Summary: UCLA researchers in the Department of Radiology have developed a novel, automated approach for tumor assessment by determining RANO-compliant tumor diameters from tumor segmentation data.
Background: MRI tumor measurements are critical in both clinical trials and clinical practice. Traditionally, radiologists manually assessed tumors by measuring the perpendicular in-plane diameters on MRI slices, a process that is prone to operator variability, time-consuming, and resource-intensive. Software and AI-based tumor segmentation methods have become commonplace, enabling volumetric measurements that greatly improve accuracy and efficiency. However, measuring tumor diameters remains essential for defining a measurable disease, a prerequisite for volumetric assessment. Additionally, response assessments based on tumor diameters may still be insightful when comparing to historical diameter-based results. Current automated approaches to obtain diameters from segmentations have rudimentary measuring capabilities, and do not take segmentation “holes” into account. These approaches often fail to comply with the Response Assessment for Neuro-Oncology (RANO) standards, or the clinical guidelines for determining disease progression. This creates a limitation in the accuracy of current methods calculating diameters from AI-based tumor segmentation, as they often do not account for surgical cavities, necrotic tissues, or cysts. Thus, there is a need for an automated, RANO-compliant approach for reliable tumor diameter measurement, aimed at disease progression monitoring and response assessment.
Innovation: Researchers at UCLA have developed a novel method for extracting RANO-compliant diameters from pre-existing tumor segmentations. The invention fits geometric objects within the segmentation region to gather size-related information of the segmented tumor. For example, it detects the presence of “measurable disease” and maximal RANO-compliant in-plane diameters. The system integrates seamlessly with existing tumor segmentation software, leveraging preexisting data. In addition, the invention introduces a customizable degree of tolerance, in two forms: (1) a closing operation that fills internal or incomplete holes to ensure that tumor segmentation represents a solid, continuous structure and (2) an overlap threshold to account for misalignment and shape irregularity. Since it adheres to RANO standards, the invention’s outputs are both legally and scientifically valid for determining measurable disease, as well as lesion diameters for longitudinal monitoring. This RANO-compliant, automated tumor diameter measurement approach has the potential to transform efficiency, reproducibility, and validity of tumor assessment and complement existing AI tools.
Potential Applications: ● Tumor segmentation and measurement ● Integration with MRI segmentation tools ● Clinical trial assessment ● Routine clinical practice ● Regulatory submissions ● Drug development ● Ground truth for algorithm validation
Advantages: ● RANO Compliance ○ Legal and clinical validity ● Improved Efficiency ● Reproducibility ● Objectivity ● Integration with Existing Segmentations ● Improved Validity
Development-To-Date: First successful demonstration of the invention completed in April 2025.
Related Papers: Ellingson BM, Sanvito F, Cloughesy TF, Huang RY, Villanueva-Meyer JE, Pope WB, Barboriak DP, Shankar LK, Smits M, Kaufmann TJ, Boxerman JL, Weller M, Galanis E, Groot J, Gilbert MR, Lassman AB, Shiroishi MS, Nabavizadeh A, Mehta M, Stupp R, Wick W, Reardon DA, Vogelbaum MA, van den Bent M, Chang SM, Wen PY. A Neuroradiologist's Guide to Operationalizing the Response Assessment in Neuro-Oncology (RANO) Criteria Version 2.0 for Gliomas in Adults. AJNR Am J Neuroradiol. 2024 Dec 9;45(12):1846-1856. doi: 10.3174/ajnr.A8396. PMID: 38926092; PMCID: PMC11630866.
Reference: UCLA Case No. 2025-077
Lead Inventor: Francesco Sanvito, MD