Improving YOLOv8 Performance Using Hyperparameter Optimization with Gray Wolf Optimizer to Detect Acute Lymphoblastic Leukemia

The identification of Acute Lymphoblastic Leukemia (ALL) using deep learning is crucial for early and proper diagnosis of the disease. Nevertheless, the efficacy of deep model learning is frequently constrained by suboptimal hyperparameter configurations. This research aims to improve the performance of the YOLOv8 model by optimizing hyperparameters using the Gray Wolf Optimizer (GWO). A thorough experimental procedure is conducted to optimize the hyperparameters. Subsequently, the hyperparameter-optimized YOLOv8 model is contrasted with the original versions of YOLOv5, YOLOv7, and YOLOv8. The experimental evidence suggests that the GWO-optimized YOLOv8 model surpass the performance of the original model as well as other YOLO models. The comparison involving the GWO-optimized YOLOv8 and the initial YOLOv8 exposes a growth in precision by almost 3.9% and an uptick in mAP50 by 6.7<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>. Furthermore, the experimental outcomes also showcase superior performance over YOLOv5 and YOLOv7 models, with precision improvements reaching up to 21.2 % and 26.3 % in the lymphocytic subtype.