NodeAI for Lung Cancer

(NodeAI Trial)

Overseen By Yogita S. Patel, BSc

Age: 18+

Sex: Any

Trial Phase: Academic

Sponsor: St. Joseph's Healthcare Hamilton

Disqualifiers: cN0 disease, peripheral tumors, others

No Placebo Group All trial participants will receive the active study treatment (no placebo)

Compensation: Varies

Number of Visits: 1

Duration: 1 day

100 Participants Needed

Trial Summary

What is the purpose of this trial?

Lung cancer is the leading cause of annual cancer deaths globally, more than breast, prostate, and colon cancers combined. The staging of chest lymph nodes (LNs) is a crucial step in the lung cancer diagnostic pathway because it aids in treatment decisions - whether a patient is a candidate for lung resection, chemotherapy, radiation, or multimodal treatments. Endobronchial Ultrasound Transbronchial Needle Aspiration (EBUS-TBNA) is the current standard for chest nodal staging for non-small cell lung cancer (NSCLC), and guidelines mandate that Systematic Sampling (SS) of at least 3 chest LN stations be routinely performed for accurate staging. Unfortunately, EBUS-TBNA yields inaccurate results in 40% of patients, leading to misinformed treatment decisions. This proportion is much higher in patients with Triple Normal LNs \[LNs that appear normal on computed tomography (CT) scans, positron emission tomography (PET) scans, and EBUS\], which have been found to have a \> 93% chance of being truly benign. This is because EBUS-TBNA is based on ultrasound, whose success highly depends on the skill of the person performing it (operator). When the operator makes an error, the entire procedure is jeopardized. This causes downstream delays in treatment due to repeated testing and ill-informed treatment decisions. Over the past decade, the investigator has been conducting a series of research studies and trials: the development and validation of the Canada Lymph Node Score (CLNS) - a surgeon-derived semi-quantitative measure of LN malignancy; an Artificial Intelligence (AI)-based version of the CLNS to predict malignancy; and a fully autonomous AI that learned to predict malignancy directly from ultrasound images, to introduce AI to the decision-making pathway in NSCLC. This resulted in the creation of an AI-powered software to predict malignancy in mediastinal LNs of patients with lung cancer. The software is currently housed in cloud storage and its applications are latent - which means that LN images must be uploaded to the software, and results are received at a future time. In its current form, the software is not ready for clinical application due to this latency. In this project, the investigator aims to build a point-of-care device which will house the software (NodeAI) and deliver real-time results to the surgeon, and this device will be tested in a clinical trial.

Will I have to stop taking my current medications?

The trial information does not specify whether you need to stop taking your current medications.

What data supports the effectiveness of the treatment NodeAI for lung cancer?

Research shows that artificial intelligence (AI) can help in early diagnosis and personalized treatment of lung cancer by analyzing patient data and predicting treatment responses. AI models have been successful in identifying cancer early and optimizing therapies, which can improve patient outcomes. ^{1 2 3 4 5}

How is the treatment NodeAI for lung cancer different from other treatments?

NodeAI is unique because it uses artificial intelligence (AI) to help detect and classify lung nodules, which are small growths in the lungs that can be cancerous. This AI approach aims to improve early detection and diagnosis, potentially leading to earlier and more effective treatment compared to traditional methods that rely heavily on human interpretation. ^{3 4 6 7 8}

Eligibility Criteria

This trial is for adults over 18 with suspected or confirmed non-small cell lung cancer (NSCLC), who have completed CT and PET scans and are referred for chest staging by EBUS-TBNA. It's not suitable for those who don't meet these specific criteria.

Inclusion Criteria

I have had both CT and PET scans done.

I am 18 or older with suspected or confirmed lung cancer, referred for a specific lung biopsy.

Timeline

Screening

Participants are screened for eligibility to participate in the trial

2-4 weeks

Treatment

Participants undergo EBUS-TBNA with real-time analysis by NodeAI and surgeon for lymph node malignancy prediction

1 day

1 visit (in-person)

Follow-up

Participants are monitored for the accuracy of NodeAI predictions compared to pathology results

3 weeks

Treatment Details

Interventions

• NodeAI

Trial Overview The trial is testing NodeAI, an AI-powered platform designed to predict lymph node malignancy in real-time during lung cancer staging procedures. The goal is to improve accuracy compared to the current standard of Systematic Sampling.

Participant Groups

Group I: NodeAI Experimental Treatment 1 Intervention

The ultrasound video and images of each LN will be analyzed by NodeAI, which will assign a CLNS for each LN based on the four ultrasonographic features of the CLNS, predict LN malignancy, and determine whether to biopsy it or not.

Group II: Surgeon Active Control 1 Intervention

The ultrasound video and images of each LN will first be analyzed by the surgeon, who will assign a CLNS for each LN based on the four ultrasonographic features of the CLNS, predict LN malignancy, and determine whether to biopsy it or not.

Who Is Running the Clinical Trial?

St. Joseph's Healthcare Hamilton

Lead Sponsor

Trials

203

Recruited

26,900+

Findings from Research

Artificial intelligence (AI) has the potential to enhance precision medicine in lung cancer by analyzing a variety of data, including molecular information and patient characteristics, to optimize treatment outcomes.

AI can assist in early cancer detection and provide personalized therapy recommendations, improving both survival rates and quality of life for lung cancer patients.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Integration of artificial intelligence in lung cancer: Rise of the machine. Ladbury, C., Amini, A., Govindarajan, A., et al.[2023]

Artificial intelligence (AI) has shown significant promise in improving lung cancer screening and diagnosis by accurately detecting and characterizing lung nodules through various imaging techniques, such as low-dose CT scans and PET-CT imaging.

