AI Determine Malignancy of GGO on Chest CT

Apply Artificial Intelligence to Determine the Benign or Malignancy for Pulmonary Ground Glass Nodules of Computed Tomography

Research Objectives To use AI computer-aided detection software to assist physicians in reading CT scans of lung nodules, providing auxiliary diagnostic tools for medical decision-making. The software can mark nodule locations and related information during routine physician reading. This study will obtain prospective consent to use patient CT images for software reading and compare with clinical physician diagnosis, in order to enhance software training and improve recognition of lung lesions for early diagnosis and treatment.

Study Design Collect CT images of untreated lung nodules 4-30mm in size that are scheduled for surgery. No limits on age, gender, disease type, with image resolution <2.5mm. AI and clinicians will judge nodule characteristics separately. Surgical resection followed by comparison with pathology reports will evaluate diagnostic accuracy.

Study Procedures A double-blinded method will be used. AI and physicians will record nodules as likely benign or malignant separately. After surgical resection, the lesions will undergo pathological staging and the diagnostic accuracy of both groups will be compared.

Expected Results Compare the diagnostic accuracy of AI and clinicians to improve AI training quality, achieve early diagnosis and treatment goals, and provide patients with better medical care quality.

Monitoring Method AI and clinicians will read separately, adhering to shared decision making without affecting patient access to diagnosis and treatment.

Keywords: lung nodules, early lung cancer, artificial intelligence, chest CT, minimally invasive surgery, lung image analysis software

Study Overview

• Status: Enrolling by invitation
• Conditions: Lung Nodules, Early Lung Cancer, Artificial Intelligence, Chest CT, Minimally Invasive Surgery, Lung Image Analysis Software
• Intervention / Treatment: Diagnostic test: AI computer-aided detection software

Detailed Description

1. Research Objectives

This study will install artificial intelligence (AI) computer-aided detection software (CADe) on the original user interface to provide auxiliary diagnostic tools for clinical medical decision-making.

During routine CT examinations, this reading software can assist in identifying nodules in chest CT scan images. When physicians routinely read lung CT scan images, the marking results will be displayed. Applicable nodule sizes are 4mm to 30mm. When suspicious nodules are detected, the physician will mark the ROI (Region of Interest) location and present nodule-related information for physician reference during diagnosis.

This study will also obtain informed consent forms from subjects on a prospective basis to acquire patients' serial chest CT images. The AI intelligent lung image reading software and clinically caring physicians will provide benign and malignant after judgment. This is to enhance the practical training of intelligent software. It is also expected to improve the recognition of benign and malignant lung lesions in future images, not just lung nodules. This can reduce overdiagnosis and treatment rates, or it is expected to improve the accuracy of early diagnosis and treatment. "AI cannot be used as the sole basis for diagnosis. It only provides auxiliary diagnostic tools for clinical medical decision-making and must not simplify or replace diagnosis/treatment procedures."

2. Study Design (Summary)

The enrollment period is limited to 2 years. There are no gender restrictions. Age >20Y/O . Patients with Low-dose chest CT scans (<2.5mm slice thickness, imaging from any hospital) detecting lung nodules (no limit on nodule type/region or nature of nodules 4-30mm) that have not yet undergone surgery, and are scheduled to undergo surgical resection at the Department of Thoracic Surgery, Chung Shan Medical University Hospital will be included.

Exclusion criteria are low-dose chest CT scans (only >2.5-5mm slice thickness, imaging from any hospital), lung tumors (>30mm, no limit on nature), lung nodules that have already undergone surgical resection (no limit on known or unknown pathology reports), and patients with known other cancers (other known cancers besides lung cancer that meets inclusion criteria must be excluded) will not be included.

Trial group: AI Control group: Attending physician from the Department of Thoracic Surgery, Chung Shan Medical University Hospital

● Study Procedures

Under the above conditions of low-dose chest CT imaging and inclusion criteria, the two groups will identify and record lung nodules and mark them as either likely benign or likely malignant prior to impending surgery (before pathology reports are known). All of these lung lesions will eventually undergo surgical resection (no limit on surgical methods) and complete pathological results. Analysis will then be performed to evaluate the accuracy of predicting benignancy or malignancy in the trial and control groups.

