Dennis did not have specific plans to pursue a career in academic medicine, much less in biomedical engineering, when he completed his MD at Johns Hopkins in 1935. He hoped to establish a private surgical practice in Minnesota, as his father had. This tentative plan was abandoned, however, during his surgical internship and residency in Owen Wangensteen's program at the University of Minnesota. Wangensteen had become the university's youngest chairman of surgery in 1930, and had quickly reorganized surgical training there to combine post-graduate surgical training (in both clinic and laboratory) with a tour of duty in a basic science department, culminating in a PhD. Dennis loved both the clinical work and the research, and, mentored by Wangensteen and physiology department chair Maurice Visscher, completed his PhD in 1940. His dissertation research on the physiological basis of appendix blockage built on his earlier work with Wangensteen on the origins of appendicitis. He joined the faculty that year as an instructor, moving up through the ranks to full professor in 1947. Though Dennis hoped to enlist during World War II, the medical school dean persuaded him that he could better contribute to the war effort as a medical educator.

Dennis soon became a well-regarded teacher, known for his dedication, conscientiousness, and inventiveness. One resident would later recall ". . . those days when an extraordinarily young-looking professor of surgery spoke up, usually from the back of the room, with great clarity, decisiveness, and frequently with a totally original idea. The room usually quieted and everyone listened with complete attention. [Dennis's] meticulous way of teaching, operating, designing, and performing research was an inspiration to all of us." Dennis contributed steadily to surgical practice during his early career, developing improved techniques to treat appendicitis, ulcerative colitis, bowel obstructions, and pancreatic cancer. These conditions would continue to draw his interest throughout his career, even as he joined the ranks of the first generation of cardiac surgeons after the war.

Like many surgeons and researchers, Dennis had invented various devices he needed for his work, e.g., measuring pressures inside the appendix, and muscular contraction in the appendix wall. For general surgery, he made other tools, including a simple apparatus to test the strength of suture materials. At the end of 1945, Dennis's mechanical aptitude prompted his former mentors, Wangensteen and Visscher, to suggest a new research project: developing a pump-oxygenator (heart-lung machine) that could take over a patient's circulation while surgeons worked on the heart. Dennis was delighted by this idea, although he had not considered it before--cardiovascular surgery was a largely unexplored frontier at the time. (Cardiology, in contrast, had developed into a distinct specialty before World War II, greatly aided by tools such as the electrocardiograph (EKG).) By 1945, the boundaries of thoracic (chest) surgery were expanding rapidly, often with dramatic results. In 1944 Alfred Blalock and Helen Taussig at Johns Hopkins had devised an arterial shunt operation that prolonged the lives of many "blue babies" born with heart defects. That same year, Robert Gross at Harvard and Clarence Crafoord and K.G.V. Nylin at Sweden's Karolinska Institut had each repaired a coarctation of the aorta (a congenital constriction of that large artery in the descending portion near the heart.) Dwight Harken's wartime success at getting pieces of shrapnel out of still-beating hearts inspired him to use the technique (after many animal trials) to correct mitral valve stenosis, a common outcome of rheumatic fever. (Prior to the advent of antibiotics in the late 1930s, rheumatic fever often followed streptococcal infections such as strep throat and scarlet fever. The infection caused stenosis, i.e., scarring and narrowing of the heart valve openings, particularly the mitral valve through which oxygen-rich blood coming from the lungs flows from the left atrium to the left ventricle. With their circulation thus restricted, such patients were often disabled by the lack of oxygen. Over a million Americans suffered from rheumatic heart disease at mid-century.) In 1948, Harken, working in Boston, Charles Bailey in Philadelphia, and Russell Brock in London, would independently perform successful procedures (mitral commissurotomy) to re-open the constricted valve. With the heart still beating, they made a small incision in the left atrium and, working by touch alone, opened the stenosed valves with a finger, sometimes fitted with a small blade. This was a "closed heart" procedure, carried out by opening the left chest cavity to gain access. It was clear, however, that cardiac surgery could advance only slowly until a way was found to stop the heart long enough to repair it, without depriving the body of oxygen. Some, like Wilfred Bigelow in Toronto, believed that controlled hypothermia might be used to decrease the metabolic rate so that the heart could be slowed or stopped for a few minutes without danger to the patient. The ideal solution, of course, would be a heart-lung substitute that could sustain circulation for as long as the surgeon needed.

