RewardQNet: Novel Reward-based Deep Learning Agent Self-structuring Framework for Untreated Intracranial Aneurysm (UIA) Segmentation

Around 3% of the world's population suffers from an unruptured intracranial aneurysm (UIA). They are even more prevalent in certain risk groups, with a prevalence of approximately 10% in individuals with a positive family history of aneurysmal subarachnoid haemorrhage (aSAH). A subarachnoid haemorrhage (aSAH) is a type of severe stroke caused by the rupture of an intracranial
aneurysm. While semantic segmentation algorithms enable image analysis and quantification in a wide variety of applications, developing specialised solutions is challenging and highly dependent on the dataset's properties and hardware configuration. We introduced RewardQNet, a novel reward-based deep learning agent capable of automatically structuring the appropriate hyperparameters for the segmentation task, including preprocessing, network architecture, training, and postprocessing, based on the actions that yield the highest reward. A set of rooted parameters, Q-actions, and an empirical criterion will be used to classify the process's critical design choices. RewardQNet will structure the deep learning model autonomously in order to perform the highestreward segmentation of UIA. This framework can assist a healthcare professional in diagnosing UIAs. It is consistent with Malaysia's National Policy on Industry 4.0 (Industry 4wrd) under T1 of "Access to Smart Technologies and Standards" and with the United
Nations' Sustainable Development Goals under Goal 3 "Good health and well-being" to ensure healthy lives and promote well-being for all, which is also consistent with Malaysia's 12th Malaysia Plan (RMK -12).