From its commercial appearance in the early 1980s, magnetic resonance imaging (MRI) has blossomed into a standard component of diagnostic services. Over this 20 year period, it has grown from devices that were claustrophobic, expensive, and difficult to site into a robust technology available almost everywhere in America. Will similar development characterize the next two decades or is MRI approaching its golden years? This attempt to peer into the crystal ball and project its impact in the year 2020 is based on trends noted in technology, utilization rates, and MRI economics.

(Caveat: Projecting the future has been likened to nailing Jell-O to a tree. You may be marginally successful but usually you just make a mess. It is important to carefully question all information used for projections. The following speculations are founded on my experience since 1983 with MRI vendors, imaging providers, and research groups. The supporting data are open to interpretations other than my own, and I welcome opinions from those willing to share differing viewpoints.)

TECHNOLOGICAL CHANGE

The first commercial MRI devices in the United States were tubular resistive and superconducting units (0.15T to 0.5T) that appeared in the early 1980s. In late 1984, the first generation of 1.5T systems began to be installed, and by the latter half of the 1980s, high field was the dominant trend. However, these were largely unshielded units that produced “footballs” of 5 G fringe fields about 60 feet in diameter and 80 feet long. They also required refills of liquid nitrogen about every 2 weeks and liquid helium about once a month. The market found high-field units produced better image quality, but the increased costs of equipment and site preparation exacerbated risk. The development of actively shielded 1T units moderated costs while retaining the advantages of high field. This caused the rapid demise of tubular superconducting systems below 1T.

In the mid 1990s, low-field, open-sided designs became available. These produced great excitement among providers who had struggled with patient anxiety and claustrophobia. By early 1998, sales of these units represented more than half of all MRI sales in the United States, even though examination times were longer than for high field, and clinical applications were somewhat limited.

Figure 1. United States MRI installed base, 1995.

About this same time, compact high-field systems began to appear. The shorter bore reduced claustrophobia, and active shielding made siting easier than ever. Further, advances in shield cooling reduced liquid helium refills to once every 2 to 3 years. Now providers could have high-field image quality, short examination times, and lower patient anxiety in one design. The appearance of “high-field,” open-sided units in the past few years (0.6T to 1.0T) has re-energized interest in open-sided designs, but at present their high cost and lower image quality appear to limit adoption in many cases. They currently represent less than 10% of the installed base of units between 0.6T and 1T.

Figure 2. United States MRI installed base, 2002.

These transitions can be seen in the distribution of MRI units by field strength between 1995 and 2002 (Figures 1 and 2). Although the overall US installed base grew by almost 75% during this period, the fraction of systems in the range of 0.5T to 0.6T fell from 20% to 3%, demonstrating the disappearance of mid-field tubular MRI. The fraction representing 0.1T to 0.49T grew from 15% to 28% and its contributors were altered substantially. The large majority of tubular systems in this sector (0.35T) were retired and low-field, open-sided systems flourished. Thus, by 2002, the distribution of MRI units was approximately one third low-field, open-sided and two thirds conventional high-field (Ž 1T) tubular systems.

The most recent development has been the advent of short-bore, ultra high-field units, most commonly 3T. The latest National Electrical Manufacturers Association sales numbers indicate 3T is now outselling the entire open-sided sector. Primary reasons for this activity are (1) the marked improvements in image quality offered by 3T and (2) site considerations that are only slightly more complicated than for modern 1.5T units.

Figure 3. Projected United States installed MRI base in 2020.

What trends can be discerned from this data? Image quality continues to be the primary motivation for equipment selection. Field strength is a prime element of image quality as it contributes to higher signal-to-noise ratio (SNR), shorter examination times, and greater range of applications. Thus, higher field strength will be in demand. Yes, users also want to reduce claustrophobia but if this entails lower image quality or substantially higher prices compared to the same field strength tubular systems, it is likely to be a weaker motivation. Thus, based on technical grounds, it appears ultra high-field systems will dominate the next generation of MRI. My projection for equipment distribution by type in 2020 is seen in Figure 3. Designs lower than 1.5T are likely to fade away as are open-sided units costing much more than their tubular counterparts. A significant fraction is assigned to specialty magnets, which will utilize technology and magnet designs largely not available today. Based on current sales activity, I predict 3T will displace 1.5T and become the most prevalent field strength in 2011 to 2012.

