Revitalizing Ultrasound

May 23, 2018

Reconciling the technique and technology of ultrasound.

I was told that globally, last year, there were around 175 ultrasound manufacturers with sales of some 225,000 ultrasound units. I suppose that’s accurate, and I can only imagine how many exams are done everywhere, every day given a pretty low rate for equipment upgrading. Obviously, ultrasound contributes a lot everywhere or it would not have such presence.

One of the problems of success is that it resists change, which can be problematic in fields where technical advances and fundamental knowledge are increasing at a rapid rate. Another problem is that when ultrasound started, it did best with obvious, advanced or end-stage disease, while a big focus in public health now is early diagnosis and preventive health.

One of the strangest things about ultrasound is that it has fit into our health care communities in a way that is contrary to basic notions of medical education and practice, which strives continuously for re-evaluation and improvement.

Ultrasound’s origins were appropriately humble, but it offered a novel way to look into patients. Over time and with increasing popularity, the goal of ultrasound is most often to image a part of a body rather than contributing to an inclusive understanding of the whole patient.

The scope of the exams has become narrow, or finely focused, as some would have it. Obviously, the system works. This approach minimizes physician involvement, which fits right in with the “team” approach paradigm. It is not demanding of equipment performance or of operator education. Patients often need separate ultrasound exams in different locations, and there is an unfortunate tendency of referrers to do shotgun scheduling of all kinds of studies in the hope of receiving a windfall of actionable information. There are additional problems arising from self-referral and from referrals from sources that do not have a full sense of what ultrasound can provide in their own institutions.


I am stressing the routine limited ultrasound, because anything seen outside the study framework constitutes an incidental. It is hard to pin down an exact definition, because it is so dependent on how and why an ultrasound is done as well as the age and comorbidities of the study population.  Incidental Findings on Bedside Ultrasonography: Detection Rate and Accuracy of Resident-Performed Examinations in the Acute Setting reports review of 196 FAST exams by EM residents, who found a 30 percent incidence of incidentals of which 70 percent were reconfirmed by radiologist reviews of available images. Other studies with high performance equipment (in preparation) yield secondary findings in 50 to 75 percent of exams depending on patient age and specialty area.

Whether there are incidentals or not and how they should be reported or evaluated can be contentious issues. It’s a little “meat” versus “bitter poison” in the original Roman adage.

A painful, plain chest film learning experience

We are the products of all of our experiences. Let me recollect some basic lessons from the start of my diagnostic radiology residency in 1970, two years after I made my first clinical ultrasound images at NIH. The basic rule was that if you needed to irradiate someone, get the most out of the procedure (and, of course, use as low an overall dose as possible). For a film or series of images, the “reader” is responsible for everything that can possibly be seen therein, no excuses. One of the painful ways this was illustrated was to put a first-year victim on the spot in front of the rest of the department with a mystery chest x-ray of someone with cleidocranial dysostosis. Absence (or hypoplasia) of both clavicles does not change the expected symmetry of the bones of the thorax, so beginners nearly always miss it. Whether you’re the victim or an observer, you never quite forget the embarrassment of failing to recognize something so obvious. There were a lot more recondite examples. Film reading for a radiologist involves everything in the film, not to a region or a system, independent of the reason for the exam. 

Fluoroscopy, the original 2D dynamic imaging technique

Fluoroscopy was the original real time viewing core of radiology. You locate radiographically visible objects by rotating the target, and you get a lot of supplemental information by how things move spontaneously and respond to stress. (Hint: the “radiologically visible” requirement is really important).  High speed ultrasound (which is all of ultrasound now) is the heir of fluoroscopy. I did lots of upper and lower GI exams. I was taught that it was mandatory to read through each patient’s chart before the exam and to use fluoroscopy, briefly, for anything that would be likely to provide valuable information for the referring physician and to modify the study as needed during the exam.

Cautionary experiences with CT and Virtual Colonoscopy

It is not too surprising that early users of CT developed protocols with a total body scanning component, but this never had a good diagnostic yield. Whole-Body Computed Tomography Screening by Lincoln Berland set the stage for ending the practice. Lincoln is a great radiologist and a principal figure in the history of ultrasound.  His paper: Incidental Extracolonic Findings in CT Colonoscopy was a meta-analysis of some 7,500 cases. Incidental findings occurred in a high percentage of exams. In about 10 percent of cases, the incidental “finding” was thought to be important enough to warrant further study. A cost-benefit analysis showed that the economic burden of the incidental findings was about five times the costs of the CT colonoscopy exams themselves. Abdominal aortic aneurysm was the only condition with a demonstrable patient benefit.

