Figure 9
From: Reinforcement learning for patient-specific optimal stenting of intracranial aneurysms
Stent optimization along deep reinforcement learning for aneurysm B. The fine line represents the evolution per episode of the instant reward, while the thick line is the moving average reward computed over the 100 latest values. The right vertical axis presents the relative variations of MWSS with respect to the setpoint of half the pre-operative value. Representative stents generated over the course of optimization are superimposed, with the stent generated at episodes 15, 25 and 35 shown in (a–c) and the optimal stent predominantly generated after episode 45 shown in (d). P and D annotations indicate the proximal and distal end sections of the stent, respectively.
