One regrettable consequence of this website may be increased anxiety in pregnant women who are having sonograms. I very much doubt that there are any fetal consequences; the purpose here is to encourage nuts-and-bolts prudence by the ultrasound community. Sadly, this community will not do much on its own and some gentle prodding by patients/clients is the only way that I can see to improve the situation. Is it not peculiar that medical users tend to be self-righteously critical of commercial fetal viewing facilities because of safety concerns?
Increase in tissue temperature caused by ultrasound was the earliest mechanism considered for potential biological effects. “The TI gives a relative indication of the potential for temperature increase at a specific point along the ultrasound beam.” (17). One related index - TIB - assesses the particular potential for heating adjacent to fetal bones. Reference 17 elaborates on the difficulties with using the TI and related indices. Only the MI is required to be displayed in B-mode; for Doppler and color-flow imaging the TI is supposed to be displayed as the overall energy intensity is relatively higher. TI may be displayed with B-mode. TI/TIB values are low for B-mode imaging, but relatively higher for color and PW Doppler. Looking at the video clips helps explain why, with markedly increased flow effects reflecting higher energy inputs.
The conventional wisdom is that one can draw considerable reassurance from the fact that B-mode ultrasound is unlikely to cause significant heating (4). I find high TI/TIB values to be rather disturbing, but do not take much reassurance from relatively low values in view of considerations surrounding mechanical effects previously outlined. Note that, “if all of the energy deposited were converted to heat, a dose of whole-body radiation that would be lethal to human beings would raise the temperature of individual cells only 0.001 oC.” (48).
2012: The 2011 Statement on Measurement of Fetal Heart Rate by the American Institute of Ultrasound in Medicine (AIUM) discourages the use of spectral (pulsed wave - PW) Doppler for measuring embryonic or fetal heart rate (49). Although this method often produces good tracings more easily than standard M-mode, with an “audible heartbeat” as well, it involves higher thermal-type exposures. If spectral/PW Doppler is used, intensity settings should be kept to a minimum with only a brief tracing. Note that the “audible heartbeat” generated by non-imaging continuous wave Doppler (eg handheld devices used in doctors’ offices and fetal monitors) involves much lower intensities.
For what it’s worth, I doubt that precise measurement of heart rate is necessary in early pregnancy when it is normal on ordinary real-time visualization.
Ionizing radiation comparisons:
Some years ago I spent time as a medical advisor (from Prince Edward Island!) to our then Atomic Energy Control Board, and retain an interest in risk evaluation of ionizing radiation (47). The most recent overview publication is the BEIR VII document from the (US) National Research Council (23). It notes the absence of genetic effects demonstrable in humans, including an extensive follow-up of 30,000 children of exposed atomic bomb survivors. Given findings obtained from laboratory studies, genetic risks are very low and “one would not expect to see an excess of adverse hereditary effects in a sample of about 30,000 children” (page 9). Approximate genetic risks are projected from experimental data (page 12). While attention has been shifting to radiation-induced cancers, minimizing gonadal exposure in diagnostic examinations remains important. Gonadal exposure is given a significant weighting factor in calculation of whole-body equivalent doses when studies are confined to a limited part of the body. So, with the comparatively puny data base for reassurance for fetal ultrasound follow-up, why do users tend to act as if fetal sonography is known to be harmless? The information deficiency is especially true for higher output equipment following the increase in regulatory limits in 1992. Do we have a double standard?
Detecting the risk of breast cancer induction from ten years of annual mammography as an increase over “natural” cancer risk by patient follow-up would require cohorts of tens of millions of women (24) and decades of observation. Risk estimates for mammography are derived from higher dose exposures, with extrapolation to the diagnostic range on a linear no-threshold assumption. This assumption is conservative and reasonable, but not universally accepted(40).
ALARA in practice:
Easier said than done in terms of downward adjustment of power settings during examinations. This is especially true when using harmonic imaging; I find that our Toshiba Aplio can only create satisfactory fetal images on harmonics. With the real-world pressure to get the optimum study in a reasonable time, constant twiddling with the power and gain settings is probably not practical. Dr. Abramowicz, in response to my question about whether there was downward adjustment of power settings in the observational study of which he was the senior author, replied that “examinations were usually performed with default settings” (4,21,22).
What to do?
My answer: Those who buy equipment for fetal ultrasound should seek out units and probes that will provide satisfactory images on the lowest default MI settings. A starting point would be to keep MI values below 0.23 to ensure that fetal eye exposure is no more than the regulatory limits following birth.
