Chronic kidney disease (CKD) is an irreversible condition that progressively impairs kidney function, leading to the need for hemodialysis or kidney transplantation. For hemodialysis (HD) patients, their dialysis access, such as the arteriovenous fistula, serves as a vital lifeline for efficient blood filtration and waste removal. Because HD patients undergo hemodialysis three times a week, the repeated needle punctures may gradually lead to vascular tissue hyperplasia, vascular intimal thickening, and fibrosis. Underlying chronic conditions in HD patients such as diabetes, cardiovascular diseases, or metabolic syndrome can further contribute to inflammation, infection, or access stenosis or occlusion. Hemodialysis fistula stenosis screening is essential to preserve vascular access, as early detection enables timely intervention, helping to ensure adequate dialysis, and minimize the risk of thrombosis or access failure. Hence, for early HD fistula stenosis detection, in this study, we propose to train a YOLOv11-based classifier with image enhancement, feature extraction, and pattern recognition to enable automatic stenosis detection. With B-mode ultrasound imaging data stream, the proposed classifier performs the automated image segmentation for distinguishing vascular wall contours, vascular lumen, and thrombotic regions and then presents them in a colored visualization pattern. By tenfold cross-validation, the classifier model based on the proposed “Two Dimension (2D) Fractional-Order Convolution Operation (FOCO) + YOLOv11 + Stochastic Gradient Descent Optimizer” achieved the following average performance metrics: Precision of 100.00 ± 0.00%, Recall of 90.00 ± 0.01%, F1 score of 0.9474 ± 0.0032, and Accuracy of 94.74 ± 0.02% for the identification of normal and stenotic regions. On the basis of the segmentation results, the key vascular quantification indexes, such as the degree of stenosis and luminal change rate, are computed to evaluate the access stenosis levels. Furthermore, by utilizing spatial information, we can reconstruct the sequential ultrasound images into a 3D visualization fistula model through Mixed Reality display devices, offering an intuitive and interactive assessment of fistula health and potential blockages.