A Prospective Study on Artificial Intelligence Guided Lung Cancer Screening for High-risk Never Smokers in Hong Kong
试验速览
- 阶段
- 不适用
- 状态
- 招募中
- 入组人数
- 3,000
- 试验地点
- 1
- 主要终点
- Sensitivity, specificity, positive predictive value and negative predictive value of AI-assisted programme in lung nodule (≥5mm) detection and monitoring compared to radiologist assessment
研究概览
简要总结
Lung cancer screening is currently not recommended in non-smokers due to paucity of evidence. Emerging evidence suggests that first-degree family history is a strong risk factor for lung cancer in Asian non-smokers. In Asia, lack of resource is a major challenge in successful implementation of lung cancer screening. Artificial intelligence (AI) is a promising tool to overcome this resource. In this study, we aim to study the clinical utility and demonstrate the feasibility of using an AI assisted programme for lung cancer screening in Asian non-smokers with a positive family history. This is a single-arm non-randomized lung cancer screening study. 3000 non-smokers, age 50 to 75 year old, with a first-degree family history of lung cancer, will be enrolled. Participants will undergo low does computed tomography (LDCT) of thorax and blood taking at enrolment. LDCT films will be interpreted by AI softwares for presence of lung nodules. Participants with lung nodules will be further investigated and followed up according to the risk of malignancy. The primary endpoint is the prevalence of early-staged lung cancer detected by first-round LDCT thorax in this population.
研究设计
- 研究类型
- Interventional
- 分配方式
- Na
- 干预模型
- Single Group
- 主要目的
- Screening
- 盲法
- None
入排标准
- 年龄范围
- 50 Years 至 75 Years（Adult, Older Adult）
- 性别
- All
- 接受健康志愿者
- 否
入选标准
- •Patients are eligible to be included in the study only if all of the following criteria apply:
- •Age 50-75 years old
- •Non-smoker (defined as less than 100 cigarettes in lifetime)
- •Having a first-degree family history of lung cancer
- •Physically fit for curative treatment if early-staged lung cancer is found
- •Able to provide written informed consent
- •Consent to follow up visits and follow up CT scan if indicated
- •Consent to blood taking for translational research
排除标准
- •Patients who meet any of the following exclusion criteria at screening are not eligible to be enrolled in this study:
- •History of malignancy
- •Smoking history (defined as more than 100 cigarettes in lifetime)
- •Clinical symptoms suspicious for lung cancer e.g. haemoptysis, chest pain, weight loss
- •Medical comorbidities that preclude curative treatment (surgery) for lung cancer, such as severe heart disease, acute or chronic respiratory failure, home oxygen therapy, bleeding disorder
- •Pregnant ladies or ladies planning for conception
- •History of tuberculosis or interstitial lung disease
- •Pneumonia requiring antibiotic treatment within the last 12 weeks
- •CT thorax or chest performed within 2 years (including LDCT, PET-CT, MRI thorax or suspicious of lung cancer)
- •Unable or unwilling to provide written informed consent
研究组 & 干预措施
Artificial intelligence-based programme (Lung-SIGHT)
Artificial intelligence (AI) algorithms have been demonstrated to function well and complement radiologists as second or concurrent readers in pulmonary nodule detection. AI Lung nodule detection and quantification solution are now widely used in the hospitals in the United Kingdom and at least eight other European countries. The sensitivity of nodule detection by radiologists increased from 72% to 80% with the aid of the AI programme. A clinical trial in Taiwan showed that using AI programme alone achieved an overall sensitivity of 95.6% in nodule detection, and superior performance in detecting nodule sized 4-5 mm comparing to radiologists. Overall, application of AI in CT analysis and lung nodule detection may significantly reduce the cost and workload of radiologist.
干预措施: Lung-SIGHT (Device)
结局指标
主要结局
Sensitivity, specificity, positive predictive value and negative predictive value of AI-assisted programme in lung nodule (≥5mm) detection and monitoring compared to radiologist assessment
时间窗: 2 years
次要结局
- Prevalence of lung cancer detected by second-round LDCT (T1) in patients with negative first-round LDCT(2 years)
- Cost effectiveness of LDCT lung cancer screening using AI-assisted programme(2 years)
- Stage distribution of lung cancer detected by LDCT screening(2 years)
- Sensitivity, specificity, positive predictive value and negative predictive value of AI-assisted programme in lung cancer detection(2 years)
- Diagnostic utility of plasma-based biomarker for detection and risk assessment of early-staged lung cancer(2 years)
- Rate of invasive workup and associated complications(2 years)
研究者
Molly SC Li
Assistant Professor
Chinese University of Hong Kong