Not only will this result in minimized exposure, but it will create market pressures to bring intensities down in general. Additionally, manufacturers should be encouraged to provide an option for automated real-time adjustments within their systems so that intensities will be dropped when the signal path passes through significant amounts of non-attenuating fluid (maternal bladder, amniotic fluid) before reaching the region of interest as identified by the focal zone selected. This should be a simple exercise for programmers. I do not want to inject an adversarial element into the experiences of individual patients, but if they politely encourage the user community perhaps some progress can be made about equipment selection. I have seen no evidence to suggest that any other approach will work.
Who guards the guards?
This ancient question (Quis custodiet ipsos custodes?) applies in surveillance of modern technology. Evaluation of safety issues in obstetric/fetal ultrasound should be transferred to some organization with no vested interest and not encumbered by the work patterns of government departments. There should be more representation from younger individuals – at present there are too many of us older folks with baggage. We also need a lot more research, although where the money for it would come from is uncertain. Maybe it sounds delusional, but could the Bill and Melinda Gates Foundation spare a few million for this orphan area?
The ALARA principle, as borrowed from radiation protection, has two pillars: justification and optimization. So far this site has largely addressed the optimization aspect in fetal ultrasound – minimizing exposure in an examination that is considered necessary. Some space for examining justification of fetal examinations is required. Most recently this has been brought up by the use of 3D/4D sonography for entertainment and bonding purposes, especially with independent commercial facilities offering this service - discussed in section 8.
Prudently applied, 3D ultrasound can actually shorten exposure duration compared to conventional 2D imaging by taking five 3D volumes of the fetus and then using them “offline” to generate images by slicing in desired planes (26). “It took a mean time of 1.1 minutes to obtain the 3D volumes…With the standard 2D technique, the structural surveys were done in a mean time of 13.9 minutes”.
Although screening ultrasound is now entrenched practice, the benefits may not be as dramatic as one might think. A Cochrane review of nine trials showed earlier detection of multiple pregnancies and reduced rate of induction of labor for post-term pregnancy, but no differences for substantive clinical outcomes such as perinatal mortality. When a search for fetal abnormalities was part of the examination, there were increased numbers of pregnancy terminations for fetal anomalies (27). At the time that screening ultrasound was becoming popularized I wrote an article (28) hoping to stimulate discussion and debate, but without success. I have to admit that increasing patient obesity now renders it harder to make reliable clinical assessments.
2010: Improvements in image quality have enhanced prenatal diagnosis, especially for detection of significant congenital heart disease. From a practical point of view, a screening sonogram at about 20 weeks is entirely reasonable.
Detection of fetal anomalies can be therapeutically important, but screening can also cause anxiety and distress when “soft markers” for chromosomal abnormalities or anatomical findings of questionable significance are detected. One such situation is to encounter mild fullness of the drainage of the fetal kidneys; despite knowing that this finding is almost certainly within normal limits the examiner may decide that it is medicolegally a good idea to suggest a repeat study later in the pregnancy. This sort of problem has been elegantly summarized in the title of an article “Antenatal diagnosis of renal tract anomalies: has it increased the sum of human happiness?” (29).
An encouraging development has been prevention of neural tube defects (spina bifida etc) with folic acid before and around conception. Will there be further progress along these lines?
2010: Additionally, I hope that ultrasound exposure issues related to first trimester screening for Down’s syndrome will be eliminated when maternal blood analysis for fetal/placental DNA or RNA will permit reliable testing that will make examinations for nuchal thickness irrelevant.
A disturbing consequence of the ability to identify fetal gender has been termination of female fetuses in some societies; it has been estimated that up to 10 million female fetuses have been aborted in India in the last 20 years (30). It would seem that fetal ultrasound has ended more life than it has saved. I am pro-choice with misgivings, and find this to be an illustration of Bouchier’s Columbus Principle: Any new activity will cause more trouble than you can possibly imagine.
Addendum September 2007: The dismal state of what clinical users know about acoustic intensities, and onscreen displays of these values, has been nicely documented in an article in the March 2007 issue of the Journal of Ultrasound in Medicine. This was a survey of users, principally obstetricians and technologists. Only 3.8% could describe the mechanical index (MI) properly and only 20% were aware that acoustic indices are displayed on the sonographic monitor during examinations (33).
