The Three Worlds of MRI

Robert Turner

Max-Planck Institute for Human Cognitive and Brain Sciences, School of Psychology, Cardiff University,

Sir Peter Mansfield Imaging Centre, University of Nottingham, U.K.

Abstract: The development of MRI from its beginnings in 1972 provides many lessons in the mutual benefits obtained

when experts in three quite different disciplines learn to communicate with each other. Important technical

breakthroughs have occurred every few years. I will describe eight of these, largely based on my own

experience, and show how the differing perspectives of basic scientists, industrial engineers and medical

professionals such as radiologists have combined fruitfully to enable a transformation in how we humans

understand our bodies and brains, in sickness and in health.

1 INTRODUCTION

The development of MRI from its beginnings in 1972

provides many lessons in the mutual benefits obtained

when experts in three quite different disciplines learn

to communicate with each other. The disciplines

concerned are of course physics, in particular the

physics of nuclear magnetic resonance and of electro-

magnetism; engineering, including radiofrequency and

audio-frequency electrical engineering, together with

cryoengineering; and radiology.

Each of these disciplines can be regarded as a

social and economic arena with its own particular

structures of authority, culture, validation and

economics. Entrepreneurial innovation can exist in

each of these arenas, but because of the essentially

scientific nature of MRI, the most penetrating

innovation has tended to come from physicists in

university laboratories, who generally have the most

freedom to explore crazy ideas. Large engineering

firms such as General Electric, Siemens and Philips

have strict administrative hierarchies which limit

unsupervised enquiry by junior staff. However, this is

not to discount the usefulness of what have been

called ‘skunkworks’, groups of engineers who

operate under the radar with the tacit encouragement

of their immediate management, pursuing potentially

useful ideas which may be far from the priorities of

their senior management. In a similar way, while

radiology departments are normally under the

autocratic authority of the senior radiologist, and

junior staff may have very little time available for

exploring and validating new techniques, there are

outstanding exceptions where entire laboratories

based in radiology departments are totally focussed

on innovation.

I will not discuss the initial inventions and

discoveries on the part of Richard Ernst, Paul

Lauterbur and Peter Mansfield that led to MRI as a

clinical modality, based entirely within the realm of

physics. Since 1980, important technical

breakthroughs have occurred every few years. I have

picked eight of these as illustrations of the type of

interaction between the three disciplines that results

in significant changes of practice, largely based on

my own experience.

These are as follows:

1) Observation of coronary arteries in adult

human heart, in Nottingham in 1986.

2) Dramatic improvements in gradient coil

technology, from 1986 to 1990, at Nottingham

and elsewhere.

3) Development by Pykett and Rzedzian of a

purpose-built whole-body MRI scanner

capable of snapshot echo-planar imaging, at

Advanced NMR Systems in Massachusetts in

1987

4) Introduction of diffusion-weighted EPI by

Turner and Le Bihan as a clinical modality in

radiology at NIH, between 1989 and 1992.

Turner R, Le Bihan D, Maier J, Vavrek R,

Hedges LK, Pekar J. Echo-planar imaging of

intravoxel incoherent motion. Radiology. 1990

Nov;177(2):407-14. doi: 10.1148/radiology.

177.2.2217777. PMID: 2217777.

5) Implementation of EPI on a clinical Siemens

scanner by Schmitt at Erlangen in 1991-2

Turner, R.

The Three Worlds of MRI.

DOI: 10.5220/0012639900003657

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2024) - Volume 2, pages 9-15

ISBN: 978-989-758-688-0; ISSN: 2184-4305

6) Discovery of BOLD contrast fMRI between

1990 and 1992, at Bell Labs, Minneapolis,

Boston and NIH

7) Establishment of routine functional MRI

scanning for studies of cognition at Functional

Imaging Laboratory, Queen’s Square, London

in 1995

8) Development of NextGen 7T MRI scanner,

Berkeley, by Feinberg and colleagues, 2022.

