Optimizing CNN Hyperparameters for Blastocyst Quality Assessment in Small Datasets

Morphological assessment of blastocyst quality is one the most significant challenges in IVF process because current still based on evaluation an embryologist, so it manual, subjective, and lacks precision. Artificial Intelligence (AI) plays a role overcoming limitations manual system, expected to increase implantation process. The study aims optimize convolutional neural networks (CNN) model by grid search method evaluate different machine learning models classifying small dataset. reliability proposed will be compared with other as logistic regression (LR), support vector (SVM), k-nearest neighbors (KNN), boosting algorithm, adding canny operator segmentation principal component analysis (PCA) feature extraction.We results performance measures like Precision, Recall, F1-measure, Accuracy, Area under receiver operating characteristic curve (AUC-ROC). final showed that our CNN achieves validation accuracy 84.00%, test 83.33%, AUC 0.844. McNemar’s statistical outperforms classifiers.