I am a physician with an M.D., Ph.D. My Ph.D. is in biophysics and theoretical biology. I have been involved in ultrasound scanning and research for over 40 years. I have extensive experience in flow imaging including Doppler techniques, decorrelation flow imaging, volume flow, and perfusion imaging. I am particularly interested in quantitative blood flow evaluations. Most, if not all, medical imaging methods for evaluating blood flow are qualitative or semi-quantitative at best. By semi-quantitative, I mean measurements that have little or no physical meaning for estimating true blood flow, i.e. metrics quantified in units of volume/unit time. Our approach uses Gauss’s divergence theorem, which measures the total flux across a surface. The method is perfect for ultrasound which can generate a surface defined across ultrasound beams at a constant distance from the center of the transducer array. With such a surface, it is possible to make flow measurements that are angle, flow profile, and vessel geometry independent. The method is robust, could be implemented in real-time on medical ultrasound scanners, and has the potential of having a major impact on medical diagnosis and treatment, where true flow assessments presently do not exist. The one caveat is that the ultrasound acquisitions must be made in 3D. Yet, many clinical ultrasound scanners have probes that can scan in 3D, and if 3D data can be acquired, the data processing in some ways is actually simpler than standard 2D imaging. Presently, we have validated the method in flow phantoms and in animal models. Clinically, we have used the method to measure blood flow in umbilical cords, surgically implanted shunts that divert blood flow in livers of patients with cirrhosis, and total brain blood flow in neonates.
Our major efforts now are making partial volume corrections to our flow estimates. Partial volume issues arise at the boundaries of vessels, where portions of a beam are in and out of flow. This requires compensation in order to calculate the correct flux. Presently, we employ methods that use Doppler signal power variations to account for ultrasound beams being partly in and out of blood flow. The method requires a 100% blood standard to make the compensation. This method has limitations, primarily due to the need for a 100% reference. Because of this requirement, we are limited in the size of vessels in which we can estimate the flow. When vessels are small relative to the ultrasound beam diameter, we can’t get a 100% blood sample. Therefore, we are working on other methods which are more robust and do not require a 100% blood standard. These methods directly employ beam geometry and correlation techniques to assess partial volume effects. We would like to include AI in this process. Successful implementation of a robust partial volume correction would make it possible to measure blood flow in vessels with diameters smaller than the beam profile, which would be a major advance in medical diagnosis.
Left hand image shows an ultrasound transducer positioned over the anterior fontanelle of a neonate. The beam is illuminating the 3 vessels that deliver blood to the brain: left internal carotid artery (lICA), right internal carotid artery (rICA), and the basilar artery (BA). The right image shows cross-sections of the 3 vessels cut by a surface through which the total flux is measured to estimate total brain blood flow.
All of my research projects are related to medical ultrasound:
1) My primary research interest is the quantification of blood flow using ultrasound. It is a surprising fact that true blood flow measurements are virtually never made in medicine. Although blood flow is imaged in many ways using different techniques, the actual quantification of flow is not done even though the number of applications for such measurements is huge. If flow data can be acquired in 3D, ultrasound measurements of flow are straightforward to quantify. Our method uses Gauss’s divergence theorem, and the results are angle independent, flow profile independent, and vessel geometry independent. We use the divergence theorem to estimate the total flux across a surface designed normal to the ultrasound beams in 3D. Using this method, real-time flow measurements are technically possible, and volume flow estimates could be made directly on a clinical ultrasound scanner. As mentioned, the clinical implications of such measurements are enormous. It would be possible to routinely measure such consequential parameters as cardiac output, blood flow to the brain, kidneys, liver, direct measurements of perfusion to tumors before and after treatment. The list is virtually endless.
2) Again, related to blood flow, we have developed a method for assessing placental structure and function in a very simple way. The placenta is crucial for the survival of a fetus in utero. It provides the fetus with oxygen and nutrients from the mother, and it reciprocally facilitates waste and CO2 removal from the fetus. Each of these is intimately related to blood flow. A particular problem related to the placenta is that it has two blood supplies, one maternal and one fetal, and it has been difficult to assess them simultaneously. These two blood supplies deliver blood to the placenta, but they don’t mix, and so typically at least two separate measurements must be made to evaluate placenta blood flow. For instance, for ultrasound, maternal flow is evaluated through Doppler studies of a uterine artery. While on the fetal side, flows to the placenta are assessed through Doppler studies of the umbilical artery.
We discovered that we could detect signals from both the maternal and fetal sides of placental flow in the envelope of the umbilical venous waveform. The umbilical vein arises in the placenta and delivers nutrients and oxygen to the fetus. The blood in the umbilical vein is fetal, and it receives blood directly from the fetus through a capillary network. This network lies within the chorionic villi in the placenta which are surrounded by maternal blood. However, as I said, the two blood supplies do not mix. Yet, we reasoned that the Doppler flow signal from the fetal blood within the umbilical vein would be modulated by signals from the mother and fetus based on their heart rates. The maternal heart rate is approximately 1/2 the fetal heart rate, and in any individual case, we know what these heart rates are. We can monitor the maternal pulse rate and sample the fetal “pulse” through Doppler signals from an umbilical artery. By performing a Fourier transform, we can see signals at these frequencies in the umbilical venous waveform envelope. Preliminarily, we have seen that the relative amplitudes of the maternal and fetal peaks in the Doppler power spectrum can discriminate between normal pregnancies and those with fetal growth restriction and pre-eclampsia. Although this analysis is very new, it holds promise for a way to directly evaluate placental function in a simple and benign way.
