Liver Lesion Tracking with AI: Highlights from RSNA 2024

Tracking liver lesions on CT scans is a critical yet highly complex task in radiological oncology. The intricate anatomy of the abdomen, coupled with the challenge of analyzing temporal changes in individual patients, makes lesion co-registration a significant hurdle. This complexity is compounded when evaluating changes in lesion size, shape, and enhancement patterns, often indicative of malignancy or treatment response.

At RSNA 2024 in Chicago, Dr Maximilian Schmidt, a key member of the radiology team at University Hospital Erlangen, presented our innovative study on a fully automated pipeline for liver lesion detection and tracking in abdominal CT scans. Conducted in collaboration with his team and Sycai Medical, the research addresses critical challenges in radiology, with a focus on improving diagnostic accuracy and streamlining workflow to improve patient care.

Dr Maximilian Schmidt during the presentation at RSNA 2024

Key Highlights of Our Study

1. Unique Lesion Identification:

Our approach assigns unique IDs to individual lesions, enabling precise co-registration across consecutive CT scans. This innovation reduces errors and provides consistent tracking over time.

2. Temporal Analysis:

The pipeline effectively tracks changes in size, shape, and enhancement patterns of liver lesions, offering critical insights into disease progression and helping differentiate between benign and malignant changes.

3. Unparalleled Accuracy:

Using state-of-the-art methods, the study demonstrated a mean accuracy of 100% in lesion co-registration. This performance underscores the reliability of our solution in handling complex abdominal anatomies.

Advancing Radiology Follow-up and Patient Care

By automating this labor-intensive task, our pipeline significantly enhances diagnostic workflows, reduces variability, and provides radiologists with the tools to better understand disease progression and treatment efficacy. Attendees at RSNA 2024 showed particular interest in how this innovation can streamline clinical workflows and improve patient care outcomes.

The feedback and discussions generated during the congress have been invaluable in shaping the next steps for our research and product development. We are excited to continue refining this technology and contributing to the advancement of AI in radiology.