Students, faculty, and alumni partner with radiologists at the Cleveland Clinic
By John Walsh
Many good ideas start with brainstorming, and this one is no different.
Brainstorming led to John Carroll students making it easier for multiple doctors to achieve consensus about a difficult diagnosis more quickly by analyzing the same X-ray remotely. The software tool, which is a result of a project called Radiollaboration, is in the midst of a yearlong testing period – after which the product could be commercialized – at the Cleveland Clinic.
A couple years ago, Dr. David Piraino, head of the section of computers in radiology of the department of diagnostic radiology and staff member in the department of orthopedic surgery at the Cleveland Clinic, was thinking about advanced computer technology that could be implemented in a short period of time. He shared that general concept with Dan Palmer, Ph.D., professor of mathematics and computer science at John Carroll, and they, in turn, discussed it with Palmer’s students. The result – after four months of work – is a Web-based image viewer for X-rays – Radiollaboration.
Piraino suggested using digital and social media concepts because doctors want a tool that can function like Facebook to view X-rays and collect data.
“It’s nothing new in terms of creating digital images, but it’s a new way of coordinating and collecting data,” Palmer says.
Radiologists can use the software tool to mark the location of a problem on an X-ray, write comments, and submit them. Then other doctors can do the same and eventually look at all diagnoses made on the X-ray. Without the tool, radiologists have to look at each other’s diagnoses on separate X-rays. For example, one X-ray would be sent to five doctors for each to analyze, a linear process that takes awhile.
“Dr. Piraino wanted a way in which all doctors wouldn’t have to read an image if it wasn’t necessary,” says Nick Orlando ’11, one of the four students who worked on the project who now works for TMW Systems, a software manufacturer for the transportation and logistics industry in Beachwood, Ohio. “As an example, if the first three doctors’ diagnoses were all the same, the last two wouldn’t have to see it.”
Of note: The testing of the tool is being done with de-identified data or anonymized images so no one is violating the Health Insurance Portability and Accountability Act (HIPAA).
Kathy Lee, Ph.D., is helping students – specifically ones interested in health care – find their way.
Lee, the director of prehealth professions, helps about 350 students – freshmen through seniors – navigate academic waters in preparation for medical or other health professional schools. Many students enter college thinking medical school is in their future without realizing the plethora of health-care opportunities available.
“There are so many students interested in health-care professions,” she says.
The fall 2011 semester was Lee’s first in the new role she started Aug. 22. Lee, who has been at Carroll for six years as an instructor and visiting biology professor, continues to teach biology. Previously, Gwen Kinebrew, Ph.D., retired associate professor of biology, managed the prehealth profession responsibilities Lee assumed.
“I love interacting with students and helping them accomplish their goals,” Lee says.
Most students Lee mentors are biology or chemistry majors, but she also counsels economics majors, physics majors, and others. Any student who’s thinking about attending medical school or considering a profession such as dentistry, nursing, physical therapy, research, pharmacy, or any of the allied health professions, can ask Lee for guidance. Some students are interested in the health-care field but aren’t sure which direction to take.
“We talk about options and what they want to do,” she says, adding that sometimes students take a test in JCU’s Career Center for additional guidance.
If students want to pursue medical school, Lee discusses their class schedule, the Medical College Admission Test (MCAT), shadowing professionals, volunteering, medical school applications, and how medical school might be a good fit for them. She also guides students through the HPAC (Health Professions Advisory Committee) process, in which students complete an extensive application followed by an interview with Lee and two faculty members who counsel them to help prepare for medical school interviews. Lee writes a committee letter for the students to send to the schools to which they’re applying. She also coordinates visits by medical school representatives.
Lee teaches a health-care professions class in which professionals from various areas, such as radiology technicians, physician assistants, and anesthesiologist assistants, guest lecture.
“It’s a good way for students to see the opportunities in the health-care field,” she says.
Generally, freshmen don’t completely understand how the medical school admissions process works.
“Sometimes I need to explain to them they don’t get to choose the med school they want to go to,” Lee says, adding that most students apply to about 15 medical schools with the hopes of getting a couple of interviews.
Sophomores tend to ask more general questions, such as “What do I need to do this year to prepare for medical school?” Again, Lee suggests shadowing and volunteering, but also taking courses required by medical schools, such as biology, chemistry, calculus, and physics.
In addition to grade-point average, extracurricular activities, the MCAT, research, letters of recommendation, and personal statements all factor into the admissions process.
For example, Lee has been working with a physics major who wants to attend medical school. He thought about becoming a physics professor but changed his mind.
“He’s a bright kid who’s playing catch up,” she says, adding that students can major in any subject they want and still get into medical school as long as they have taken the class requirements and satisfy the other criteria to be a competitive applicant.
If students aren’t accepted to medical school, they have other options: graduate school for a biology-related master’s degree or a post-baccalaureate program that doesn’t result in a master’s but helps when reapplying to medical school, for example. Physician assistant and nurse practitioner programs are two other areas students can enter in addition to medical school.
Additionally, the University has partnered with neighboring Ursuline College to have select admission into its accelerated B.S.N. program. Students also may choose to enter a cooperative program with Case Western Reserve University that leads to a Doctor of Nursing degree. Carroll’s new interdisciplinary minor in Population and Public Health complements these programs.
Eventually, Lee will harness the data she’s been tracking to determine trends: How many students want to go to medical school; how many apply; how many get in; where do they go; and if they don’t go to medical school, what they do instead.
“We want to track these students even after they graduate so they can serve as a resource for JCU and the community,” she says.
One trend in information technology in general is crowdsourcing – the act of sourcing tasks traditionally performed by individuals to a group of people through an open call. The theory is a group of people can provide better answers than one person.
