Sachin Jambawalikar, PhD, Appointed Vice Chair of Physics, IT, and Clinical Informatics

Sachin Jambawalikar, PhD, assistant professor of radiology and vice chair of physics, information technology, and clinical informatics

The Department of Radiology is pleased to announce Dr. Sachin Jambawalikar's appointment as Vice Chair of Physics, Information Technology, and Clinical Informatics. This new position strengthens the collaboration between physics, IT, and informatics to benefit the department and patient care.

Jambawalikar is an assistant professor of radiology at Columbia University Irving Medical Center and chief of the Division of Physics. He will lead new initiatives designed to increase the efficiency and quality of data-driven tasks. "I'm excited to work more closely and efficiently with our IT team to facilitate solutions for the clinical and research needs of the department," he said.

Jambawalikar joined the department in 2011 and was appointed chief of the Division of Physics in 2014. He is board certified in MR Physics and Diagnostic Physics and is a Certified Imaging Informatics Professional (CIIP).

Q & A With Dr. Jambawalikar

What will your role be as Vice Chair of Physics, IT, and Clinical Informatics?

This vice chair role will help connect the technological needs of the department with the best solutions.

We have a physics division that looks at aspects of data acquisition, an IT division that looks at data after it's acquired, and then there are radiologists who need data analysis. Because I'm able to understand both the needs of the radiologists and the possible solutions, this role basically closes the loop between these groups.

What are the needs in the department that you want to address?

In a general sense, I see the department's needs falling into three categories: automation, custom IT solutions, and streamlining our IT-related processes.

Automation is a key aspect. For example, with QA or protocol dissemination to different scanners, there is a lot of manual work involved. There is a pressing need to automate these and other similar processes. The goal is to eliminate repetitive tasks that merely maintain the status quo without fostering any growth or improvement. For example, manually entering 1000 numbers daily may fulfill a job requirement but doesn’t contribute to substantial advancement or growth. The question becomes, how do you make the department grow by transforming these maintenance tasks into growth opportunities?

Secondly, custom IT solutions are vital for effectively addressing our unique needs. Unlike preexisting off-the-shelf solutions, which require us to adapt our processes to their limitations, custom solutions can provide simpler and flexible workflows.

Lastly, streamlining our IT-related processes will have a significant downstream effect, facilitating more effective communication and problem-solving within the department. Right now, people don't know who to reach out to for software queries or issues. One of the goals will be to create a well-defined and clearly laid out process for addressing hardware/software proposal management, so that requests will be researched and will lead to an answer in reasonable turnaround times. I think everybody will be happy to get a yes or no answer. That is what I'm trying to achieve.

What are some of the projects you are currently working on?

One of the things we're working on is making QA reports and things like that more efficient. Right now, we have different Excel files, which is highly inefficient. I'd like to implement a web-based database which can be customized and would give us records for compliance related information. We are looking into custom IT solutions and working with a vendor for this. We'll be able to create dashboards and also look at trends in the records, so if a physicist is doing QA for a particular machine, they can see if there's a trend in a particular value over time.

Another current project involves CT and MR protocols. We have protocols on a Teams site, which are written manually on the scanner, either by the technologist or by somebody else. There are a lot of issues when you do manual input and it's not efficient. With automation, using data driven technologies, we will be able to get this done more efficiently.

What future projects do you have in mind?

We are planning two key projects for the future: the creation of a research registry and providing server access to other department investigators.

We aim to develop a research registry that will house data from investigators in radiology, particularly those with expertise in specific domains. This registry will include scans with distinct radiological features, alongside any additional clinical information that investigators may have collected. All of the data, extracted from PACS, will be anonymized and securely stored on our local radiology servers. This structure will enable collaborations between investigators from our department and other departments, with the potential for a single principal investigator (PI) from radiology to work with several investigators using the same dataset. This setup not only facilitates inter-departmental collaborations, but also ensures the imaging data remains securely within the radiology department, thereby creating a mutually beneficial scenario for all parties involved.

Our second project aims to extend access to our servers to Columbia faculty and investigators from other departments. This will enable them to interact with our anonymized registry data directly and generate outcomes using their own machine learning or deep learning algorithms, while ensuring the data does not leave the radiology department. Currently, investigators who have developed an algorithm with a small number of cases often require a larger sample size for testing. They typically approach us to access a substantial number of cases from our department. We anonymize and provide this data for a fee, but subsequently lose visibility of its usage once it's transferred to their labs. Our new strategy ensures that we maintain oversight over the data, and it acknowledges the contributions of radiologists in our department who have annotated and extracted electronic health records (EHR) and other features for these cases.

How will you be involved in implementing AI tools into radiology?

AI is becoming more and more a part of the radiologists' workday. And as more and more AI solutions become available, we'll need to pick and choose different applications for different aspects of imaging and tools that also help us integrate EHR information. One of my roles will be to work with the CUIMC and NYP IT teams to ensure that we get tools that are actually useful.

A single solution might be a half solution, and I can bring my physics understanding to the table and explore whether there might be integrative approaches and solutions out there that meet all of our exact needs.