1) Observation of Coronary Arteries Using MPI

In Living Human Heart, 1986.

The demonstration of coronary vessels in 1986

(Stehling 1986) was ground-breaking, showing that

magnetic resonance imaging could be performed

sufficiently rapidly to obtain snapshots of the human

heart with sufficient spatial resolution to reveal

proximal portions of coronary vessels. By then over

400 MRI scanners had already been installed in

hospitals worldwide, many of them with 1.5 T

superconducting magnets, but the imaging sequences

which used hundreds of radiofrequency pulses spaced

at least 100 milliseconds apart, were too slow to

capture the rapidly moving coronary vessels. It was

the work of Peter Mansfield and his team of

physicists, working in the Physics Department of

Nottingham University with an MRI scanner at only

0.1T magnetic field, that showed that echo-planar

imaging (EPI), with which an entire 2-D image can

be acquired in 30 milliseconds, could freeze the

cardiac motion and thus enable viewing of structures

clearly identifiable as coronary veins and arteries.

This breakthrough was driven by the physicists and

engineers of the Nottingham Physics Department, and

facilitated by a radiologist, Brian Worthington, at the

Queen’s Medical Centre in Nottingham. No engineers

from the growing MRI manufacturing industry were

involved. The electronic hardware was largely home-

built, and the MRI sequences and reconstruction code

were written by members of Peter Mansfield’s team

(including the author).

2) Improvements In Gradient Coil Technology,

1986-88

The key to much more efficient MRI sequences, such

as EPI, was the development of coils delivering

strong magnetic field gradients that were accurately

linear and providing fast rise times with few eddy

currents. Existing gradient coils in 1985 were

primitive in their performance in every one of these

respects, and few researchers were making any

attempts to improve them. I could see a choice of two

basic design improvements: either make the coils

much smaller than the magnet bore radius, or shield

their fields to prevent their interaction with other

conducting surfaces that would impair their

performance. In Peter Mansfield’s group we explored

both possibilities. Even before I joined this group in

1984, his students had devised and built several types

of gradient coil, but it was during 1985 that I derived

the formal mathematics of cylindrical coil design, the

‘target field method’ which included the simple

shielding equation, allowing the fabrication of coil

assemblies--primary and shield--with very low fringe

fields and fast rise times (Turner 1993). Barry

Chapman (a brilliant post-doc) and I built such coils

by hand at first, using patterns printed out from the

computer and pasted on to strong GRP tubes, onto

which we glued the wire with a glue gun. Later Peter

Mansfield persuaded Oxford Magnets (at that time

the world’s major supplier of MRI hardware) to

manufacture a double-screened coil set, to my design,

thus establishing the concept of multi-layer gradient

coils that has proven to have critical importance in

state-of-the-art gradient coils that are currently being

built (see below).

During the same period, completely

independently, Peter Roemer and Bill Edelstein,

engineers working in GE’s MRI research lab in

Schenectady, had been developing ‘fingerprint’

design gradient coils which were a major step forward

from previous designs. A few months after my

derivation, as it turned out, they had themselves

discovered the shielding equation. GE went on to

incorporate these design principles in manufactured

gradient coils, firstly for a preclinical scanner

intended for small animal scanning, and then for their

1.5 T Signa clinical MRI scanners. After some

arguing about patent priorities, GE conceded and

large royalty payments duly arrived at Nottingham

University. GE and the other major MRI

manufacturers redesigned and rebuilt their scanners

over the next few years. The result was a dramatic

improvement in gradient coil performance.

For this major development in MRI both teams

consisted of physicists, one in a university research

lab and the other in an industrial laboratory, in the

years before most large multinational companies

were closing down their research labs. Such industrial

labs would often maintain the ethos of university

labs—opportunities for independent research,

rewards for innovation, minimal management

intervention, etc., but by the 1990s became viewed as

unprofitable by senior management. Radiologists at

the time were completely unaware that MRI

performance could be greatly improved by better

engineering physics; they were mostly simply

BIOSTEC 2024 - 17th International Joint Conference on Biomedical Engineering Systems and Technologies

delighted by the diagnostic quality of the images

provided by the first round of installations.

