AI-Enhanced Trial at Toronto General Seeks to Boost Lung Cancer Biopsy Accuracy

In a groundbreaking initiative, the University Health Network and NodeAI, a Hamilton-based medical AI startup, have launched a clinical trial aimed at improving the accuracy of endobronchial ultrasound (EBUS) lung cancer biopsies. This trial is being conducted at Toronto General Hospital, an institution pivotal in advancing the EBUS technique, which has revolutionized lung cancer diagnosis since its inception in the late 1990s.

EBUS, specifically EBUS-TBNA, allows for minimally invasive biopsies to determine if lung cancer has spread to lymph nodes, and it has largely replaced more invasive procedures. However, it still presents a significant challenge: about 40% of EBUS procedures yield inconclusive results due to human error, which varies by the operator's experience. These inconclusive results can lead to delayed diagnoses, unnecessary repeat biopsies, and worsen patient outcomes in a disease where timely intervention is crucial.

NodeAI's clinical trial intends to validate an AI algorithm capable of providing real-time predictions regarding lymph node malignancy during EBUS procedures, effectively reducing the variability of human performance. The AI analyzes ultrasound imaging to identify lymph nodes and predict whether they are cancerous, aiming to enhance biopsy accuracy.

The trial's involvement of Dr. Kazuhiro Yasufuku, a pioneering thoracic surgeon associated with the original EBUS advancements, adds credibility and expertise to this initiative. Dr. Yasufuku emphasizes the potential impact of the AI system to improve diagnostic precision and enhance patient care globally by integrating validated anatomical data and patterns recognized by experienced bronchoscopists.

NodeAI’s platform is designed for seamless integration into existing EBUS clinical workflows, utilizing a cloud-based interface that does not require hardware modifications. The AI performs real-time analysis during procedures, allowing less experienced operators access to expert-level diagnostics instantaneously, thereby addressing equity issues in patient care across different medical facilities.

The trial aims to enroll 100 patients over three months and will evaluate the AI's capability in efficiently processing EBUS imaging and providing accurate malignancy predictions for over 90% of the images collected during procedures. By comparing diagnostic yields from AI-assisted EBUS against standard procedures, the trial seeks to illustrate improvements in reducing inconclusive outcomes and operator variability.

Conducting this trial at Toronto General Hospital is seen as vital due to the institution’s status as the birthplace of EBUS technology, allowing for rigorous validation under the guidance of leading experts like Dr. Yasufuku. The successful data derived from this study could lead to significant advancements in lung cancer diagnostics.

Beyond the clinical trial, the broader implications of NodeAI's work include increasing access to expert-level EBUS technology across various healthcare settings, not just high-volume academic centers. With lung cancer continuing to be the leading cause of cancer-related deaths globally, innovations that enhance diagnostic accuracy are critical. A correct and timely diagnosis influences treatment pathways, including surgery, chemotherapy, radiation, or palliative care, making this trial a potentially transformative step in lung cancer management. NodeAI envisions a subscription-based model that makes their AI technology widely accessible, enhancing capabilities in both large hospitals and smaller community health centers.