UPDATE JULY 2008
1. Recently I saw a large ovarian cyst with homogeneous echoes in which
intensity-dependant motion of the contents was demonstrated with B-mode, color
Doppler and PW Doppler using our conventional probe for abdominal and obstetric
ultrasound. The cyst had extended into the abdomen, permitting evaluation
with this probe through about 2.2 cm of overlying skin and tissues. I had previously sought
this effect in superficially located amniotic fluid with echoes without success,
but perhaps amniotic fluid characteristics are not as favourable. Findings
could be reproduced with the same probe in skim milk corresponding to videos in
section 6 (but not as pretty); flow with PW Doppler was about 1.5 cm/sec.
Note that this was with our standard 3.5 MHz Toshiba Aplio probe for obstetrics.
Can someone else, possessing academic credibility, not work with this sort of
model with a variety of situations and publish relevant results?
2. A new article has demonstrated the value of a conventional diagnostic
ultrasound system with PW Doppler in accelerating clot dissolution with TPA
("clotbuster") in middle cerebral artery stroke (34). Acoustic
intensity given by the manufacturer was 179 mW/cm2. The fraction of this
that reaches the vessel of interest is probably less than would be calculated
from the conventional derating formula, as reference 13 cites a study of cadaver
skulls which demonstrated a 65-90 percent energy loss due to ultrasound
accumulation in the overlying temporal bone.
Several orders of magnitude separate radiation doses used in conventional
radiology imaging and those used in therapy. While comparison with
ultrasound should be applied cautiously, as radiation therapy is usually for
cancer, the overlap between diagnostic and therapeutic ultrasound levels is
somewhat disturbing to me. Knowing that their equipment may well be generating
intensities with biological effects sufficient for therapeutic purposes should
make even the most complacent users of fetal imaging a little uneasy.
Nonetheless, I have little optimism that there will ever be any pressure from
the broad medical user community on manufacturers to reduce intensities.
3. The April 2008 issue of the Journal of Ultrasound in Medicine is
largely devoted to safety issues. The lowest level of information for
evaluation of epidemiological studies is a report from a committee of experts;
the highest level is the randomized controlled trial (35). Unfortunately,
randomized controlled studies are short in supply (36), provide some mildly
unsettling information (1), and probably new studies will be difficult or impossible
to implement in a world of routine prenatal ultrasound.
Unrealistically optimistic risk evaluation of subprime mortgages, resulting in
AAA ratings from supposedly independent rating agencies for "asset backed
commercial paper", has led to disastrous financial consequences. Wishful
thinking is not a good basis for risk assessment and reduction. Anyone who
thinks that 40 years of experience with fetal ultrasound can permit confident
reassuring statements should note that fetal alcohol syndrome was not described
in the United States until 1973, with subsequent recognition of a range of fetal alcohol spectrum disorders. The 1975 edition of the standard pharmacology textbook stated “alcohol gains free access to the fetal circulation, but it does not seem to harm the fetus” (41). I am disappointed at physicians and
sonographers who say they repeatedly looked at their own fetuses with
ultrasound, see no adverse effects in their children after birth, and are
therefore confident that fetal exposure to diagnostic ultrasound is safe; this
reasoning would earn them an F- in a clinical epidemiology course. Most
cigarette smokers do not develop lung cancer.
4. The possibilities for complex interactions of environmental influences
in fetal life and delayed non-cancerous effects in much later life are explored
in the July 3rd 2008 issue of The New England Journal of Medicine (37).
5. This website has about 1000 page views/ visits per month. Despite
the invitation under "contact", no-one has responded with information rebutting
any of the content. I have also sought responses directly from several
expert sources without any substantiated critical replies - actually a couple
that were complimentary. If new information is supplied, or if I am justly
taken to task, an apropriate addendum will be added to the website.
UPDATE OCTOBER/NOVEMBER 2009
Reassurance for Mothers: To repeat the point made in the first paragraph of the introduction, and also the first paragraph of the miscellaneous items section: I doubt that there are significant short or long term adverse effects from fetal exposure to ultrasound. There are no proven hazards that I am aware of. I do not lose any sleep over the ultrasound examinations of my grandchildren.