Completing the MRI hardware transformation

was another major improvement in gradient driver

technology, led by Siemens, whose engineers

(notably Franz Schmitt) designed extremely powerful

gradient current amplifiers with water cooling to

enable a much higher duty-cycle.

3) Development by Pykett and Rzedzian of a

Purpose-Built Whole-Body MRI Scanner

Capable of Snapshot Echo-Planar Imaging, at

Advanced NMR Systems in Massachusetts in

1987

Another variation on the theme of a breakthrough at

the combined hands of scientists, engineers and

radiologists came about between 1986 and 1990. Two

of Peter Mansfield’s former PhD students, the

physicists Ian Pykett and Richard Rzedzian, left the

UK and set up a venture-capital funded development

company, Advanced NMR Systems, Inc (ANMR).

This was located in Woburn, Massachusetts, and it

was greatly aided by the financial and logistic support

of the Radiology Department of Massachusetts

General Hospital (MGH), then headed by Thomas J

Brady, a man of great vision who could already see

the enormous potential of an imaging technique that

took only a few tens of milliseconds to acquire an

entire image. Remarkably, the company attracted

significant investment capital, and they were able to

recruit highly competent additional physicists and

engineers. By 1987 they had constructed a 2.0T

whole-body EPI scanner capable of producing

cardiac images of diagnostic quality (Rzedzian 1987;

Cohen 2012). The scanner was striking in its novelty,

using new concepts in gradient driver and coil

technology to generate powerful resonant oscillating

gradients, and novel data sampling and analysis

strategies. ANMR was soon employed by GE to

retrofit a GE Signa 1.5 T scanner installed at the MGH

NMR research laboratory in Charlestown, Boston

with the special hardware and software required for

EPI. This GE/ANMR hybrid scanner was used to

obtain the first non-invasive images of human brain

function in 1991.

The MGH Radiology Department played a crucial

role in the success of this development, giving it

institutional backing and international credibility, as

well as providing seed funding and backing for NIH

grant funding proposals. However, according to Mark

Cohen, one of ANMR’s recruited physicists, it was

the engineering perfectionism of the physicist

Rzedzian which drove the project to its ground-

breaking fulfilment.

4) Introduction of Diffusion-Weighted EPI by

Turner and Le Bihan as a Clinical Modality in

Radiology at NIH, between 1989 and 1992

That nuclear magnetic resonance could be used to

study the self-diffusion of liquids was recognised by

the physicists Stejskal and Tanner in 1965. Using

magnetic field gradients applied after spin excitation

but before data acquisition enables the tracking of

molecular-level Brownian motion, but also sensitizes

the signal to bulk movement of the sample. It was not

until the implementation of MRI in the 1980s that the

idea of spatially mapping the diffusion constant of

water in tissue was conceived as a potentially clinical

biomarker. This was largely the work of Denis

LeBihan, a French MD/PhD radiologist, then

working at the Ecole Polytechnique, Palaiseau,

France. In 1987 he was recruited to the Diagnostic

Radiology Department at the Clinical Center of the

National Institutes of Health, Maryland, which is

where I met him when I moved there in early 1988.

The diffusion-weighted images LeBihan had been

obtaining with the current standard slow technique of

spin-warp imaging often suffered from artifact due to

involuntary head or body movements, which could

completely rule out any diagnostic value. I had

already proposed the use of EPI for imaging diffusion

at a summer school in Italy in 1986, realizing that its

ability to acquire images as a snapshot would avoid

these artifacts. At NIH I was busy implementing EPI

on all the scanners available. This worked best on a

2.0T GE animal scanner fitted with an actively-

shielded gradient coil that had recently been

developed by the team at Schenectady mentioned

earlier. LeBihan and I were able to produce excellent

maps of diffusion in cat brain. The hope I then shared

with LeBihan was that brain MR images with

diffusion weighting would not only reveal basic water

diffusion parameters, but would also provide a means

for mapping rates of cortical blood flow and thus

brain activity.