UTILIZATION RATES

Projecting future growth requires an examination of past performance, growth in similar modalities, a review of continuing clinical utility, consideration of competing technologies, and a healthy dose of chutzpah. Each of the first four is discussed below in order to estimate MRI examination volume in 2020.

Figure 4. US MRI utilization rates by year.

MRI usage rates (expressed as the number of examinations per 1,000 of population per year) have increased every year since 1983 except for 1993 during the health care debate in the first Clinton administration (see Figure 4).

Over the period 1998-2002, the average growth has been 16% annually and from 2001 to 2002 was 22%. If one projects growth at 5%, 10%, and 15% per year, the resultant volume in 2020 is shown in Figure 5, beginning with 25 million examinations in 2003, the last complete year for which data are available. Based on 15%, the 2020 volume is about 270 million examinations, almost one per person in that year presuming the United States population stabilizes at about 300 million. Even at 10% growth, the 2020 volume is about 126 million.

Are such examination volumes realistic? CT is a similar device where an additional 10 years of clinical history are available. According to IMV International, Des Plaines, Ill, annual CT growth in the United States is currently about 15% (2001 to 2002) and has averaged 18.2% from 1998 to 2002. Thus, presuming MRI tracks CT performance, growth of 10% to 15% annually is quite reasonable.

Is there evidence that MRI information is coming less useful or that risks in the technology exceed patient benefit? Based on the volume of clinical papers written on MRI and on the rate of development of new applications (eg, breast, spectroscopy, cardiac, MR cholangiopancreatography, MRI angiography, diffusion tensor imaging, functional MRI), it appears that the data argues quite the opposite. Further, the safety of MRI, when practiced according to established standards, is better than virtually any other diagnostic modality with the possible exception of ultrasound.

What other modalities might be expected to garner a significant fraction of examinations now conducted by MRI? Simultaneous multislice CT has great promise for expansion into angiography but is more likely to shift examination volume from conventional x-ray angiography, not MRI. Ultrasound is improving, but is not applicable to most MRI examinations due to the absence of acoustic windows. Positron emission tomography (PET) is far more expensive than most other diagnostic examinations and is unlikely to grow substantially in a time of federal deficits and other pressures on health care costs.

What are potential sources for additional MRI examination? Of the estimated 52 million CT examinations conducted in the United States in 2003, about 50% were body examinations, yet for MRI, this fraction was less than 10%. Either MRI is not an appropriate examination for many body applications or this may be a prime area for its expansion. I tend to favor the latter view. Other opportunities may include cardiac studies, breast, psychiatric follow-up, functional MRI, therapy planning and evaluation, spectroscopy (proton and other nuclei), OB-GYN, prostate, surgical preplanning, stroke assessment, rheumatoid arthritis, and autism.

MRI ECONOMICS

MRI is a high fixed asset investment, one in which the majority of expenses are incurred independent of examination volume. As such, it is relatively simple to calculate the break-even volume, the number of examinations needed to meet costs. A typical freestanding MRI facility costs about $900K per year to operate. If average technical reimbursement is $650 per examination, it requires about 1,400 examinations to meet those expenses (breakeven). The downside to such an enterprise is failure to achieve breakeven, which can result in potentially large losses (commitments for equipment, the site, basic personnel, etc) must be made prior to any examination revenue. However, once breakeven is achieved, the large fraction of any additional revenue becomes profit since most operating expenses have been paid. As shown in Table 1, cost per examination varies substantially depending on the scan volume over which expenses are spread. The perceptive MRI operator will understand that volume and efficiency can compensate for lower reimbursement per examination. This is a powerful trend that will continue to shape MRI into the year 2020.

Table 1. Example of Typical MRI Contributions To Examination Costs

But will not MRI reimbursement increase over time? Double-digit annual increases in health care costs over the past few years have again focused intense scrutiny on the cost of medicine in America. The current federal deficit of more than $450B and projected deficits in the range of $300B to $400B over the next few years augurs poorly for increased reimbursement. Indeed, in the past 5 years the Centers for Medicare and Medicaid Services (CMS) has repeatedly tried to markedly lower reimbursement rates for MRI (by 21% in 1999 alone). I believe substantial reductions across the board should be anticipated since private payors are likely to follow Medicare changes. Thus, efficiency and expense control will be vital to the future of MRI.

