Chapter 1: What's the use for biomedical engineers to learn about clinical practice?
By Yi Wang, PhD
What use is it for biomedical engineers to learn about clinical practice? This may be the first and one of most important questions for an engineering student or a practicing engineer interested in this book. This question can be partially addressed by examining the very definition of biomedical engineering (BME). I will use my experience as an engineer working on medical imaging technology to demonstrate how important it is for biomedical engineers to have some knowledge about medicine and clinical practice.
1.1 Biomedical Engineering (BME) and medicineClinicians, scientists, and engineers are all integral parts of the healthcare team and must work cooperatively in the long, challenging battle against disease. The challenges in translating basic research advances in life sciences from laboratory test tubes and mice to the clinical arena are formidable because test tubes and mice, though instrumental for scientific investigation, cannot fully replicate human disease. To address these challenges, the discipline of biomedical engineering (BME) has been formed to integrate engineering with life sciences and medicine. BME's objective is to develop medical technology and subsequently improve medical care.
Medical technology created to monitor, prevent, diagnose, control and cure a growing number of health conditions has greatly influenced medical practice and healthcare outcomes. Medical technology has become an integral and significant component of healthcare and has permeated into all major aspects of medical practice, including surgical procedures (angioplasty, joint replacements, organ transplants), diagnostic tests (laboratory tests, biopsies, imaging), drugs (biologic agents, pharmaceutics, vaccines), interventional devices (implantable defibrillators, stents, prosthetics) and support systems (electronic medical records and telemedicine). In fact, MRI (magnetic resonance imaging) and CT (computed tomography) scanning are regarded as the top medical innovations in recent times (Fuchs VR, Sox HC, Physicians' Views Of The Relative Importance Of Thirty Medical Innovations, Health Affair, 20(5): 30-42, 2001).
In order for biomedical engineers to be effective in their endeavor of improving medical technology, they must not only understand basic biological and engineering sciences, but also comprehend the clinical requirements of both patients and clinicians and the impact BME has on healthcare and society. Clinical experiences reveal how diseases affect patients, how diseases are diagnosed and treated in medical practice, how medical technologies are invented, disseminated and utilized in healthcare facilities, and how medical technologies influence medical practice, the economy and society.
1.2 Gap between intense and focused engineering study and busy clinical practice
I did my PhD study on MRI. Though we did research on the MRI scanners in the hospital and our office was in the hospital, we never spent time in the reading room to see how radiologists made diagnoses from MRI images. We were too focused on learning the complex spin dynamics underlying MRI and too busy with coding and testing new pulse sequences. Few of the new pulse sequences that we were developing would show enough promise to warrant the evaluation of the sequences by radiologists in a clinical setting. Consequently, very few of the PhD students working on MRI actually worked closely with radiologists.
Like all clinicians, radiologists are very busy. I do recall peeking into a reading room. It was dimly lit so that radiologists could read the films displaced on the huge panel of light boxes (nowadays, the reading room is full of large monitors that display the digital images, but it is still dim.) Radiologists were busy flipping through panels of images and making dictations, and interrupting them in order to gain their attention was difficult and made me feel uncomfortable. Though we were testing pulse sequences on the same MRI scanners that were being used in clinical practice, we gained access to the scanners only after the clinical hours.
My four years of PhD thesis work on developing MRI to image coronary arteries went by very quickly. Coronary MRI continues to challenge researchers and is still in the investigational stage 20 years after I wrote my thesis. Despite having completed four years of study, I did not have the chance to work with clinicians in evaluating my pulse sequence for imaging coronary arteries and I did not learn much about the use of MRI in clinical practice.
1.3 Clinical practice is a fertile ground for ideas of technical innovation
Following the completion of my PhD in early 1994, I continued to work on the development of coronary MRI in my postdoctoral training. Imaging the moving coronary arteries was a challenge that opened opportunities for many technical innovations. I took these opportunities, came up with many technical improvements, and generated many papers. However, my efforts could not make coronary MRI sufficiently robust for clinical practice. I was not sure what to do next in my research, which was quite disturbing for a postdoc looking for a job. To determine if I should stay in MRI, I needed to know something about the MRI market. I needed to know how MRI was used in clinical practice.
Fortunately, my postdoc advisor, Dr. Richard Ehman, is a great clinician, in addition to being a first rate physicist. Because of Dr. Ehman, I was able to spend some time in the radiology reading room. I was amazed by the many diseases, images, and the empirical or semi-empirical connections between them. While I admired radiologists' good memory, I realized that for a technique to be used in clinical practice, it has to be highly reproducible. A good idea that leads to a cute paper is not enough; a great idea is one that works in practice.
My brief exposure to the radiology reading room led me to reassess my own research work. Coronary MRI was a prime area for technical innovation, but what could bridge the gap between its research development and its incorporation in the clinic? Seeing this gap was both refreshing and frightening. The refreshing aspect would later define one R01 (a major type of NIH grants) project on coronary MRI when I became an independent investigator. The frightening aspect would prompt me to diversify applications of techniques developed by my research, such as imaging arteries other than coronary arteries, which would later define another R01 project. I would like to relate specific stories on how I developed time-resolved imaging and bolus chase techniques for contrast enhanced magnetic resonance angiography (CEMRA). CEMRA is a major method to image blood vessels in MRI without x-ray radiation. However, CEMRA faced two major problems in clinical problems: 1) clinicians had to guess the time of contrast arriving at a targeted body part to start imaging, but varieties in diseases make timing-guess difficult and CEMRA quality suffers with mistiming; 2) different body parts had to be imaged by repositioning the patients and repeating contrast injection.
