Wednesday, 1 December 2010

2D / 3D / 4D Baby Ultrasounds

Being able to see your unborn child is truly an amazing experience. Ultrasound (diagnostic sonography) is a common diagnostic tool for, among other things, imaging the foetus to determine its age, look for abnormalities and observe blood flow in the umbilical cord. But possibly its most memorable effect is seeing your baby's heart beat - and in 3D/4D ultrasounds, seeing your baby's face.

The term "ultrasound" applies to acoustic energy (sound) with a frequency above the audible range of human hearing (20 Hz -20 kHz). When used in medical imaging, an ultrasonic sensor (or transducer) is placed on the mother's belly and produces pulses of sound. The frequencies used for medical imaging are generally in the range of 1 to 18 MHz. High frequencies (7-18 MHz) can be used to look for fine details but have low penetration, so to image deep tissue, lower frequencies (1-6 MHz) are used.

The sound waves are partially reflected from layers between different tissues inside the mother's body. Sound is reflected anywhere there are density changes - for example, at the baby's skin where it meets the amniotic fluid. The baby's internal organs can also be imaged depending on what frequencies you use. The reflected sound is then "heard" by the transducer, and the data analysed to produce the image. The amount of time it takes for the echo to rebound relates to how deep the sound penetrated, and the strength of the return signal relates to both the material it is reflecting off and its depth. The deeper the tissue from which the signal is being echoed, the quieter the return, simply because there is more sound loss (attenuation) the further the sound travels (it gets absorbed, scattered and reflected along the way). This information allows an image to be built up, whereby pixels at the appropriate depth are coloured by the strength of the return at that point. Generally, the sound waves are not 100% reflected at any stage - you can see "behind" objects because some sound penetrates through. However, as less sound is penetrating the deeper you go, the signals become fainter.

2D Ultrasounds

Baby face 2D scan

The typical ultrasound image is a "2D" image like the one above. In this image, the transducer is at the top and is sending sound waves down. The image is essentially a slice through the mother. It's called a 2D image as we can only see two dimensions - left/right and up/down. The 2D image is built by firing a sound beam down, waiting for the return echoes, and then firing a new pulse at a slightly different angle. This continues until an arc is swept. Combining the data from each line after the arc is swept gives the 2D image. The following images come from the excellent resource Basic ultrasound, echocardiography and Doppler for clinicians, by Asbjorn St√łylen. The left image shows the transducer scanning whilst the right image shows how the pulses are sent down in lines.



Continual rescanning means that a 2D video can be produced with roughly 50 frames per second. The human eye can see about 25 frames per second and so the video looks smooth. This frame rate is also more than enough for 2D temporal visualisation of the baby's heartbeat (~70-150 beats per minute depending on age) and to watch blood flow through Doppler ultrasound. Due to the Doppler effect, the sound pulse will rebound with a higher frequency if it hits something moving towards it, and a lower frequency if it echoes from something moving away from it - this is the same reason the noise of a car has a high pitch when moving towards you, and a low pitch as it moves away. As blood is moving in the umbilical cord, the ultrasound can be coloured by the Doppler information to show the blood flow.

3D Ultrasounds

Baby face

3D images are a fairly recent advance in diagnostic sonography. Instead of just seeing a slice through the mother, the images can show a surface - essentially adding depth (the third dimension) to the 2D image. Imagine you are looking at a car from front on - you have no idea how long the car is and you have no information on how many doors it has or if the boot is open. However, if you look at the car from another angle, you can figure this out, and the more angles you look down, the more depth information you can gain. This is essentially what a 3D ultrasound does - it stitches together multiple 2D shots from different angles to produce the image. Modern transducers have the ability to scan multiple cross-sections. If the baby is moving, there may be some blur, but as image processing is becoming quicker, the 3D images are becoming clearer. The colour of the image is not real as there is no way to see colour inside the mother. 3D scans provide information for the diagnosis of facial anomalies, evaluation of neural tube defects, and skeletal malformations, and also helps the parents bond with their unborn child (it's very cool). However, when compared to 2D scans, they aren't as useful for the diagnosis of congenital heart disease and central nervous system anomalies. One of the reasons why this is the case is because they are static, which leads us to...

4D Ultrasounds

The term 4D refers to the addition of time to 3D scans. This is a very recent advance as it is only in the last few years that we have had the computing power to not only stitch together the 2D images to make the 3D images, but to create the 3D images quickly enough to play them consecutively as a video. Modern 4D scans play at roughly 12 frames per second, so they are a little jumpy.

Here is a little video I put together of our 4D scan.



I don't know if there is an upper bound on what ultrasound technology can do - as the speed of sound is ~1540 m/s in human soft tissue, and you have no choice but to wait for the return signal before you can process the image, it may be that a high video frame rate with decent resolution is unobtainable. Resolution depends on how many different lines you fire down to make the first 2D image - more lines mean better resolution, but currently you have to wait for the echo from one line before sending down the next, which means it takes longer to produce an image. I imagine one way of improving this would be to send down all the lines at once with slightly different frequencies or waveforms, and as such when the echo is received you would know where it came from. Perhaps this is already being done - let me know if you know more!

Check out the video of Massive Attack's Teardrop in which there is a singing foetus, and I also have more images over at my ultrasound set on flickr.

References:
  • Kurjak, A., Miskovic, B., Andonotopo, W., Stanojevic, M., Azumendi, G., & Vrcic, H. (2007). How useful is 3D and 4D ultrasound in perinatal medicine? Journal of Perinatal Medicine, 35 (1), 10-27 DOI: 10.1515/JPM.2007.002 
  • Carrera, J.M. (2006). Donald School Atlas of Clin. Application of Ultrasound in Obs/ Gyn www.jaypeebrothers.com DOI: 10.5005/jp/books/10226
  • Khanem, N. (2007). Donald School Textbook of Ultrasound in Obstetrics & Gynecology The Obstetrician & Gynaecologist, 9 (2), 140-140 DOI: 10.1576/toag.9.2.140.27325 

2 comments:

  1. This comment has been removed by a blog administrator.

    ReplyDelete
  2. This comment has been removed by a blog administrator.

    ReplyDelete