AI algorithms can enhance treatment strategies by predicting patient responses to therapies and optimizing radiation treatment, potentially leading to better patient outcomes and reduced mortality rates in lung cancer.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Artificial Intelligence and Lung Cancer: Impact on Improving Patient Outcomes. Gandhi, Z., Gurram, P., Amgai, B., et al.[2023]

Deep learning, a form of artificial intelligence, is being explored as a powerful tool for diagnosing and researching lung cancer by mimicking human brain processes to analyze malignant cells.

This review highlights the current advancements and challenges in using deep learning for lung cancer pathology, suggesting it could significantly improve diagnosis and treatment strategies in the future.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Deep Learning Model as a New Trend in Computer-aided Diagnosis of Tumor Pathology for Lung Cancer. Cong, L., Feng, W., Yao, Z., et al.[2020]

A novel explainable AI-guided deep learning framework successfully identified seven significant biomarkers for non-small cell lung cancer (NSCLC) using methylation data, achieving a high classification accuracy of 91.53% with the Multilayer Perceptron model.

The discovered biomarkers, including C18orf18 and TNPO2, show potential for druggability and prognostic relevance, indicating their ability to predict patient survival and target specific biological pathways, which could lead to personalized therapies for NSCLC.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Enlightening the path to NSCLC biomarkers: Utilizing the power of XAI-guided deep learning. Dwivedi, K., Rajpal, A., Rajpal, S., et al.[2023]

This study explored the use of deep learning techniques on histological tissue slides to predict responses to chemoradiotherapy in 35 patients with stage III non-small cell lung cancer (NSCLC), identifying 34.7% as responders based on treatment outcomes.

GoogLeNet, one of the convolutional neural networks tested, demonstrated high specificity (90.1%) and good sensitivity (75%), suggesting that AI can enhance predictive capabilities beyond current clinical methods, paving the way for more personalized treatment strategies.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Deep pathomics: A new image-based tool for predicting response to treatment in stage III non-small cell lung cancer. Nibid, L., Greco, C., Cordelli, E., et al.[2023]

Early detection of pulmonary nodules is crucial for preventing bronchogenic carcinoma, but current methods often miss small cancerous lesions due to subjective variations.

Artificial intelligence (AI) and computer-aided diagnosis (CAD) systems are being developed to improve the accuracy of lung segmentation and the detection and classification of pulmonary nodules, potentially enhancing early diagnosis and treatment.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

How AI Can Help in the Diagnostic Dilemma of Pulmonary Nodules. Fahmy, D., Kandil, H., Khelifi, A., et al.[2022]

LungNet, a shallow convolutional neural network, was developed to predict outcomes for non-small cell lung cancer (NSCLC) patients and showed predictive accuracy for overall survival across four independent patient cohorts totaling 709 individuals.

The model not only predicts survival but also improves the classification of lung nodules as benign or malignant, achieving an area under the curve (AUC) of 0.85, indicating its potential as a noninvasive tool for lung cancer prognosis and CT image interpretation.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

A Shallow Convolutional Neural Network Predicts Prognosis of Lung Cancer Patients in Multi-Institutional CT-Image Data. Mukherjee, P., Zhou, M., Lee, E., et al.[2023]

A deep learning signature developed using the Swin Transformer achieved an impressive accuracy in predicting lymph node metastasis in lung adenocarcinoma patients, with an area under the curve (AUC) of 0.948-0.961 across three patient cohorts totaling 612 individuals.

This deep learning model outperformed traditional clinical-semantic and radiomics approaches, indicating its potential for improving noninvasive mediastinal lymph node staging and guiding personalized treatment strategies.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Development and validation of a deep learning signature for predicting lymph node metastasis in lung adenocarcinoma: comparison with radiomics signature and clinical-semantic model. Ma, X., Xia, L., Chen, J., et al.[2023]

References

1. United States pubmed.ncbi.nlm.nih.gov

Integration of artificial intelligence in lung cancer: Rise of the machine. [2023]

2. Switzerland pubmed.ncbi.nlm.nih.gov

Artificial Intelligence and Lung Cancer: Impact on Improving Patient Outcomes. [2023]

3. Australia pubmed.ncbi.nlm.nih.gov

Deep Learning Model as a New Trend in Computer-aided Diagnosis of Tumor Pathology for Lung Cancer. [2020]

4. Ireland pubmed.ncbi.nlm.nih.gov

Enlightening the path to NSCLC biomarkers: Utilizing the power of XAI-guided deep learning. [2023]

5. United States pubmed.ncbi.nlm.nih.gov

Deep pathomics: A new image-based tool for predicting response to treatment in stage III non-small cell lung cancer. [2023]

6. Switzerland pubmed.ncbi.nlm.nih.gov

How AI Can Help in the Diagnostic Dilemma of Pulmonary Nodules. [2022]

7. England pubmed.ncbi.nlm.nih.gov

A Shallow Convolutional Neural Network Predicts Prognosis of Lung Cancer Patients in Multi-Institutional CT-Image Data. [2023]

8. Germany pubmed.ncbi.nlm.nih.gov

Development and validation of a deep learning signature for predicting lymph node metastasis in lung adenocarcinoma: comparison with radiomics signature and clinical-semantic model. [2023]