System for Malignant Lymph Node Identification

Project Summary Accurate identification of malignant lymph nodes (LNs) is crucial for managing patients with solid cancers, as treatment planning—whether surgical, radiation, or systemic—depends on knowing the presence and location of malignant LNs. However, the malignancy status of small or less FDG-avid LNs is often difficult to determine from standard imaging alone, leading to the overtreatment or unnecessary resection of benign LNs, which significantly impacts patients' quality of life. To address this challenge, we have developed a high- performing artificial intelligence (AI)-based LN malignancy classification model. This model has been successfully implemented in two prospective Phase II clinical trials for head and neck cancer radiation therapy, allowing clinicians to target malignant LNs while sparing benign ones from therapeutic doses. The results were highly promising: the risk of solitary elective nodal failure was 0%, and patients experienced significantly improved quality of life compared to those treated with conventional whole-neck radiation therapy. Recognizing the growing clinical demand for accurate LN malignancy identification, particularly with the increasing use of immunotherapy, we have filed a patent application to protect our technology and established DAIgnostx LLC to support its commercialization. In this Phase I STTR project, we aim to develop the essential components of our LN malignancy classification system to facilitate widespread clinical adoption. Aim 1 focuses on creating a localization-guided LN segmentation model to provide accurate LN contours, enabling radiomic feature extraction critical for classification. Aim 2 involves developing two deployment options for the model: a cloud- based platform and a standalone desktop application, ensuring broad accessibility and adaptability. Additionally, we will rigorously evaluate the model’s performance following FDA guidelines with input from RMQ+. Successful completion of this project will pave the way for regulatory approval and the launch of the first dedicated product for cervical LNs segmentation and malignancy classification.