Fetal Ultrasound and Informed Consent: Give Given that fetal eye exposure is commonly well above the ophthalmic ultrasound guidelines, the question of obtaining informed consent from the mother arises. After communication with a prominent Canadian medical ethicist, I sent a submission to the Canadian Medical Association Journal which was
"published" as an e-letter and is
Delays in Pediatric CT Dose Reduction: CT (computed tomography) scans involve large radiation doses by diagnostic imaging standards. This is of particular importance in children, where radiation risks are greater. It is only in the 21st century, however, that dose reduction protocols for the smaller child body have been published and promoted. The American Roentgen Ray Society was gracious enough to give me space on their web site,
viewable here, for comparison of issues in pediatric CT and fetal ultrasound. (Typo in the item - my town is Amherst, not Amyherst)
MRI, Gadolinium and NSF (nephrogenic systemic fibrosis): Visualization of abnormalities with magnetic resonance imaging (MRI) can often be improved by intravenous injection of gadolinium-based compounds. For many years this was thought to be innocuous, and was freely used even in patients in kidney failure. More recently the unusual and unpleasant delayed complication of nephrogenic systemic fibrosis (NSF) has been documented, especially in patients with kidney failure. An October 2009 article describes 36 patients from one institution (38). This unexpected and unintended consequence has probably come to attention because it is such a distinctive disorder; had it been a corresponding increase in stroke or heart attacks obscured by a high background frequency would more prudent user attitudes towards gadolinium have developed or would they still be cavalier?
Progress: Presentation on safety issues in fetal ultrasound are now part of some refresher courses.
AN ALTERNATIVE TO DOPPLER FETAL MONITORING? While my interest as a radiologist is in imaging ultrasound, probably the largest amount of fetal exposure is from continuous-wave Doppler ultrasound for fetal heart monitoring during labor, and also for nonstress testing for assessment of fetal welfare in late pregnancy. In 1984 I wrote “it is hard to believe that modern technology cannot provide satisfactory tracings from audible fetal heart sounds in most patients. Pressure on manufacturers from the physician-consumer would likely produce results.” (28). A beautiful two-page article in The Lancet the same year provided an example of how such tracings can be obtained (39). If this was a reasonable proposal 25 years ago, surely it is far more so now. My memory was recently jolted while listening to a medical technology lecture which included a discussion of the extraordinary capabilities in submarines for detecting and identifying low-frequency propeller sounds from other vessels. Can we not use military technology for real human advantage (swords into plowshares – Isaiah 2:4)?
PubMed review shows some progress in this area and also potential applications to fetal welfare assessment by analysis of heart sounds. I would be particularly interested in feedback about this, as it is outside my usual domain.
Ultrasound and the Cell Membrane:
An important article appeared in the Feb 22, 2011 edition of the Proceedings of
the National Academy of Sciences (PNAS) giving both a theoretical framework for,
and electron-micrograph illustrations of, ultrasound effects on the cell
membrane (42). The work comes out of the Technion, Israel’s
equivalent of MIT, and the article is now open access. The schematic
diagrams in figure 4, and illustration of experimental effects in figure 3,
provide a nice visual summary. The experimental model uses goldfish epithelium.
Those who would dismiss it on these grounds should note that the cell membrane
is a generally fundamental structure. A relevant historical precedent is
the work of Hermann Muller, who reported the first demonstration of genetic
effects of radiation in 1927 using fruit flies and was later awarded the 1946
Nobel Prize in medicine and physiology for this. Fruit fly research was
then used for exploration of radiation genetics long before the structure of DNA
was described in 1953. The goldfish is at least a vertebrate, and the
relative simplicity of the experimental set-up (compared to mouse and human
studies) will allow for assessment of effects of a range of exposure conditions
and also for replication by other investigators. Dr. Eitan Kimmel, the
senior investigator, tells me that this will be the focus of his future research
and I look forwards to seeing how it works with pulsed diagnostic ultrasound.
I gather he has interesting preliminary findings in this regard.
The proposed mechanism is a cavitation process between the two lipid molecular
layers of the cell membrane, which can occur at much lower pressures than
conventional cavitation. The mathematical expression underlying this,
given in the abstract, actually closely resembles that for the MI (mechanical
index) value used as an onscreen intensity measure in clinical ultrasound.
There is also a much less elegant way of looking at ultrasound and the cell
membrane. The sort of pressures occurring in the ultrasound pulse have
been described as only capable of moving a particle by 50 nanometres (nm); since
the cell membrane measures 3-10 nm in thickness this displacement is
less reassuring to me than to the authors (43).