It turned out that the scanner’s limited sensitivity

ruled out this latter possibility, because flowing blood

contributes only about 2% of the MRI signal.

However, the by-product of this enterprise was the

method of diffusion-weighted EPI that has now

become standard across the world for clinical

diffusion imaging. In 1989, in order to image the

human head, I implemented this method on a GE

Signa clinical MRI scanner. I designed a local, single-

axis head-only gradient coil which I arranged to be

constructed by NIH engineers in the Biomedical

Engineering and Instrumentation Program. When

driven by a conventional Techron power audio

amplifier, this coil could generate very strong linear

The Three Worlds of MRI

gradients along the z-axis, which could be rapidly

switched, sufficient to enable EPI and to provide

excellent diffusion-weighting of the signal. The coil

was light enough to be portable and could easily be

installed in the GE Signa 1.5T scanner housed in the

NIH NMR Imaging Center. To adapt the gradient

power supply to its new load, only a small control

board of electronic components needed to be fitted.

Two GE engineers, Bob Vavrek and Joe Maier,

based at GE’s MRI factory at Waukesha, Wisconsin,

had already been attempting to implement EPI on a

GE Signa scanner, and had written some of the

sequence software, with the tacit connivance of their

line manager. This was a classic instance of a

‘skunkworks’, far from the official job description of

the employees concerned. However, they had had no

experience of the practical details required to produce

good images, and no access to the gradient coil

hardware needed to give sufficiently powerful rapidly

switchable gradients. I was introduced to Maier and

Vavrek by an MRI scientist called James McFall, who

had previously worked for GE. In several visits to

Waukesha, I passed on to them my detailed

experience from my time with Peter Mansfield in

Nottingham and further work at NIH, and we worked

together to perfect the imaging sequence code (Turner

1990). Once I had it working at NIH, I took my coil

to their lab in Waukesha.

It is an interesting reflection that despite the rapid

success of this unofficial project in implementing

EPI, GE’s senior MRI management decided that the

company would instead focus its efforts on obtaining

the EPI technology developed by ANMR, which was

held by its management as a zealously guarded secret.

GE progressively bought a larger and larger share of

ANMR and undertook the responsibility of marketing

the GE/ANMR hybrid scanner, hoping that eventually

ANMR would reveal everything. Meanwhile GE’s

gradient technology continued to improve, and by

1994 their sequence development engineers were able

to implement EPI in a more conventional way on their

standard Signa 1.5T scanner. This implementation had

serious limitations, however.

Meanwhile, diffusion-weighted EPI (DWI) was

being adopted in several other labs around the USA.

Michael Moseley, working at the University of

California at San Francisco, found that rat brain

ischemia showed up strikingly in such images. This

led to DWI’s rapid clinical uptake worldwide as a

means of stroke diagnosis and evaluation. Moseley

also found, working with celery stalks and then with

rat brain, that DWI could be used to detect very

clearly the alignment of fibres in tissues, such as in

the white matter of the brain. This discovery has

primed an immense field of enquiry directed at

exploring the human brain’s connectome, the pattern

of connections across the brain that underpins

cognition and thought.

So it was that the combined efforts of physicists,

radiologists and engineers broke open entirely new

areas of diagnostic medicine and the scientific

understanding of how the brain’s circuitry is

connected. Without the recognized status of Le Bihan

in the field of radiology, evidenced by his invitation

as a speaker at the 1989 annual meeting of the

Radiological Society of North America, it is entirely

possible that measurement of water diffusion would

not have caught on as a clinical methodology in that

conservative and sometimes complacent discipline.