I was a premed, chemistry major in college. I had a particular interest in math and physics, although medicine ultimately seemed like a logical choice for me. So, when I went to medical school, I entered an M.D./Ph.D. program at the University of Chicago. I received my M.D. in 1974 and my Ph.D. in Biophysics and Theoretical Biology in 1977. For my Ph.D. research, I worked in a lab that was a mixture of biologists, computer scientists, and physicists all focusing on the study of a simple “developing” organism, Dictyostelium discoideum, which is a type of slim mold. The lab was dedicated to understanding how D. discoideum cells communicated and organized into multicellular structures. I made and analyzed time lapse movies of their development using a very early device that permitted computer analysis of these movies. Given that image analysis was a fundamental part of my research, it made sense to go into Radiology as a medical specialty. In fact, during my Radiology residency, I used the same computer analysis system to produce 3D angiograms of the brain, which I overlaid on computed tomograms to show blood vessels in the brain. These were very crude, to say the least, but they were unique displays of brain blood vessels at the time.
During my Radiology residency, I was strongly influenced by a Radiology faculty member, James Bowie, M.D., who specialized in ultrasound imaging. I began to focus on ultrasound, and I have spent the rest of my clinical and research life in ultrasound. My first ultrasound project was related to neurosurgery operations. Ultrasound is the perfect tool for guiding procedures; it is safe, easy to use, and real time. Neurosurgery was a great target. Once the bone is removed, both the brain and spinal cord are beautifully imaged with ultrasound. Neurosurgeons have to be very precise with small mistakes having huge consequences. Using ultrasound, they could actually see where they were going. Ultrasound guidance for their operations was a revelation, and the method is still used today largely unchanged from our initial work.
Although I loved the University of Chicago, I really wanted to collaborate with a physicist who specialized in ultrasound research, and there was none at the U of C. However, Michigan had a great one, Paul Carson. Therefore in 1984, I came to Michigan to work with Paul. We have been close collaborators for now over 40 years.
Since coming to Michigan, I have worked on multiple ultrasound related projects with Paul and others, notably Brian Fowlkes, Steven Pinter, and Oliver Kripfgans. These projects included blood flow evaluations such as volume flow, decorrelation flow imaging, contrast agents, ultrasound backscatter assessment, and tissue elasticity measurements to name a few. Among the most notable is something called power Doppler, which we as a group optimized for clinical use. When we started our work, power Doppler was available on some machines, but it was never used and was buried deep in the bowels of the hardware. Now power Doppler is a standard flow imaging mode. It is routinely assigned its own button on an ultrasound machine’s control panel, and our initial paper describing power Doppler now has over 1200 citations.
We are now looking at AI as a means of partial volume correction of volume flow measurements. Hopefully, when we’re done, it will also be assigned its own button on an ultrasound console.
As I said above, my aim is to make blood volume flow measurements a routine part of the medical diagnostic and therapeutic armamentarium. The ramifications of such a measurement are huge and game changing. I was the head of medical ultrasound in Radiology for about 40 years when I retired in 2017. I am now committed to seeing volume flow implemented on clinical ultrasound scanners. To me that’s a goal worth focusing on.
1) My main hobby is reading. I love to read. I’ll read anything as long as it is well written. However, I don’t generally like cookbooks (I don’t cook.), fantasy, science fiction, and most mysteries. Everything else is fair game.
2) I have a 2nd degree black belt in karate, although I’m way out of practice. I haven’t done karate in over 45 years. However, I have some great stories. First, I have no basic karate ability. I am very stiff jointed and can barely kick up to my waist. I also am a bit of a klutz. Although I haven’t investigated this in depth, I am the only person I know of who flunked every belt test, except for 2nd degree black belt, at least once. Belt color denotes the rank of the wearer, and in the style of karate I took, Shotokan, the awarded colors range from yellow to black. All beginners start at white belts. In order to get a yellow belt, one need only be able to stand and move without falling down. I don’t remember falling, but the judges must have thought what I did during my belt test was close enough that they failed me the first time. I have never heard of anyone else failing the yellow belt test. However, I persevered and after many years and a lot of effort, I received a second-degree black belt on my first try. In addition, I actually won the free sparring competition in a karate tournament. I won by having my nose broken. I was ecstatic. The only way I could have beaten my opponent was for him to hit me in the face. Our karate style was non-contact, and my opponent was the best fighter I have ever seen. In order to win against him, he had to hit me and foul out. Which he did, and I won. He was extremely upset, but I was delighted. I still have the trophy somewhere in our house.