics and, if manufactured in large quantities, could be relatively inexpensive.

[Ravenous]

Not a chance. You'll need a much higher frequency and a totally different emitter to those low resolution sensors operating in air.

(I would guess the raspi can't handle the sampling rates needed for useful imaging either.)

[Mark_T]

Indeed, ultrasound in air is totally different from ultrasound in liquid/solids (this is true of
acoustics generally).

I believe you need some serious digital signal processing at > 1MSPS rates to image
with sonar this way, since you have to cross-correlate the amplitude-adjusted signal against
the transmitted chirp - but that can probably be done offline.

[pguillem]

You will need a 1 to 5Mhz piezoelectric transducer in order to nearly come close to biological sonography.

Bare in mind the period of a 1 Mhz wave is 1/1000000 seconds, that being 1ns, which is at least 20 times smaller than a microphone. Following nynquist, you would have to read twice as fast (>2 MSPs) to get a reliable signal without aliasing.

Unless you build yourself some sort of intermediate interface to produce and read echoes at those frequencies, you might find yourself on a dead end with a raspi.

There is a reason for a sonograph to cost US25.000... the cheapest DSP processor for that application is roughly US 400, not to mention needing a decent background in electronics, filters, engineering, transforms and signal analysis to actually understand what is going on.

The raspi GPU can yield a 2million bar Fourier transform in 10ms (fft_gpu), but you will have a hard time gathering data.

Most modern sonograph equipment uses polyphase waves to create a reliable image, that is, using not 1 but up to 120 phase angles at the same time (120 threads)... which requires some preety sophisticated equipment and code.

Good luck on this one

[kelu124]

Hey, I just stumbled on this page by chance, and it's something I've been working on! I've put most of the good read and documentation at https://kelu124.gitbooks.io/echomods/content/RPI.html, but basically, for ultrasound imaging, I used a 10/20Msps raspberry extension with parallel ADC connected to the Pi GPIOs. I have 9 bits over a 0 - 3.3V range, which is sufficient if you have a good amplifier coming back from the piezo and with an analog enveloppe detection (which is the process that transforms the sensor signal into an image).

Moreover, with a AC mode, it's possible to acquire the amplified signal from the sensor, and even if I've tested a signal centered around 3.5MHz (and the piezo has a 50% bandwidth) at 11Msps, it's borderline. Double the frequency would be useful.

See image below:

[jchaput82]

Hi There,

Thanks so much for this! (2 years later, that is). I am a physicist and a tinkering Pi muddler who has a much better understanding of wavefield physics than electronics, but I've been developing the mathematical foundations for a steady state wavefield inversion approach that could potentially make the prospect of slice tomography tractable at extremely low costs (i.e., by skipping the datalogging portion, in part). If this project is still active, I'll be digging into it.
Cheers

J