Detection of malaria parasites from blood images using random forest algorithm in comparison with k-nearest neighbors to maximize the accuracy, precision and sensitivity
Yogesh Chandra, N. P. G. Bhavani · AIP conference proceedings · 2024
Detection of malaria parasites using supervised machine learning algorithms.The aim of this study is to compare the performance of the Innovative Random forest algorithm with the K-Nearest Neighbors algorithm (KNN) in detecting malaria parasites from blood images.Materials and Methods: A Dataset of 600 images of parasite and uninfected cell images at different resolutions were collected from the Kaggle website.The entire dataset is split into two sets: one for training (70 percent) and one for testing (30 percent).In Random Forest (Group 1), 450 malaria cell images were used in the training phase and 150 malaria cell images were used in the testing phase, which was compared to KNN (Group 2).Accuracy, precision, and sensitivity values are calculated using a confusion matrix to quantify the performance of the Random forest algorithm.Results: Random forest algorithm has achieved accuracy, precision and sensitivity of 91.80%, 97.62% and 90.12% respectively compared to KNN algorithm of 86.60%, 96.20% and 83.50%.As compared to KNN, the Random forest algorithm has a significantly higher detection rate (p=0.02).Conclusion: In this study, it was found that the Random forest algorithm gives better performance than the KNN algorithm in detection and classification of malaria parasites.