Methodological Framework and Study Protocolof Intracranial Hemorrhage Detection Using the "Athou-scan®" Artificial Intelligence Model at University Clinics of Kinshasa, Democratic Republic of Kinshasa

Intracranial hemorrhage (ICH) requires prompt diagnosis, with computed tomography (CT) serving as the imaging modality of choice to ensure timely and appropriate management. Delays in image interpretation are associated with poorer clinical outcomes. Artificial intelligence (AI)-based solutions have demonstrated high diagnostic performance in the literature; however, their implementation in resource-limited settings remains challenging due to technical constraints (dataset development and annotation, model training, optimization, and validation), economic barriers (acquisition and maintenance costs), and organizational issues (training, user acceptance, and accountability). As a low-resource country, the Democratic Republic of the Congo (DRC) faces these challenges, which limit the effective deployment of AI-assisted diagnostic systems. The primary objective of this study is to evaluate the short-term diagnostic and functional performance of the Athou-scan® model for the detection and classification of intracranial hemorrhage in the DRC.This mixed-methods study comprises three components: (1) the development of Athou-scan®, a convolutional neural network (CNN)-based AI model for the detection and classification of intracranial hemorrhage on CT images; (2) a prospective clinical validation comparing its diagnostic performance with that of radiologists; and (3) a qualitative assessment of user acceptability, organizational barriers, and ethical considerations related to its implementation