And without that clinical uptake it is hard to see how

DWI could have come to play such a major role in

neuroscience.

5) Implementation of EPI on a Clinical Siemens

Scanner by Schmitt at Erlangen in 1987-92

In the early 1980s, engineers at the major

multinational companies investing in MRI

technology took little note of echo-planar imaging,

because the spatial resolution was obviously poorer

than desirable and the hardware demands were much

greater than the first commercial scanners could

provide. However, within one Siemens development

laboratory in Erlangen, Germany, the engineer Franz

Schmitt and his colleagues realized that this much

faster method might have some potential, and they

began to implement it in 1987. A group of six

engineers led by Franz Schmitt was brought together

within the Siemens Medizintechnik basic R&D

group, and they considered all aspects of the

challenge, from gradient coil development to imaging

sequence coding. Encouraged by the early ANMR

results shown by Rzedzian and Pykett in 1987, they

were able to produce their first images later that year,

using a home-built single-axis head gradient coil

fabricated in another Siemens workshop (Cohen

2012). Interaction with Peter Mansfield’s lab at

Nottingham was soon established. Initially the main

target was fast cardiac imaging, as it was in

Nottingham, but one of Mansfield’s PhD students, an

MD named Michael Stehling, joined Schmitt’s team

from 1990 and 1992 and encouraged the use of EPI

for neurological investigations.

The striking results of this independent foray into

making EPI work on a commercial MRI scanner were

a thoroughly well-engineered robust implementation

with performance superior to that of the GE

counterpart in several respects, and an enduring

collaboration with me, which later bore rich fruit.

BIOSTEC 2024 - 17th International Joint Conference on Biomedical Engineering Systems and Technologies

6) Discovery of BOLD contrast fMRI between

1990 and 1992, at Bell Labs, Minneapolis,

Boston and NIH

It was Linus Pauling who recognized in 1936 that

deoxygenated blood has a higher magnetic

susceptibility than oxygenated blood. However, it

was not until 1990 that it became clear that this

difference would affect MRI brain image intensity, as

shown by Seiji Ogawa and his team of biophysicists

working with rats at AT&T Bell Laboratories in

Murray Hill, New Jersey. The key to this discovery

was the use of an imaging sequence that did not use

spin echoes, which largely remove the effect of the

magnetic field inhomogeneities caused by the

presence of deoxyhaemoglobin. When Ogawa

changed the animal’s breathing gas the MRI contrast

changed. He used the relatively slow multi-pulse

technique known as FLASH (Fast Low-Angle Shot)

to obtain images clearly showing the cortical veins.

He dubbed the phenomenon BOLD, standing for

Blood Oxygenation Level Dependent imaging.

Having heard Ogawa present his results at the annual

meeting of the Society for Magnetic Resonance in

Medicine (SMRM) in 1990, I realized that a time

course of single-shot EPI images would provide much

better temporal resolution of any changes in blood

oxygenation, and I quickly performed a series of

experiments with cats on the 2T animal scanner at

NIH mentioned earlier. The normal breathing gas was

changed to pure nitrogen for 60 seconds while a brain

EPI image was obtained every 3 seconds. The results

were dramatic. The image of the cat brain grey matter

darkened by over 20%, and recovered and over-shot

as soon as the normal breathing gas was restored.

When I showed these results to an MGH researcher,

Ken Kwong, at a meeting in Washington in April

1991, and suggested that we could try such an

experiment in humans with a breath-hold, he became

very interested. He told me that he would try the

experiment, using the hybrid GE/ANMR scanner and

giving the volunteer subjects periods of visual

stimulation. By May 1991 he had the first positive

results. In August the MGH Radiology Department

head, Tom Brady, presented the preliminary results at

that year’s SMRM meeting, which caused a

sensation. For the first time in history, the activity of

the human brain in response to a sensory stimulus

could be mapped entirely noninvasively.