The Fundamental Importance of Fundamental Frequency:
Until 12 – 15 years ago diagnostic ultrasound used the same
frequency emitted by the probe for receiving and processing the returning echoes
to create the image. A 4 MHz probe would thus emit at a frequency centered
around 4 MHz, and then receive and process returning echoes centered around 4
MHz. This approach does not inherently require large pressure fluctuations
in the ultrasound pulse to work. In more recent years tissue harmonic
imaging (THI/”harmonics”) has become embedded in practice, as it does offer some
image improvements. The subsection on Tissue Harmonic Imaging in section 3
of the website describes why THI requires large pressure fluctuations in the
ultrasound pulse and therefore relatively high MI values. To review: When
the peak positive pressure in the ultrasound pulse is high enough, tissue is
compressed to a point where the speed of sound is increased, and during the
following maximum negative pressure the speed of sound is reduced because
of tissue expansion. As the pulse moves through the body this effect
allows for development of a harmonic at twice the frequency of the fundamental
frequency which was emitted by the probe. For example, a probe emitting at
2 MHz would receive 4 MHz components in the returning signal for image creation.
Think about it – the process requires pressures sufficient to compress or expand
tissue enough to alter the speed of sound! Most current equipment appears
to be tuned to functioning with harmonics and has forfeited the ability to
deliver decent images on fundamental frequency at low intensities. An MI
value of 1 appears numerically unimpressive, but it actually corresponds to
about 30 atmospheres of pressure variation in a 2 MHz ultrasound pulse.
The following illustrations should give some real-world context to the above
considerations. The MI values and Tis (thermal index) values are in the
upper right corner of the images. In fairness, the pictures do
not adequately convey the improvements in visualization that have occurred.
There is certainly no criticism of General Electric, a company for which I have
considerable respect. I do not know of another manufacturer of ultrasound equipment that can deliver
satisfactory images at truly low output values.
The first two pictures
(picture 1, picture 2) are from a 10 week pregnancy with a GE Logiq 9 on
fundamental frequency with default settings. Note that the MI is 0.19,
below the 0.23 guideline for ophthalmic ultrasound, and the Tis is zero.
These images are certainly good enough for dating and twin exclusion, and are
really quite acceptable by any standards. Higher output techniques may be
needed with more obese patients.
The third image is from a 20-week pregnancy on fundamental frequency with
the same low MI and Tis values as above. The fetus is well demonstrated –
it is in a photogenic position. While the mother is not obese, there is an
anterior placenta which attenuates ultrasound. The fourth image is from
the same pregnancy with harmonics turned on, resulting in MI of 0.7 and Tis of
The fifth image is from the newer Logiq E9, which appears to be directed
at functioning on harmonics. Note that the MI is now 1.1 and Tis is 0.7.
Fetal visualization for this 17 week pregnancy was excellent, but is the extra
exposure necessary? The matter deserves systematic exploration. One
major impediment is that patients have expectations for the prettiest picture
possible; another is that user-based stimulation for development of equipment
optimized for fundamental frequency died years ago.
Of Mice and Chicks
The now well-known article demonstrating alteration of neuronal migration in fetal mice exposed to ultrasound has been discussed in section 4. A more recent study reports adverse effects of ultrasound on fetal mice with respect to subsequent learning and memory (44). While the experiment was done with a diagnostic unit, the conditions were sufficient to increase the maternal body temperature by about 1 degree and thus likely represent more exposure than usual clinical situations. On the other hand, the findings are sufficiently impressive to suggest that some effect would likely have been seen with lesser duration and intensity. There was evidence of increased sensitivity with exposure on day 14 compared to day 16 of pregnancy. All such studies need independent replication.
Exposure of chick brains to PW Doppler ultrasound for 4-5 minutes two days prior to hatching has been shown to cause remarkable negative consequences for learning behaviour two days after hatching (45). Standard imaging B-mode ultrasound for 5 or 10 minutes had no effect. The study was done with a high resolution probe in the context of concern about neonatal head ultrasound examinations in premature babies, but reinforces recommendations against use of PW Doppler in fetal ultrasound unless really necessary. I thank Dr. Jacques Abramowicz for drawing my attention to this article.
Would anyone who sees mistakes in the above please let me know at
firstname.lastname@example.org. Sonnet 116: “If this be error and upon me proved..”
Maybe I should just give this up these efforts and instead offer prayers to St.
Rita, patron saint of hopeless causes.