How have MRI costs changed? In 1985, one could expect to pay about $1.6M for a 1.5T system. Additionally, service costs were about $140K per year and typical site expenses were $750K. Today, a new 1.5T MRI can be purchased for less than $1M ($567K), service costs are about $110K/yr ($62.3K) and site costs are typically less than $250K ($142K). The numbers in parentheses are the 2004 numbers inflation adjusted to equivalent 1985 dollars per United States Bureau of Labor statistics. Thus, the actual costs of MRI equipment, service, and site expenses have plummeted over the past 20 years by factors of approximately 65%, 55%, and 80%, respectively. Even a modern 3T unit at $2.3M is only $1.3M in 1985 dollars. And what of reimbursement? Presuming the average MRI technical fee in 2004 is $650 ($368), it has also fallen by 70% compared to a 1985 technical component of $1,200.

Are similar reductions likely in the next 20 years? I believe similar inflation-corrected decreases are very probable. Technical reimbursement is also likely to fall into the range of $300 to $350, which, if the past is a proper model, should be worth about $200 today corrected for inflation.

Some progress has been made in scanning efficiency, but far more needs to be done. In a survey Robert A. Bell and Associates, Encinitas, Calif, conducted in 1994, we found average time per examination of 1.3 hours at hospitals and 1.1 hours at freestanding MRI facilities. In 2002, there were approximately 22 million MRI examinations conducted in the United States on about 6,500 units. Thus, the average MRI conducted about 3,400 examinations. According to IMV International, the average hospital MRI site scheduled services about 64 hours per week. Assuming 50 weeks per year (2 weeks for downtime and holidays), this results in about 3,200 hours per year or slightly less than 1 hour per MRI examination at the average site.

Over the next 10 years, I anticipate average scan efficiency improving by at least a factor of two and possibly as much as a factor of three. Where is the precedent? Consider CT. In the 1980s, the typical CT examination also took 45 minutes to an hour. According to IMV International, in 2002, there were approximately 45 million CT examinations conducted in the United States on about 8,500 units. Thus, the average CT conducted about 5,300 examinations. The average hospital CT site scheduled services about 53 hours per week. Assuming 50 weeks per year (2 weeks for downtime and holidays), this results in about 2,650 hours per year or 30 minutes per CT examination at the average site. One might argue that CT is easier to conduct but it uses a much higher fraction of contrast and CT patients are often not as mobile as MRI patients. Exceptional MRI sites in the United States today are scanning at less than 30 minutes per examination with no decrease in quality.

Figure 5. Projected MRI examination volume by year through 2020.

The market is likely to shift toward technology that provides improved image quality at lower cost per examination. This can be achieved more readily by increasing throughput than by cutting equipment costs. Therefore, if 3T (or higher fields) can process patients faster and still maintain excellent image quality, they will be in demand despite higher equipment acquisition costs.

SPECIALTY MRI

Predicting new equipment opportunities is perhaps the hardest aspect of the future. Prior to the early 1970s, it would have been very difficult to predict the impact CT has had on x-ray. But one should also note that x-ray has not disappeared. On the contrary, about six plain film examinations are conducted annually for each CT examination, largely driven by the cost-effectiveness of this basic examination. If lower cost MRI devices can be developed, they may be capable of providing inexpensive basic MRI information. Advances in magnet design, superconducting technology, and other areas may offer fertile ground for such products. A vendor in Carlsbad, Calif, is manufacturing portable units that use an 80-pound magnet and produce images about two inches on a side. I have predicted that by 2020 about 15% of MRI devices would fall into this category, but if an MRI equivalent of plain film x-ray is developed, it could have enormous impact. Perhaps shoe stores will even install them so we can see if our toes have enough room!

SUMMARY

Three powerful trends that will help shape the future of MRI are (1) the desire for improved image quality, (2) the need to process patients faster, and (3) substantially increasing volume of MRI examinations at lower reimbursement per examination. These can be mutually satisfied through higher-field systems used efficiently. 3T units are likely to replace older systems in hospitals and in freestanding clinics.

Robert Bell, PhD, is president of R.A. Bell and Associates, in Encinitas, Calif. He welcomes your comments and criticisms, (858) 759-0150.