AI-based feature parameters extraction from color images

The article presents a novel artificial Intelligent based feature parameters extraction that utilizes Mixed descriptors for analyzing white blood cell images. The proposed approach involves integrating these matrices into an artificial intelligent and Mixed descriptor architecture called AI-MD. The performance of the AI-MD model is evaluated using four cross-validation methods and various evaluation metrics including loss and accuracy curves, F1-score, recall, and receiver operating characteristic curve. The study concludes with a comparison of the proposed method with other recognition approaches. In order to assess the effectiveness of our architecture, we utilized the four cross-validation method. Recognition accuracy was evaluated using the area under the curve (AUC) metric for specific classes for all based on the original image matrices. The AUC values attained for these classes were as follows: 99.49%, 99.75%, 98.60%, and 99.72% respectively. Our method yielded highly favorable outcomes, with an impressive AUC value of 93.93%. This result outperformed existing approaches, highlighting the significant potential of our method. Consequently, it underscores the necessity for additional research in the realm of screening methods within medical applications, we also validate this experimentation using dataset of UCI has suggested to classify the fetus into three classes: normal, suspicious, and pathological.