CT has always enjoyed more credibility than ultrasound. It’s easy for the uninformed person to assume that if CT cannot do something, then little orphan ultrasound definitely cannot. That is all wrong from the physical nature of tissue mapping with streams of photons versus pulses of phonons.

Branchpoint 1975

January, 1975, I started a new job as an assistant professor of radiology at Stanford and the chief of a new (and very small) ultrasound division. I became the clinical site for the Searle manual scan gray scale unit, which was the very best unit of its type at the time. A little later, I was invited to a private unveiling of the Varian electronic sector scanner by the late Leslie Zatz, head of radiology at the Palo Alto VA and one of the revered first wave of ultrasound pioneers. The unit was intended for cardiac imaging. After the live demo when two other physicians and a lot of Varian people were celebrating with champagne (which they deserved, the unit blew away everything else at the time for high speed, high quality imaging), I asked if I might have a look myself and a got a desultory mid ‘70s equivalent of “knock yourself out.”

The model could not have been more perfect, being a super-healthy, super-fit man in his 20s with no body fat. There was a 3.5 MHz probe designed for intercostal viewing, which is just right for the liver and spleen, so I went immediately to the upper abdomen. I was stunned. The first thing I noticed was that you could adjust the unit for good image quality while scanning. I can’t tell you how important that was and how time saving it would be in practice. I saw that you could zip through an organ with any probe orientation, and you could move the probe around to look wherever and however you wanted. I mentioned to the company vice-president that there was more to the body than the heart, and he agreed. The first two new units wound up just a few floors away from each other at Stanford; cardiology on an upper floor, and radiology in the basement.

I set up ultrasound the way that the best fluoroscopy units were organized. I, or one of my fellows, did the initial exam and documentation with the Varian and then a technologist was tasked with highest detail static images of a specific region with the Searle unit. The chart had to be available, and the examiner was free to look wherever clinical concerns and extant information led. The Varian unit was big and bulky, but it was on wheels and portable exams were done in the main hospital and the Children’s Hospital on the other side of the medical campus. There was not a lot of call for remote ultrasound then, because the general value of ultrasound was not common knowledge or supported by published papers. I replicated the setup and procedures the following year at the Peter Bent Brigham Hospital and the Boston Hospital for Women, which combined and modernized a few years later. Cardiac imaging was a component of a big part of the Brigham population.

One of the papers that came out of the initial Stanford experience with physician-done, dynamic exams was  Combined ultrasound imaging and physical examination: Editorial review of phased array sector scanning techniques and applications (Journal Belge de Radiologie 1978, 61(5):463-469). Ultrasound expands and improves the physical examination, like an opthalmoscope, but it does not replace it.

Large aperature single element manual scanned systems were doomed. Within just a few years, all ultrasound was being done with high speed imaging devices with transducers specific to particular application areas. It was time to unify the exam, but in what way?

The dominant ultrasound pathway

The early days of ultrasound were grim, just like the earliest days of MRI. You couldn’t see much, and it took a really long time to generate images. Early adopters had to slog through exams with bistable display devices and were overjoyed when TV grayscale imaging became available. Few paid any attention to the phased array, and they tended to employ surrogates to scan and document, although they were typically on site and available during exams when questions arose. The most efficient plan of operation for them was to limit attention to a specific clinical issue or anatomic field. Image information content and equipment performance left a lot to be desired, there were few academic departments with full time ultrasound physicians and dedicated ultrasound technologists.

Ultrasound is treated administratively very much like nuclear medicine, which had fixed location equipment, relatively few isotopes and few clinical uses. It was never easy in most institutions to purchase new equipment, update older units or expand the range of transducers as new ones became available. Detractors were quick to identify the subjective nature of ultrasound exams as a weakness; there was really never the support necessary to convert the subjective part into a reservoir of strength. 

An example
A 29-year-old woman is referred for a routine first trimester pregnancy exam. Why was I doing it? Because, I examine every patient myself. I also have a personal interest in embryonic development. Why was the patient referred to me? I can’t say, but my referrers all know that I never treat any exam as routine.

(See Figure 1)

A bicornuate uterus may have been unexpected, but is hardly incidental. And, yes, both kidneys were present. I usually start with a quick abdominal survey in new OB patients with no big expectation of finding anything; it serves to settle the patient down, before we get to the emotionally charged part of the study.

(See Figure 2)

IBD is not usually discovered first in someone in their late 20s, but it is not rare either. Inflammatory reactions appear to be due to mitochondrial DNA release in the gut, symptomatic flares may not be recognized by the patient for what they are and fade from memory when or if symptoms subside and other worries appear.