The history of this discovery has been described

in many publications. At NIH I worked rapidly to

obtain the hardware needed to try out this effect at the

higher field strength of 4T, using a scanner that had

been donated to NIH by GE as surplus to their needs.

By February 1992 I was getting good results, with the

help of my postdoc Peter Jezzard. Soon I was

bombarded by requests from researchers at the

National Institute for Mental Health (NIMH) who

wanted my help to explore several fundamental

questions of localized brain function (Turner 2012).

In this discovery of BOLD contrast and its initial

application to imaging of human brain, only

physicists were involved. This development took

place without the need for advice or assistance from

MRI industry engineers or radiologists.

7) Establishment of Routine Functional MRI

Scanning for Studies of Cognition at

Functional Imaging Laboratory, Queen’s

Square, London in 1995

Only three years after its discovery, abstracts

describing research using BOLD contrast to observe

functional activity in human brain started to appear at

the annual meeting of the Society for Neuroscience,

in 1993. I was a co-author of four of them. At that

time, most of the research which explored functional

localization and integration in the human brain was

performed using positron emission tomography and

electroencephalography (EEG). The first of these was

invasive, using a radioactive tracer (oxygen-15 water)

injected into the bloodstream, and had a poor

resolution of about 5 mm. Furthermore, the radiation

dose was such that volunteer subjects were permitted

to participate only once in their respective lifetimes.

EEG is entirely non-invasive, and provides excellent

temporal resolution, but the electric fields due to

neural activity that can be detected on the scalp have

experienced significant attenuation as they pass

through the brain tissue and skull, and their sources

are difficult to localize without making speculative

assumptions.

BOLD fMRI thus comprised a scientific

revolution. Human subjects could be scanned again

and again, and activity in their brains could be

observed with a spatial resolution of one or two

millimetres, throughout the entire brain volume. The

main limitation arose from the fact that the BOLD

functional signal depicted changes in blood

oxygenation that were a downstream consequence of

neuronal electrical activity, an indirect measure, a

much slower counterpart with a time scale of seconds

compared with the millisecond timescale of neuronal

firing.

My introduction of the techniques of EPI, and thus

DWI and BOLD functional MRI, to the world-class

brain researchers at NIH and NIMH resulted in

several new research programmes and plenty of

publications. It also led to the offer of a professorship

at the Institute of Neurology in London, to become a

The Three Worlds of MRI

co-founder of the first purpose-built laboratory in the

world to study human cognition using brain mapping

techniques, funded generously by the Wellcome

Trust.

This laboratory, known as the Functional Imaging

Laboratory (FIL), included a PET scanner and small

cyclotron for generating oxygen-15, originally

conceived as the mainstay of the planned brain

research. The decision had already been made to

install a Siemens MRI scanner, based on the advice of

an excellent MRI scientist already working in

London, David Gadian. But at that time it was still

uncertain how useful functional MRI might become,

and it was my challenge to establish techniques that

were robust and user-friendly. Most of the research

team recruited by my lab director colleagues Richard

Frackowiak, Ray Dolan, Karl Friston, Chris Frith and

Cathy Price were psychologists, psychiatrists and

neurologists by training. They had little or no

experience of the technology of MRI.

I returned to the UK in late 1993 in order to assist

the preparations for the new lab and to establish fMRI

data analysis strategies with Karl Friston, the

computational neuroscientist. The FIL opened its

doors in 1995, with a 2.0T MRI scanner and closed

cycle helium liquefier. By 1998 a stream of high-

quality research papers was flowing from this

laboratory, and by 2000 it was recognized

internationally to be the most influential research

institution in the growing field of imaging

neuroscience. One of the keys to this success was the

excellent support which I experienced from Siemens

engineers based in Erlangen, Germany. Before the

scanner was installed in 1995, I spent several weeks

in Erlangen getting to know the development

environment and the specific engineers who might be

helpful to the FIL’s research needs. Of particular

importance were Franz Schmitt, a physicist and

engineer, and Edgar Mueller, an applications

manager, who became personal friends and ensured

that our scanner always had the best hardware

available. With this excellent partnership support, I

and my small team of physicists at the FIL were able

to introduce standards of data format, quality,

processing availability, and reliability which enabled

the lab’s cognitive science researchers to explore a

wide range of profound questions regarding

perception, cognitive control, emotion, memory and

consciousness (Turner 1998). Our success prompted

a growing number of imaging neuroscience labs

around the world to invest in Siemens scanners, thus

amply rewarding the company for their special

attention to our research needs.

