Validation of an AI algorithm for preoperative planning for lung cancer patients
Principal Investigator
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-1389
Initial CDAS Request Approval
Feb 19, 2025
Title
Validation of an AI algorithm for preoperative planning for lung cancer patients
Summary
Objective:
This study aims to validate an already developed AI-based algorithm for pre-operative planning in lung cancer patients, focusing on improving tumor localization, anatomical structure visualization, and surgical planning on US patient data.
Background:
Lung cancer remains one of the leading causes of cancer-related mortality, with surgery being a primary treatment modality. However, the success of surgical interventions heavily depends on accurate pre-operative planning, which traditionally relies on manual interpretation of imaging data by radiologists. This process is not only time-consuming but also prone to human error. AI technologies offer the potential to automate and enhance these tasks, thereby improving surgical precision, reducing errors, and enabling more personalized treatment strategies.
Current pre-operative planning methods primarily involve manual segmentation and tumor localization using CT scans. These methods can be subjective and dependent on the expertise of the clinician. The proposed AI algorithm aims to streamline this process by accurately identifying tumor boundaries, visualizing proximity to critical structures, thus supporting surgeons in making sound decisions.
Methodology:
The study will involve CT scans from NLST patients who went through surgical resection. Pre-operative imaging will be processed by the AI algorithm. The algorithm, developed using deep learning techniques, has been trained to analyze these images, segment tumor regions, and evaluate anatomical structures relevant to surgery.
The validation process will occur in the following phases:
1. Data Collection: Imaging data from NLST patients who went through lung cancer surgeries.
2. Algorithm Performance Assessment: The algorithm’s output (tumor segmentation and anatomical structure visualization) will be compared to expert radiologists’ evaluations, focusing on accuracy.
This study aims to validate an already developed AI-based algorithm for pre-operative planning in lung cancer patients, focusing on improving tumor localization, anatomical structure visualization, and surgical planning on US patient data.
Background:
Lung cancer remains one of the leading causes of cancer-related mortality, with surgery being a primary treatment modality. However, the success of surgical interventions heavily depends on accurate pre-operative planning, which traditionally relies on manual interpretation of imaging data by radiologists. This process is not only time-consuming but also prone to human error. AI technologies offer the potential to automate and enhance these tasks, thereby improving surgical precision, reducing errors, and enabling more personalized treatment strategies.
Current pre-operative planning methods primarily involve manual segmentation and tumor localization using CT scans. These methods can be subjective and dependent on the expertise of the clinician. The proposed AI algorithm aims to streamline this process by accurately identifying tumor boundaries, visualizing proximity to critical structures, thus supporting surgeons in making sound decisions.
Methodology:
The study will involve CT scans from NLST patients who went through surgical resection. Pre-operative imaging will be processed by the AI algorithm. The algorithm, developed using deep learning techniques, has been trained to analyze these images, segment tumor regions, and evaluate anatomical structures relevant to surgery.
The validation process will occur in the following phases:
1. Data Collection: Imaging data from NLST patients who went through lung cancer surgeries.
2. Algorithm Performance Assessment: The algorithm’s output (tumor segmentation and anatomical structure visualization) will be compared to expert radiologists’ evaluations, focusing on accuracy.
Aims
1. Validate the AI Algorithm’s Accuracy:
Compare the AI algorithm’s tumor localization and segmentation results with expert radiologists’ evaluations.
Assess the robustness of AI algorithm on a patient population (US population) which differs from the training dataset population.
Analyze imaging protocol's influence on AI performance
Collaborators
None