Notes:

1. Identification records will be collected by co-PI (Tsai) in a double-blinded manner.
2. Trial group - AI by V5 lung image reading software
3. Control group - Attending physician from Dept of Thoracic Surgery, Chung Shan Medical University Hospital. Cannot rely solely on V5 lung image reading software for diagnosis. Cannot simplify or replace diagnosis/treatment procedures. Cannot affect patient's rights to diagnosis and treatment.

4. This study cannot affect or intervene with patients receiving diagnosis and treatment. Must respect shared decision making between doctors and patients, and respect patient autonomy.

   • Inclusion Criteria

1. Age>20 y/o
2. Gender (no limit)
3. Disease type (no limit on known lung cancer history, acute or chronic non-cancerous conditions)
4. Low-dose chest CT (<2.5mm slice thickness, imaging from any hospital)
5. Lung nodules (no limit on nodule type/region, or nature of nodules <4-30mm)

6. Lesions not yet operated on, expected to undergo surgical resection at Dept of Thoracic Surgery, Chung Shan Medical University Hospital

   • Exclusion Criteria

1. Low-dose chest CT (only >2.5-5mm slice thickness, imaging from any hospital)
2. Lung tumors (>30mm, no limit on nature)
3. Lung nodules that have undergone surgical resection (no limit on known or unknown pathology)

4. Patients with known other cancers (other known cancers besides lung cancer meeting inclusion criteria must be excluded)

   ● Statistical Analysis Method

   Clinical trials will be conducted in a double-blinded manner, under the premise of not affecting disease diagnosis and treatment procedures.

   ROC curves

   • Evaluation Indices

   Comparison with post-operative pathology reports will serve as imaging evaluation indices for accuracy and ratios of benignancy vs. malignancy.

   • Withdrawal Criteria

   This study does not conflict with clinical medical shared decision-making nor affect any original established treatments. Results will only serve as future auxiliary clinical tools to assist in identification and utilization, with the goal of providing advantages in clinical decision-making for lung cancer diagnosis and treatment.

   • Rescue Treatments

   None, does not affect any impending medical shared decision-making and treatments. No risk impacts.

   • Target enrollment number

     100 patients over 2 years study period.

   • Expected Study Duration

   3/1/2024 - 2/28/2026

• Study Type: Observational
• Enrollment (Estimated): 100

Contacts and Locations

Study Locations

• Taiwan
  • South
    • Taichung, South, Taiwan, 402
      • Chung Shan Medical University Hospital

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: N/A

• Sampling Method: Non-Probability Sample

Study Population

Low-dose chest CT (only >2.5-5mm slice thickness, imaging from any hospital) Lung tumors (>30mm, no limit on nature) Lung nodules that have undergone surgical resection (no limit on known or unknown pathology) Patients with known other cancers (other known cancers besides lung cancer meeting inclusion criteria must be excluded)

Description

Inclusion Criteria:

- Age >20 Gender (no limit) Disease type (no limit on known lung cancer history, acute or chronic non-cancerous conditions) Low-dose chest CT (<2.5mm slice thickness, imaging from any hospital) Lung nodules (no limit on nodule type/region, or nature of nodules <4-30mm) Lesions not yet operated on, expected to undergo surgical resection at Dept of Thoracic Surgery, Chung Shan Medical University Hospital

Exclusion Criteria:

-

Study Plan

How is the study designed?

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Time Frame
The accuricy of the diagnostic power.	2024/3/1,2026/5/31

Collaborators and Investigators

• Sponsor: Chung Shan Medical University

Study record dates

Study Major Dates

• Study Start (Estimated): March 1, 2024
• Primary Completion (Estimated): February 28, 2026
• Study Completion (Estimated): February 28, 2026

Study Registration Dates

• First Submitted: February 21, 2024
• First Submitted That Met QC Criteria: February 21, 2024
• First Posted (Estimated): February 28, 2024

Study Record Updates

• Last Update Posted (Estimated): February 28, 2024
• Last Update Submitted That Met QC Criteria: February 21, 2024
• Last Verified: February 1, 2024

More Information

Terms related to this study

• Additional Relevant MeSH Terms: Neoplasms
• Other Study ID Numbers: CS2-23178

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: UNDECIDED

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No