The research project is a component of another research project that applies the protocol of the World Health Organization for screening of cervical cancer, with testing of high-risk Human Papilloma Virus (hrHPV) as first screening. In the screen, triage and treat approach women who tested positive for hrHPV are undergoing Visual Inspection of the cervix with Acetic Acid (VIA). This procedure is applied in Uganda, India and Bangladesh. However the quality of VIA by lower-trained staff is variable because Low and Middle Income Countries face limited numbers of qualified health care professionals. Artificial intelligence (AI) might be a solution to improve consistency of VIA assessment. This research validates an AI decision support system (AI-DSS) under field conditions.
This research project aims to validate an image processing algorithm in order to study if the AI can function as decision support system (DSS) for healthcare workers, who perform VIA. The AI-DSS has been designed by the Manipal Academy of Higher Education in India and has shown a sensitivity of 99.05% and specificity of 97.16% for detection of aceto-white lesions. The first step of the validation process is a dry run using a database of VIA pictures with pathology confirmation. The accuracy, sensitivity and specificity of healthcare workers, an expert team and the AI will be analysed, based on the ground truth of database pictures. Thereafter, the AI-DSS will be validated under field conditions. Women who are hrHPV-positive in Uganda, India and Bangladesh will undergo a VIA. The AI is incorporated in a device, that can take pictures of the cervix and provide and instant report (positive = needs further investigation or negative = no lesions found). The user experience of the device will be evaluated. The AI-DSS will be validated by comparing the VIA assessment by health workers, the AI with assessment by the expert panel. Accuracy, sensitivity and specificity of the AI algorithm will be assessed.
Study Type
INTERVENTIONAL
Allocation
NON_RANDOMIZED
Purpose
SCREENING
Masking
NONE
Enrollment
600
VIA procedure, taking pre-colouring and post-colouring pictures. First VIA assessment by health worker, followed by AI-DSS assessment. Independent panel of gynaecologists also assessed pictures. The panel assessment serves as reference value to calculate accuracy, sensitivity and specificity of health worker and AI-DSS.
VIA procedure, taking pre-colouring and post-colouring pictures. First VIA assessment by health worker, followed by AI-DSS assessment. Independent panel of gynaecologists also assessed pictures. The panel assessment serves as reference value to calculate accuracy, sensitivity and specificity of health worker and AI-DSS.
VIA procedure, taking pre-colouring and post-colouring pictures. First VIA assessment by health worker, followed by AI-DSS assessment. Independent panel of gynaecologists also assessed pictures. The panel assessment serves as reference value to calculate accuracy, sensitivity and specificity of health worker and AI-DSS.
Mpasana
Kakumiro, Uganda
Accuracy, sensitivity, specificity of AI-DSS
Measurement of true and false positive and true and false negative results of the AI DSS in relation to assessment by expert panel
Time frame: through study completion, estimated 2 years
Percentage of quality pictures
Percentage of pictures taken during the VIA procedures, which meet according to the expert panel the quality standards suitable for AI-DSS assessment, based on visibility of cervix and Squamo-Columnar Junction (SCJ), light, reflection of speculum, mucus-free surface
Time frame: through study completion, estimated 2 years
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.