This example involves arbitrarily different systems or specialty interests, but they are in the same anatomic space. Is this an incidental finding? Could be-if one thought that the only reason to do an early first trimester ultrasound is to identify one or more embryos, establish an intrauterine location, and confirm vitality. That does not make a lot of sense medically, nor is it in the best interests of the patient or the referring physician. 

Multitasking & digital magic

There is always cognitive multitasking whenever a physician interacts with a patient. Medical education involves building this ability within prevailing medical knowledge and biomedical science for application in interacting with patients. The theory can be humors, vital essences, or molecular genetics, and treatment options can be leeches or robotic surgery, the concept is as vital now as in was in antiquity. Specialty residencies expand foundational multitasking by reinforcing and selectively enriching perceptual channels. One of the previous articles in this series referenced the unique way that radiologists perceive medical images. I would be astounded if there were not many works showing something identical for psychiatrists and auditory processing.

Ultrasound in its earliest and simplest form was analog. The advent of one and two dimensional arrays lead to hybrid digital and analog systems. Now, we are into fully digital systems except for transducer and display surface materials and power source elements. It’s pretty easy to simulate a 1990s-era hybrid device in a physically small unit with a functionally or physically limited set of transducers. Several senior ultrasound industry people have admitted to me that these units are not intended for radiology use. The proposed target niches are physicians, domestic and third world, primary care and non-ultrasound using specialists, who are presumed to become happy with simple uses. After all, isn’t some ultrasound better than none? Perhaps there is also the notion that this is very much like the same services that they have been referring patients for years, one gall bladder, breast, thyroid, or pelvis at a time, now made available as if no other ultrasound service is available nearby.

We have been immersed in competent, focused, clinically simple ultrasound for so long, and the referrals so narrow in scope, that the broad search applications have been lost uses have been lost. Narrow ultrasound is appropriate in many situations, like an injury in a healthy athlete or pneumothorax in a premie in a NICU, and it may be efficient  for sick, hospitalized people in whom there are no unresolved diagnostic issues, but there is the polar opposite of a patient who has few or no complaints who seeks medical aid. Physicians multitask and need to accrue information to make a treatment plan; ultrasound images are a trivial part of the process. This was a truism when ultrasound started. From the standpoint of the physician managing a patient with a serious complaint, it is still just as true. The physician forms a partnership with the patient that starts with the current issue but which also extends far into the future via preventive health measures.

Primary care physicians would be great in using ultrasound themselves in full-on diagnostic mode, but that will only work with the best equipment (meaning really effective noise reduction, especially at high frequencies) that can be used to look wherever clinical intuition and suspicion leads as the exam progresses. Really good image quality also translates to rapid exams, because normal fields are obvious and unequivocal. A physician scanner can be expected to leverage his or her intimate knowledge of pathology and pathophysiology.

Digital processing has hit a teenage growth spurt and is transforming into an alien place. In the New Horizons lecture at the 2017 RSNA, Dr. Daniel Sodickson referred to the digital landscape for MRI as a place “where things are seriously weird” (If that is not exact, it is very close). He went on to show some spectacular examples from undersampling and/or discarding data that is irrelevant to what one wants to find.  A few ultrasound manufacturers have been quietly challenging conventional design wisdom and coming up with much higher fidelity ultrasound images. There are a few such multi-probe, full-size units FDA approved and CE marked and one ultraportable tablet.

Challenging conventions for beam forming and data processing requires a massive amount of near-instant computation and some mighty clever software engineering. Traditional units did one set of operations for each image, forming a stereotyped image one line at a time. A new device might run a bunch of separate programs simultaneously on send and receive, forming a composite image from snippets that are optimal for each part of the imaging field. The initial fallout are features like full field of focus at high frame rates, greatly increased Doppler sensitivity, and noise reduction, which increases both spatial and contrast resolutions.

The Point of Care is where the doctor scans

Industry, administrators, and regulatory agencies all need to understand that the needs of physicians go far beyond fragmentary bits of information generated by narrowly focused or limited exams. The concept of Point of Care Ultrasound needs to be defined as where the physician is, whenever he or she interacts with a patient. 

The newest forms of ultrasound imaging make it possible to move away from a fixed location without any sacrifice in image quality or diagnostic yield. Newer units all have full connectivity so that images and video clips can be shared at the time of the patient interaction, not for radiological review but for management help on the spot. Tele-consultation might be about the operability of some lesion in one case, antibiotic choice in another, or referral for some other kind of exam.

Radiology should step up in providing diagnostic ultrasound services in the field by physicians with high performance portable equipment. This is easily enough tested in any academic department. As the range of equipment increases and performance advances, radiology should take the lead in integrating ultrasound into residency programs in other specialties, when it can be merged with the way a new physician learns to merge new skills sets into his or her multitasking repertoire.