8) Development of NextGen 7T MRI Scanner,

Berkeley, by David Feinberg and Colleagues,

2022

For my final example of fruitful interactions between

physics, engineering and radiology, I have chosen the

recent remarkable combination of skills and expertise

leading to the NextGen 7.0T MRI scanner now

installed at the University of California at Berkeley.

From its inception in 1995, for the next 20 years,

imaging neuroscience using fMRI made only

incremental improvements in spatial resolution and

quantitative accuracy, largely because the preferred

analysis strategies Sused spatial smoothing of the

images in the process of aligning and averaging

results from sufficiently large cohorts of experimental

subjects. Thus more detailed exploration of the

patterns of brain activity associated with specific

tasks was neither feasible nor desired. It was not until

2010, when Joseph Polimeni and his colleagues at

MGH showed that fMRI at 7.0T could begin to

resolve brain functional activity at the level of cortical

layers (Polimeni 2010), that interest grew in the

possibility that fMRI could distinguish between top-

down and bottom-up signalling by means of their

relative position within the cortical thickness, and

thereby add greatly to the causally realistic modelling

of brain function. Progress was greatly helped by the

use of 7T MRI scanners, introduced by Siemens, GE

and Philips in the early 2000s, which provide a much

greater signal-to-noise ratio than the usual clinical

scanners at 1.5T and 3.0T. In 2011 at the Max-Planck

Institute for Human Cognitive and Brain Science,

Leipzig, again at 7.0T field strength, Trampel and

colleagues (Turner 2016) were able to demonstrate a

difference between layers of motor cortex involved in

motor ideation compared with actual finger

movement.

By 2018 it was clear to a Berkeley physicist and

radiologist, David Feinberg, one of the pioneering

developers of MRI in the 1980’s, that the performance

of available 7T MRI scanners had not yet been

optimized, and that major improvements in

sensitivity could be made by radical redesign of two

of the hardware components. These were the gradient

coils and the radiofrequency coils. He was successful

in obtaining a grant to develop such a novel system of

$14M from the National Science Foundation and

NIH, and work could begin. The gradient coil

development group at Siemens, led by Peter Dietz,

enthusiastically took up the challenge of designing

and building a far more powerful head-only gradient

coil, and Bernhard Gruber, working at the RF coil

development lab of the Athinoula A. Martinos Center

for Biomedical Imaging, in the Department of

BIOSTEC 2024 - 17th International Joint Conference on Biomedical Engineering Systems and Technologies

Radiology, Massachusetts General Hospital

undertook the building of multi-element RF coils with

unprecedented performance. The results of this

remarkable collaboration are described in a very

recent paper in the journal Nature Methods (Feinberg

2023). A massive improvement in spatial resolution

has become available on a scanner which is still

relatively easy to operate and has excellent properties

in regard to human imaging.

2 SUMMARY AND

CONCLUSIONS

I have described a series of breakthrough

technological and biophysical developments taking

place over a period of 37 years. Each of these has been

fundamental to the next one, and to the current state

of the art in imaging neuroscience.

