Core Concepts
A novel Deep Boosted and Ensemble Learning framework is proposed for accurate malaria parasite screening.
Abstract
The content discusses the development of a new Deep Boosted and Ensemble Learning (DBEL) framework for screening malaria parasite images. The framework includes a new dilated-convolutional block-based Split Transform Merge (STM) and feature-map Squeezing-Boosting (SB) ideas to improve discrimination ability and generalization of ensemble learning. The proposed DBEL outperforms existing techniques on the NIH malaria dataset, achieving high accuracy, sensitivity, F-score, and AUC values. Data enhancement using discrete wavelet transformations (DWT) and data augmentation techniques are employed to improve computational complexity and model robustness. The content also covers the utilization of customized CNNs, performance evaluation metrics, feature space visualization through PCA analysis, graphical analysis with ROC and PR curves, as well as future implications in healthcare applications.
Abstract:
Introduction to Malaria Parasite Detection using DBEL framework.
Overview of manual screening challenges in clinical practice.
Description of the proposed Boosted-BR-STM CNN architecture.
Explanation of ensemble learning strategies for improved detection.
Comparison with existing CNN models on the NIH malaria dataset.
Performance evaluation metrics including Accuracy, Sensitivity, F-score.
Feature space visualization through PCA analysis.
Graphical analysis with ROC and PR curves.
Future implications in healthcare applications.
Stats
The proposed DBEL framework achieved Accuracy (98.50%), Sensitivity (0.9920), F-score (0.9850), and AUC (0.9960).
Quotes
"The proposed DBEL framework implicates the stacking of prominent and diverse boosted channels."
"The developed Boosted-BR-STM improved the generalizability of the proposed DBEL scheme."
"The hybrid learning approach evaluates the performance of deep boosted feature maps."