Artificial Intelligence-powered Advanced Ultrasound for Early Diagnosis of Intraocular Melanoma

SUMMARY Choroidal melanoma is the most common primary intraocular malignancy in adults and a potentially fatal cancer that continues to be challenging in all respects. Choroidal melanoma can arise from preexisting choroidal nevus or de novo. Distinguishing early-stage choroidal melanoma from nevus can be challenging as they share similar ophthalmoscopic appearances when small. Up now, no single objective gold standard modality can successfully discriminate between these two entities. Formation of new and aberrant vascular networks are seen in de novo cases of choroidal melanoma as well as in the transformation of an existing nevus to melanoma. Our long-term goal is to combine novel artificial intelligence techniques with such new noninvasive tool for the diagnosis of early-stage choroidal melanoma. A secondary gain from such an imaging method will be assessment of treatment monitoring. Here, we propose to advance a new ultrasound-based technology, quantitative high definition microvessel imaging (qHDMI) that reveals tumor microvessels as small as 150µm and quantifies the microvessel morphological structures as new quantitative imaging biomarkers. Our goal is to combine novel artificial intelligence techniques with such new noninvasive tool for the diagnosis of early-stage choroidal melanoma. An advantage of the proposed qHDMI technique is that it does not require the use of contrast agents to produce high-resolution images of the microvasculature. We plan to address two aspects: Diagnose early- stage choroidal melanoma and assess the response to radiation therapy or transpupillary thermotherapy in patients with choroidal melanoma. The project includes 2 specific aims: Specific Aim #1 includes two sub-aims: Sub-Aim1.1: Determine the feasibility of the combined AI techniques with new qHDMI biomarkers and images for differentiation of choroidal melanoma from choroidal nevus and correlate with the clinical diagnosis based on all multimodality ophthalmic imaging together and gene expression profiling if clinically recommended. Sub-Aim 1.2: Longitudinal monitoring of the clinically diagnosed benign choroidal nevi in Aim 1.1 and determine the feasibility of qHDMI for detecting the transformation of choroidal nevus to Melanoma. Specific Aim #2: Determine the feasibility of qHDMI for treatment monitoring in patients with choroidal melanoma and correlate the results with the clinical diagnosis based on all multimodal ophthalmic imaging together. This proposal is the result of collaboration among leading experts in the field and benefits from the world-class research environment at the Mayo Clinic. Successful completion of this research will pave the way for novel, non-invasive, low-cost, and user-friendly technology for patients with ocular tumors. The qHDMI technique, enhanced with advanced AI- powered deep learning algorithms, holds strong potential for clinical translation and broad accessibility. We anticipate it will significantly benefit patients with choroidal melanoma by enabling earlier detection and timely treatment—ultimately improving the chances of preserving both vision and life. Project Number: 1R01CA304060-01A1 | Fiscal Year: 2026 | NIH Institute/Center: National Cancer Institute (NCI) | Principal Investigator: Azra Alizad (+1 co-PI) | Institution: MAYO CLINIC ROCHESTER, ROCHESTER, MN | Award Amount: $643,606 | Activity Code: R01 | Study Section: Special Emphasis Panel[ZRG1 NICD-H (01)] View on NIH RePORTER: https://reporter.nih.gov/project-details/11366800