Accurate diagnosis and treatment planning for ischemic stroke present significant challenges, largely due to reliance on subjective visual assessments and complex image processing. Traditional thrombectomy eligibility assessment often utilizes Time to Maximum (Tmax) maps derived from Perfusion Weighted Imaging (PWI), requiring deconvolution techniques that are both time-consuming and reliant on costly, sophisticated software for precise hypoperfused tissue quantification. This study proposes an innovative AI-based approach that stream-lines thrombectomy candidate selection by directly analyzing Time-to-Peak (TTP) maps, bypassing the need for Tmax computation. We employed a U-Net segmentation model to efficiently and accurately identify hypoperfused regions, reducing the procedural complexity, time, and costs associated with traditional methods. Our retrospective study utilized a dataset of 14 patients with 339 TTP images to train and validate the AI model, achieving high accuracy with a Pearson correlation coefficient of 0.96 and a Dice similarity coefficient of 0.92 compared to conventional Tmax-based analyses. Notably, this method reduced analysis time from 4.5 min to 45 s per patient, maintaining both accuracy and reliability. The direct application of AI to TTP maps marks a transformative step in ischemic stroke assessment, simplifying the evaluation process and making advanced care potentially more accessible. Further research on larger datasets is required to confirm the broader clinical applicability of this technique.