New research aims to use AI to make cancer diagnostics for pets more available and affordable
A pioneering research project led by Ph.D. student Christina Pacholec at the Virginia-Maryland College of Veterinary Medicine aims to make significant strides toward revolutionizing lymphoma diagnosis in dogs.
Lymphoma, one of the most common cancers affecting dogs, typically needs multiple vet visits and tests for a correct diagnosis. Traditional methods, such as biopsies, are not only invasive but also costly and time-consuming. This project aims to harness the power of artificial intelligence (AI) to make cancer diagnostics more accessible and affordable for pet owners.
Leveraging AI to analyze cytological images — examining cell samples under a microscope — could enable early detection of lymphoma through less invasive, quicker, and more cost-effective methods.
Training an AI tool to find cancers
The project's initial phase is training an AI tool to distinguish between lymphoma-affected and healthy dogs by analyzing cytology images. To achieve this, the team will use over 10,000 lymph node aspirate images already on file at the Veterinary Teaching Hospital for AI training.
For the second phase, the research will focus on the pre-emptive identification of lymphoma. This involves modifying the original AI model using a technique known as transference, enabling it to detect much earlier emerging disease patterns. This adaptation could lead to more timely, earlier interventions and an enhanced approach to managing lymphoma.
Early detection of lymphoma can significantly influence treatment strategies, leading to improved outcomes for affected pets. The testing method, based on cytology, is minimally invasive and could be a low-cost alternative for pet owners. The research aligns with the Spectrum of Care Iniative, to increase the range of affordable diagnostic and treatment options available. This research not only has the potential to transform veterinary medicine but also could offer insights that could benefit human cancer diagnostics.
Becoming a specialist in veterinary pathology
Pacholec's path to her Ph.D. and specialization in veterinary pathology was fueled by a desire to contribute meaningfully to veterinary medicine. Her years of clinical work have provided her with perspective on patient care and diagnostics, enriching her research endeavors.
"The clinical work only helped because I have seen the differentials, I know the patients. ... So I would say daily I pull from that knowledge," said Pacholec.
Pacholec has undertaken multiple machine learning and image analysis courses to equip herself for the research, alongside participating in AI workshops and meetings.
"Christina's commitment to enhancing her skills in bioinformatics and computational pathology is crucial. Her proactive approach in attending these courses and workshops will undoubtedly contribute to her success in this research field,” said Kurt Zimmerman, Pacholec's mentor for her Morris Animal Foundation Training Fellowship, professor of pathology ,and associate department head of Biomedical Sciences and Pathobiology.
"Christina's diagnostic insight and understanding of disease pathogenesis are exceptionally well-developed. She is an incredible clinical pathologist,” Zimmerman said. “I am fortunate to have the opportunity to work with a graduate student of Dr. Pacholec's caliber."
Funding and support
The research is supported by a series of grants, including start-up internal funding from the college via the Veterinary Memorial Fund, contributions from the AKC Canine Health Foundation, and a significant grant from the Morris Animal Foundation. This financial backing has been crucial in advancing the project, enabling the team to undertake both the retrospective analysis and the prospective study. The commitment of these organizations to Pacholec's research underscores their support for upcoming researchers and the potential impact of her work on veterinary medicine.
Future directions
The hope is to build on this research advance veterinary diagnostics through AI, with aspirations to enhance early detection and treatment of lymphoma in dogs. The project aims to explore the AI model's capability for broader diagnostic applications, particularly in identifying minimal residual disease. "Can we detect minimal residual disease in these patients?" Pacholec said.
The project is dedicated to sharing its findings openly and working with computer science and engineering experts. "Collaborating would be a goal down the road," said Pacholec, pointing out the plan to partner with specialists from different fields.
"Why this research is important," said Pacholec, "is because at the heart of veterinary medicine is the desire to improve the lives of our patients. By developing affordable, accurate, and minimally invasive diagnostics, we can hopefully significantly impact pet cancer treatment, ensuring a better quality of life."
