FY 2023 SBIR TOPIC 402 PHASE II. ENHANCE THE PERFORMANCE OF THE AI FOR LYMPH NODE DETECTION, SEGMENTATION AND MEASUREMENTS AND DEVELOP ADDITIONAL AI MODELS FOR MALIGNANCY CLASSIFICATION LEVERAGING MU

由于患者的预后和随后的治疗与疾病的分期有着内在的联系，因此正确确定淋巴结转移性疾病对于肿瘤患者的管理至关重要。在影像学上检测/分割淋巴结是一个繁琐、耗时的过程，它本身就受制于观察者内部/观察者之间的可变性。淋巴结的恶性分类提高了诊断评估和治疗计划。人工智能软件OncoAI在I期成功开发，可以自动检测和分割MRI和CT中肿大的淋巴结，并实现全自动RECIST测量。本二期提案的总体目标是进一步提高人工智能模型在淋巴结检测、分割和测量方面的性能，并利用多模态成像开发其他用于恶性肿瘤分类的人工智能模型。软件功能和可用性将进一步改善，以无缝整合到临床工作流程中。最后，将进行一项多机构验证研究，以证明OncoAI在临床实践中的安全性和有效性，并获得监管部门的批准。拟议的目标将为OncoAI建立强大的技术和监管基础，不仅有助于商业上的成功，而且对癌症治疗的临床实践产生更广泛的影响。

The correct determination of nodal metastatic disease is imperative for patient management in oncology, since the patient’s prognosis and subsequent treatment are inherently linked to the stage of disease. Detection/segmentation of lymph node on imaging is a tedious, highly time-consuming process that is inherently subject to intra-/inter-observer variability. Malignancy classification of the lymph node improves both the diagnostic evaluation and treatment planning. An AI software, OncoAI, was successfully developed in Phase I that automatically detects and segments enlarged lymph nodes from MRI and CT and enables fully automated RECIST measurements. The overall goal of this Phase II proposal is to further enhance the performance of the AI models for lymph node detection, segmentation, and measurements and develop additional AI models for malignancy classification leveraging multi-modality imaging. Software functionality and usability will be further improved towards seamless incorporation within the clinical workflow. Finally, a multi-institutional validation study will be conducted to demonstrate the safety and effectiveness of OncoAI in clinical practice and obtain regulatory approval. The proposed aims will set a strong technical and regulatory foundation for OncoAI and contribute to not only commercial success, but also broader impact to the clinical practice of cancer care.