“It’s an interesting concept to apply to medical imaging,” Piraino says.
The tie that binds
The relationship between the Cleveland Clinic and John Carroll has been strengthened by three alumni – Roseann Spitznagel ’95, Jim Wetzel ’01, and John Hurt ’06 – who were computer science majors and worked or are working in IT departments at the clinic, as well as James Spitznagel ’93, senior network infrastructure engineer at JCU, and Carl Spitznagel, Ph.D., professor of mathematics and computer science. Roseann Spitznagel is the manager of software development and RIS systems, imaging informatics.
“When the clinic wanted to do collaborative work, John Carroll had its cheering section,” Palmer says.
A few years ago, Thomas Masaryk, M.D., chairman of the department of diagnostic radiology at the Cleveland Clinic, discussed ways to link health care and information technology with Roseann Spitznagel, and she suggested working with John Carroll. Spitznagel, in turn, worked with Palmer to develop a health-care computer science track, including the types of classes the University should offer. As part of that track, physicians and alumni in IT in the health-care field visit campus to teach students about their professions. The track is likely to evolve into a major.
“If you’re going to be in computer science, health care is the place to be in Cleveland,” she says.
In the classroom
At Carroll, Palmer teaches a software engineering course for computer science majors that provides experience about what it’s like to program in the real world. In the class, students choose what projects and teams they want to work on. Boban Josipovic ’11, now working for Parker Hannifin in e-commerce tech support; Anthony Lanese ’11, now a graduate student; Tom Drescher ’11, now working for Rosetta, an interactive marketing agency; and Orlando worked on the Radiollaboration project in the fall of 2010.
“Dr. Palmer gave us four choices, and I liked health care, so I worked on this project,” Josipovic says.
“I was drawn to the project because it allowed us to be creative about how the application should work,” Orlando says.
“The team gelled very well and made progress quickly,” Palmer says. “It was one of best projects we’ve had, and the team did a nice job.”
A core component of the project is open- source software called SVG-edit, a tool that creates and edits scalable vector graphic images to allow users to draw shapes and modify images. The team stripped out most of the features, left the highlighting (with different shapes) feature, and developed new text and image-saving capabilities. Other program features remained but were nonfunctional. The team wrote code for the social media component to allow doctors to write on X-rays and communicate their notes, allowing for collaboration.
“I was extremely lucky to be paired with this group because our individual talents were a good fit for the project,” Orlando says.
During the project, Piraino communicated with the students about once a week to suggest usability changes, such as how to display images, what to do about doctors’ overlapping marks, and whether everyone’s marks were weighted the same.
“In software development, there are times when what you said isn’t exactly what you meant; and when you build and test something and it doesn’t make sense for the users, we make changes,” Piraino says.
“During the project, Dr. Piraino liked where we were going,” Orlando says. “Although, he did specify that not every user should have the same level of authority, so we adjusted that.”
Toward the end of the project, the team – which worked 10 hours a week, two in the classroom and eight outside it – used pair programming, an agile software development technique in which two people use one computer, allowing more work to be done compared to each person working on his own computer.
“We needed to demonstrate the product could be used in a hospital,” Orlando says. “If decision makers at the clinic like the concept, they can develop it into a functioning application.”
How it works
Using the tool is easy. Doctors load an image on a computer screen and mark it using a mouse. Highlighting the same area on the screen multiple times creates a heat-map-like detection, adding the feedback that eventually reaches a threshold at which a consensus is reached. Radiologists then can email the image to other doctors or notify others the image is ready for their input. They don’t have to talk face to face because they can communicate through a Web browser.
The tool, which can be applied to analyzing CT scans and MRIs, wouldn’t be used routinely, only in difficult cases in which one doctor needs to ask several colleagues for input, Piraino says. A nonmedical example is the Where’s Waldo game in which it’s easier for several people to find him. A medical example is where something is wrong with a patient, and it’s easier for multiple doctors to pinpoint it.
“It improves diagnosis on more complicated radiology images,” he says.
“This software leverages social media by supporting remote, asynchronous interaction to simplify the task of coming to a consensus diagnosis,” Palmer says.
Here it is
When the team finished, it presented the project to Palmer and Piraino, who liked it. After the presentation, a few more changes, such as creating another image with the final diagnosis, were made.
“Dr. Piraino was thrilled we produced exactly what he wanted,” Josipovic says.
After the project, Josipovic added email functionality to the software tool.
“I was very impressed,” Piranio says. “Many of the software functions are the same as the ones the University of Pennsylvania used for CT dose tracking. Similar software is being used in the research community.”
Josipovic liked working in a real-life situation and enjoyed the freedom to develop a tool that met Piraino’s needs.
“I liked the teamwork,” he says. “I was excited the tool is something Dr. Piraino would actually use. The project was great exposure for seniors working with real life customers or clients.”
The team knew there was no possibility of owning the tool because it used open-source software and because of HIPAA requirements, which requires an involved certification process.
A look ahead
Doctors at the clinic still are testing the software tool and will continue doing so for about another five months to aggregate comprehensive results.
“I’m blinded to the results right now,” Piraino says, adding that testing for one year is the norm for multiple physicians to evaluate a computer program and determine a truth. “It’s taken us longer to evaluate the program than it did for the students to do the coding.”
Most likely more changes will be made to the software tool as a result of additional testing. For example, all JPEG images will have to be converted to the DICOM (Digital Imaging and Communications in Medicine) format, which is standard for all medical images. Additional security measures also would have to be implemented.
The next step – marketing and selling the product – is a big one.
“I’m unaware of any similar products,” says Piraino. “It’s a big jump to go from a research project to commercializing a product. We would involve other companies. It’s another whole level of processes.” JCU