The story of time-resolved imaging for CEMRA.
The tail end of the second year of my postdoctoral training was soon upon me, and I began to actively explore my job options. During my visits to reading rooms, I was very fortunate to meet with Dr. John Huston, a great neuroradiologist who was interested in MRI research. Dr. Huston performs neuro interventional procedures, such as placing stents to open carotid arteries, in addition to reading MRIs. Dr. Huston allowed me to observe his next interventional case. Though the sight of blood at the femoral artery puncture made me light-headed, I was fascinated by the use of imaging in guiding therapy, in addition to imaging based diagnosis. It was also amazing to see the speed of x-ray imaging (relative to MRI) and to witness clinicians' skill in performing treatment under limited anatomical information in x-ray fluoroscopy. Blood vessels were imaged using x-ray digital subtraction angiography (DSA), which was invented by my PhD thesis advisor, Prof. Charles Mistretta. DSA is based on serial or time-resolved imaging of a body part to capture contrast bolus passing through the vasculature. The subtraction of contrast enhanced 2D images (thick slab or projection without resolution along depth) by a mask image acquired prior to contrast arrival generates an image of the vessels containing contrast agent. Right after my trip to the interventional radiology suite with Dr. Huston, I realized that the same thing could be done in CEMRA to address the mistiming problem. As I looked into x-ray angiography more, I also realized that the moving table used in x-ray fluoroscopy to do bolus chase angiography in the lower extremity could be borrowed for MRI, which is the idea of the multi-station-stepping-table as described below. Immediately, I started to develop a time-resolved imaging approach for CEMRA, as well as mask subtraction. Soon, we published a paper on magnetic resonance digital subtraction angiography (MRDSA). MRDSA ushered in the field of time-resolved CEMRA that remains to be active for both clinical practice and scientific research.
The story of the multi-station-stepping-table (MSST) platform in MRI and bolus chase CEMRA
Near the end of my postdoc, I felt that I could do a lot in improving MRI for clinical practice and I could realize this by developing my own MRI research program in an academic center. In 1997, I started my own research lab at Cornell University. I worked closely with interventional radiologists Drs. Neil Khilnani and David Trost to develop the MSST platform for MRI to address the problem that many fields-of-view are needed to image a long territory, such as in the lower extremity. This MSST solution was particularly useful for imaging arteries in the legs following a single contrast bolus. Soon we developed the bolus chase CEMRA using MSST to image peripheral arteries from feet to abdomen, in a manner similar to x-ray bolus chase angiography. This was a very fruitful pursuit, as we published many papers and received an R01 grant. We were awarded several patents but they were licensed very cheaply. Still, it was gratifying to see our work become commercial products (Siemens' total imaging matrix, TIM) and be used in routine clinical practice.
1.4 The impact of a biomedical engineering idea has to be assessed in clinical practice
My career as a biomedical imaging engineer in an academic setting has benefited tremendously from collaborating with clinicians: I would learn about the unmet clinical needs and think about technical solutions to address these unmet needs; I would learn different ways to do things and incorporate them into what I was doing. This point has been illustrated in my stories as described above. I would like to exemplify another need for biomedical engineers to work with clinicians by evaluating and disseminating information about developed technology. Bringing new techniques to clinical practice may help improve patient care and is critical for generating a societal impact.
The issue of improving patient care with new technology is quite complex, involving many players besides engineers. Clinicians are needed to do clinical trials to demonstrate efficacy; business people are needed to invest into the product development and marketing; bureaucrats and lawyers are involved in FDA approval. Perhaps the most important initial step is obtaining preliminary clinical data to demonstrate the clinical utility of a new technology. When a new technique is developed specifically to address an unmet need in clinical practice, the clinical utility of the new technique is well defined, and clinicians are eager to perform evaluation to find out how much the new technique will improve their practice. Often we engineers may come up with ideas from insights in technologies, not from responding to unmet clinical needs. There are many cases where the clinical utilities are not clearly defined, and engineers have to actively seek out clinicians to identify any clinical use of their technologies.
To illustrate the necessity of engineers reaching out to clinicians to establish clinical utilities, I shall describe one example from my own experience. Recently, my lab has been working on solving the inverse problem from magnetic field to tissue magnetic susceptibility. The magnetic field can be estimated from MRI data. Tissue magnetic susceptibility reflects molecular electron polarization, an important piece of molecular information of the tissue, in the magnetic field of MRI. This field-to-susceptibility inverse problem is a fundamental problem in physics that has not been solved. We feel fortunate to have found an important physics problem to work on and we are excited that this problem is solvable in the context of MRI. About 5 years ago (2010), we came up the Bayesian approach with a reasonable solution and published our technique as quantitative susceptibility mapping (QSM). How would QSM benefit patient care? We have been struggling with this question ever since. QSM may be used to improve hemorrhage diagnosis. QSM may be used to improve visualization of targets in deep brain stimulation. QSM may be used to study multiple sclerosis. QSM may be used to assess brain iron accumulation that is associated with neurodegeneration in ailments such as Parkinson's disease and Alzheimer's disease. We are actively working with neurologists and surgeons to identify the clinical efficacy of QSM.
1.5 The purpose of this book is to help bridge the gap between engineers and clinicians
Getting your feet into the door of a medical center and approaching busy clinicians can be difficult and intimidating, as I experienced initially. This problem is partly caused by the engineer's lack of knowledge about medical practice. Here we try to ease this problem for engineers by introducing the basic clinical procedures, vocabulary, and culture.