School of Computer Science and Engineering, Central South University, Changsha, 410083, People's Republic of China.
Department of Network Center, Pingdingshan University, Pingdingshan, 467000, People's Republic of China.
Interdiscip Sci. 2020 Jun;12(2):217-225. doi: 10.1007/s12539-020-00367-7. Epub 2020 May 11.
Malaria is one of the epidemics that can cause human death. Accurate and rapid diagnosis of malaria is important for treatment. Due to the limited number of data and human factors, the prediction performance and reliability of traditional classification methods are often affected. In this study, we propose an efficient and novel classification network named Attentive Dense Circular Net (ADCN) which based on Convolutional Neural Networks (CNN). The ADCN is inspired by the ideas of residual and dense networks and combines with the attention mechanism. We train and evaluate our proposed model on a publicly available red blood cells (RBC) dataset and compare ADCN with several well-established CNN models. Compared to other best performing CNN model in our experiments, ADCN shows superiority in all performance criteria such as accuracy (97.47% vs 94.61%), sensitivity (97.86% vs 95.20%) and specificity (97.07% vs 92.87%). Finally, we discuss the obtained results and analyze the difficulties of RBCs classification.
疟疾是可导致人类死亡的传染病之一。疟疾的准确快速诊断对治疗很重要。由于数据和人为因素的限制，传统分类方法的预测性能和可靠性往往受到影响。在这项研究中，我们提出了一种名为 Attentive Dense Circular Net（ADCN）的高效新颖分类网络，它基于卷积神经网络（CNN）。ADCN 的灵感来自残差网络和密集网络的思想，并结合了注意力机制。我们在公开可用的红细胞（RBC）数据集上训练和评估我们提出的模型，并将 ADCN 与几个成熟的 CNN 模型进行比较。与实验中表现最好的其他 CNN 模型相比，ADCN 在所有性能标准（如准确性（97.47%对 94.61%）、敏感性（97.86%对 95.20%）和特异性（97.07%对 92.87%））方面均表现出优势。最后，我们讨论了获得的结果，并分析了 RBC 分类的困难。