A set of take-home messages can be inferred from

these examples, as follows:

a) In fields like MRI, the inventiveness of

physicists is vital but also greatly needs the

amplification factor of radiological interest to

make any change to standard equipment and

practise. Physicists need to learn to think and

speak like radiologists.

b) MRI scientists are strongly encouraged to

develop personal links based on shared

research interest and mutual respect with

engineers in the companies that provide their

equipment.

c) Wise management is required when the skills

and objectives of all three worlds, physics,

radiology and engineering, need to be

combined. Interests and rewards need to be

carefully balanced.

d) Large companies often benefit by quietly

encouraging the independent enterprise of

their more ambitious engineers.

e) The limiting factor for scientific discovery

should never be the inadequacy of the

equipment, but only the creative imagination

of the innovator.

REFERENCES

Cohen MS, Schmitt F. Echo planar imaging before and after

fMRI: a personal history. Neuroimage. 2012 Aug

15;62(2):652-9.

Feinberg DA, Beckett AJS, Vu AT, Stockmann J, Huber L,

Ma S, Ahn S, Setsompop K, Cao X, Park S, Liu C, Wald

LL, Polimeni JR, Mareyam A, Gruber B, Stirnberg R,

Liao C, Yacoub E, Davids M, Bell P, Rummert E,

Koehler M, Potthast A, Gonzalez-Insua I, Stocker S,

Gunamony S, Dietz P. Next-generation MRI scanner

designed for ultra-high-resolution human brain imaging

at 7 Tesla. Nat Methods. 2023 Dec;20(12):2048-2057.

Kwong KK, Belliveau JW, Chesler DA, Goldberg IE,

Weisskoff RM, Poncelet BP, Kennedy DN, Hoppel BE,

Cohen MS, Turner R, et al. Dynamic magnetic

resonance imaging of human brain activity during

primary sensory stimulation. Proc Natl Acad Sci U S A.

1992 Jun 15;89(12):5675-9.

Ogawa S, Lee TM, Kay AR, Tank DW. Brain magnetic

resonance imaging with contrast dependent on blood

oxygenation. Proc Natl Acad Sci U S A. 1990

Dec;87(24):9868-72.

Pauling L, Coryell CD, The magnetic properties and

structure of hemoglobin, oxyhemoglobin and

carbonmonoxyhemoglobin. Proc Natl Acad Sci USA

1936 22: 210–216.

Polimeni JR, Fischl B, Greve DN, Wald LL. Laminar

analysis of 7T BOLD using an imposed spatial

activation pattern in human V1. Neuroimage. 2010 Oct

1;52(4):1334-46.

Rzedzian RR, Pykett IL. Instant images of the human heart

using a new, whole-body MR imaging system. AJR Am

J Roentgenol. 1987 Aug;149(2):245-50. doi:

10.2214/ajr.149.2.245.

Stehling MK, Howseman AM, Chapman B, Coxon R,

Glover P, Turner R, Ordidge RJ, Jaroszkiewicz G,

Mansfield P, Morris GK, Dutka D, Worthington BS,

Coupland RE. Real-time NMR imaging of the heart:

observation of coronary vessels. Lancet 1987; ii: 964-

965.

Turner R, Le Bihan D, Maier J, Vavrek R, Hedges LK,

Pekar J. Echo-planar imaging of intravoxel incoherent

motion. Radiology. 1990 Nov;177(2):407-14.

Turner R, Le Bihan D, Moonen CT, Despres D, Frank J.

Echo-planar time course MRI of cat brain oxygenation

changes. Magn Reson Med. 1991 Nov;22(1):159-66.

Turner R. Gradient coil design: a review of methods.

Magnetic Resonance Imaging 1993; 11: 903-920.

Turner R, Howseman A, Rees G, Josephs O, Friston KJ.

Functional magnetic resonance imaging of the human

brain: data acquisition and analysis. Exp. Brain Res.

1998; 123: 5-12

Turner R. The NIH experience in first advancing fMRI.

Neuroimage. 2012 Aug 15;62(2):632-6.

Turner R. Uses, misuses, new uses and fundamental

limitations of magnetic resonance imaging in cognitive

science. Philos Trans R Soc Lond B Biol Sci. 2016 Oct

5;371(1705).

The Three Worlds of MRI