When Dennis began work on his pump-oxygenator in 1946, his first step was to find out who else might be developing such a device. Earlier investigators--including surgical pioneer Alexis Carrel, and engineer/aviator Charles A. Lindbergh--had attempted to artificially circulate blood through individual organs with various devices. But only John Gibbon, at Jefferson Medical College in Philadelphia, had done substantial work on a heart-lung machine. Prior to World War II, Gibbon had maintained cats on artificial circulation for up to 25 minutes with a prototype machine. When Dennis went to see him, Gibbon welcomed him as "someone else in the world who didn't consider him an impossible dreamer." Gibbon had just returned from military service, and his equipment was still in storage, but he gave Dennis reprints of his work and the two men exchanged ideas and research results during the next few years, as they worked on their parallel projects. Dennis, assisted by surgical resident Karl E. Karlson, began his own work at Minnesota in 1947, with funding from the National Institutes of Health (NIH), which would continue to support this and other projects until 1972.

There were formidable technical challenges involved in replacing the body's basic blood pumping and oxygenating functions. Four to five liters of blood per minute had to be collected, exposed to oxygen, and infused under pressure back into the patient, for 30 minutes or more, all without damaging the blood cells or disrupting the blood chemistry. The most difficult component was the oxygenator, which had to maximize the blood's oxygen uptake, while avoiding both blood damage (hemolysis) and the formation of bubbles that would cause air embolism in the patient's circulation. Researchers experimented with a number of ways to maximize the blood surface area in an oxygen-filled chamber, including spreading a thin film of blood over fine steel mesh cylinders or discs, or running it through membranes. Dennis's model filmed the blood (with heparin added to prevent clotting) across a series of slowly rotating steel mesh discs, and collected it at the bottom of the oxygen chamber, to be pumped back to the patient. The pump component had to move blood through the apparatus at the proper volume and pressure, again with minimal hemolysis. Gibbon had been using a modified DeBakey roller pump, but Dennis and his colleagues found that a Dale-Schuster type bellows pump carried higher flows and caused less damage. Other challenges included keeping the machine's components disinfected--many of the new materials they employed (e.g. polyvinyl chloride and methyl methacrylate) couldn't be heat-sterilized; determining how much circulating blood pressure was actually needed; how to keep the blood from cooling too much as it went through the machine, and how best to reverse the anticoagulant effects of the heparin after the operation.

By early 1951, after four years of animal experiments, Dennis and his team believed that their pump-oxygenator was ready for a clinical trial. On April 5, they performed the world's first open-heart operation using artificial circulation. The patient was a six-year-old girl dying from an apparent atrial septal defect (a hole in the wall that separates the heart's two upper chambers.) The pump oxygenator worked well during the 40 minute procedure. However, there was unexpected bleeding from the heart wall as the heart was opened, which made it hard to see at first, and then the surgeons found that the patient's heart defect was much larger than expected; they were unable to repair it and the child died. Two weeks later, Dennis operated on a second child with a small atrial septal defect, but lost this patient when the technician operating the machine failed to turn on a level-control device in the collection reservoir; when the blood level in the reservoir got too low, air was blown into the patient's aorta, causing a fatal embolism. These demoralizing failures raised serious questions about the design of the machine, and Dennis did no further cases for several years. Dennis's experiences were not unique; failure was far more common than success during the early years of cardiac surgery, for many reasons. The heart was only beginning to be explored, and techniques just starting to be developed through trial and error. And the patients willing to risk such procedures were almost always in near-terminal condition, so they sometimes couldn't survive the stresses of an operation.

In the meantime, Dennis had been recruited to head the surgery department at SUNY Downstate. In August 1951, he and six members of his team, including Karlson, moved to Brooklyn, where they would eventually use their improved heart-lung machine for a successful operation in 1955.