Malaria Diagnosis Using a Lightweight Deep Convolutional Neural Network

Abstract

The applications of AI in the healthcare sector are increasing day by day. The application of convolutional neural network (CNN) and mask-region-based CNN (Mask-RCCN) to the medical domain has really revolutionized medical image analysis. CNNs have been prominently used for identification, classification, and feature extraction tasks, and they have delivered a great performance at these tasks. In our study, we propose a lightweight CNN, which requires less time to train, for identifying malaria parasitic red blood cells and distinguishing them from healthy red blood cells. To compare the accuracy of our model, we used transfer learning on two models, namely, the VGG-19 and the Inception v3. We train our model in three different configurations depending on the proportion of data being fed to the model for training. For all three configurations, our proposed model is able to achieve an accuracy of around 96%, which is higher than both the other models that we trained for the same three configurations. It shows that our model is able to perform better along with low computational requirements. Therefore, it can be used more efficiently and can be easily deployed for detecting malaria cells.

iMAGING: a novel automated system for malaria diagnosis by using artificial intelligence tools and a universal low-cost robotized microscope

“…In contrast, the addition of different images in terms of visual differences would affect the final descriptive parameters of the algorithm, although it would confer robustness to detect diverse preparations (Maron et al, 2021). Thin blood smear algorithms for parasite detection usually have higher values of precision, recall and, consequently, F-score, when compared with thick blood smears (Loddo et al, 2022;Magotra and Rohil, 2022). In addition, the customization of CNNs to improve detection results is generating optimal algorithms, such as the REONet method (modified ResNet-18) to classify malaria parasites on thin blood smears with 96.68% specificity, 94.79% sensitivity, and a 95.69% F-score (Zhu et al, 2022).…”

An optimized features selection approach based on Manta Ray Foraging Optimization (MRFO) method for parasite malaria classification

“…The malaria cells are classified using the proposed convolutional neural network-based transfer learning model such as efficientnet-b0 and shuffle-net. The pre-trained efficientnet-b0 consists of 290 layers such as convolutional (65), batch-normalization (49), sigmoid (65), element-wise multiplication (65), convolution group (15), average global pooling (16), addition (9), fully connected (FC), addition (15), classification, softmax, and global average pool (16). The shufflenet consists of 172 layers such as input (1), convolution ( 49…”

“…),(49) batch-normalization, ReLU(33), max-pooling (01), average pooling (02), 16 shuffling channels, 1 fully connected (FC), 1 softmax, 1 classification, 15 addition, 01 average global pool, and 2 depth concatenation. This research extracted features from the MatMul FC layer of efficient-netb0 and node-202 FC layer of shuffle-net.…”

Malaria cell identification using improved machine learning and modified deep learning architecture

“…By taking the original image and converting it into several images using transformation techniques like rotation, shear, and translation, augmentation expands the dataset and helps the model perform more accurately. Convolutional neural networks (CNNs) [17] are popular and computationally efficient for classification tasks [6], [13].…”