Artificial Intelligence, Elastography, Endobronchial Ultrasound, Lung Cancer, NSCLC
Conditions
Brief summary
Before any treatment decisions are made for patients with lung cancer, it is crucial to determine whether the cancer has spread to the lymph nodes in the chest. Traditionally, this is determined by taking biopsy samples from these lymph nodes, using the Endobronchial Ultrasound Transbronchial Needle Aspiration (EBUS-TBNA) procedure. Unfortunately, in 40% of the time, the results of EBUS-TBNA are not informative and wrong treatment decisions are made. There is, therefore, a recognized need for a better way to determine whether the cancer has spread to the lymph nodes in the chest. The investigators believe that elastography, a recently discovered imaging technology, can fulfill this need. In this study, the investigators are proposing to determine whether elastography can diagnose cancer in the lymph nodes. Elastography determines the tissue stiffness in the different parts of the lymph node and generates a colour map, where the stiffest part of the lymph node appears blue, and the softest part appears red. It has been proposed that if a lymph node is predominantly blue, then it contains cancer, and if it is predominantly red, then it is benign. To study this, the investigators have designed an experiment where the lymph nodes are imaged by EBUS-Elastography, and the images are subsequently analyzed by a computer algorithm using Artificial Intelligence. The algorithm will be trained to read the images first, and then predict whether these images show cancer in the lymph node. To evaluate the success of the algorithm, the investigators will compare its predictions to the pathology results from the lymph node biopsies or surgical specimens.
Interventions
DEVICEEBUS-Elastography
Patients undergoing LN staging for lung cancer with EBUS-TBNA will have digital images and biopsy of every LN obtained in accordance with standards of care. Prior to the lymph node biopsy by EBUS-TBNA, elastography will be performed. The relative strain of tissues in the scanned area of the LNs will be displayed as a colour map, with stiffer areas in blue and softer tissue in red. Elastography and B-mode images will be displayed side by side and images recorded and saved onto an external drive for analysis. Elastography images will be fed to the NeuralSeg algorithm which has a network architecture similar to the standard U-Net for image segmentation. The automatically identified regions of interest will be overlaid onto the EBUS Elastography images to extract the LN stiffness measurements. After overlaying, NeuralSeg will determine the proportion of the LN area within 9 previously defined stiffness thresholds.
Sponsors
St. Joseph's Healthcare Hamilton
Study design
Allocation
NA
Intervention model
SINGLE_GROUP
Primary purpose
DIAGNOSTIC
Masking
NONE
Intervention model description
This is a single-centre, prospective clinical trial, in which patients will be enrolled in a consecutive sample and patient involvement will conclude when the procedure ends. No follow-up will be required after the study.
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
* Patients that are diagnosed with suspected or confirmed NSCLC that have been referred to mediastinal staging through EBUS-TBNA at St. Joseph's Healthcare Hamilton will be eligible for this study.
Exclusion criteria
* No
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Stiffness Area Ratio | 8 months | Identifying whether the percent area of a lymph node above a defined blue colour threshold is independently associated with malignancy |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| NeuralSeg's prediction of lymph node malignancy | 2 months | Determine whether NeuralSeg can accurately predict malignancy in lymph nodes when compared to biopsy results of the lymph nodes that were examined |
| The agreement between NeuralSeg's predictions and pathology results, as measured by diagnostic accuracy, sensitivity, specificity, positive and negative predictive values | 2 months | The agreement between NeuralSeg's predictions and pathology results, as measured by diagnostic accuracy, sensitivity, specificity, positive and negative predictive values |
Countries
Canada
Outcome results
None listed