YOLO-TumorNet: An innovative model for enhancing brain tumor detection performance

Brain tumors are high-risk conditions where early detection and precise localization are crucial for improving patient prognosis. However, existing automated detection methods still exhibit limitations in robustness within complex backgrounds, boundary recognition, and the detection of small tumors, making it challenging to meet the high precision requirements of clinical applications. To address these issues, this paper proposes an improved YOLOv10-based model, YOLO-TumorNet. Specifically, YOLO-TumorNet integrates the InceptionNeXt architecture, Multi-Scale Spatial Pyramid Attention (MSPA), and Bidirectional Feature Pyramid Network (BiFPN) modules to enhance multi-scale feature extraction and channel attention mechanisms, thereby improving the model’s accuracy and robustness in brain tumor detection. Additionally, extensive experiments conducted on the Br35H and Roboflow datasets demonstrate the superior performance of YOLO-TumorNet in terms of boundary clarity, detail capture, and small tumor detection.