Detection and stage classification of Plasmodium falciparum from images of Giemsa stained thin blood films using random forest classifiers

Abstract

Background The conventional method for the diagnosis of malaria parasites is the microscopic examination of stained blood films, which is time consuming and requires expertise. We introduce computer-based image segmentation and life stage classification with a random forest classifier. Segmentation and stage classification are performed on a large dataset of malaria parasites with ground truth labels provided by experts. Methods We made use of Giemsa stained images obtained from the blood of 16 patients infected with Plasmodium falciparum. Experts labeled the parasite types from each of the images. We applied a two-step approach: image segmentation followed by life stage classification. In segmentation, we classified each pixel as a parasite or non-parasite pixel using a random forest classifier. Performance was evaluated with classification accuracy, Dice coefficient and free-response receiver operating characteristic (FROC) analysis. In life stage classification, we classified each of the segmented objects into one of 8 classes: 6 parasite life stages, early ring, late ring or early trophozoite, mid trophozoite, early schizont, late schizont or segmented, and two other classes, white blood cell or debris. Results Our segmentation method gives an average cross-validated Dice coefficient of 0.82 which is a 13% improvement compared to the Otsu method. The Otsu method achieved a True Positive Fraction (TPF) of 0.925 at the expense of a False Positive Rate (FPR) of 2.45. At the same TPF of 0.925, our method achieved an FPR of 0.92, an improvement by more than a factor two. We find that inclusion of average intensity of the whole image as feature for the random forest considerably improves segmentation performance. We obtain an overall accuracy of 58.8% when classifying all life stages. Stages are mostly confused with their neighboring stages. When we reduce the life stages to ring, trophozoite and schizont only, we obtain an accuracy of 82.7%. Conclusion Pixel classification gives better segmentation performance than the conventional Otsu method. Effects of staining and background variations can be reduced with the inclusion of average intensity features. The proposed method and data set can be used in the development of automatic tools for the detection and stage classification of malaria parasites. The data set is publicly available as a benchmark for future studies.

Automated detection and staging of malaria parasites from cytological smears using convolutional neural networks

“…Finally, we have introduced a staging model that recapitulates IDC progression of the parasite and allows for more refined interpretation when reading slides. Recent studies have aimed to enhance IDC life stage classification by adding early, mid, and late substages (25) . Using a regression model, we are able to differentiate the IDC stage of the parasite more precisely and prevent penalizing predictions that have been classified as a different substage of the parasite but are developmentally not far removed.…”

“…Only a few studies have also combined parasite detection with the classification of the different stages of the intraerythrocytic development cycle (IDC). Furthermore, treating parasite development as a classification problem disregards information on progression within and between the individual stages; progression through the IDC is a continuous process, and experts disagree on the boundaries between the different life stages (10,21,25,26) . Automation has the potential to save time and add to the number of RBCs sampled for both diagnosis and lab usage; however, its usage in diagnosis would require confidence in the analysis process; if results are accessible for review post-analysis by a microscopist, such a system is more likely to be implemented as a robust decision support tool.…”

Reducing data dimension boosts neural network-based stage-specific malaria detection

“…(i) All types of image sets were recorded by our group in our study, except for Giemsa-stained light microscopy data, which were taken from Abbas et al 39 .…”

“…For segmentation into single-RBC cutouts, that were eventually used for the NN-based four-category classification, we used 173 AFM images, 1232 fluorescence microscopy images and 792 Giemsa-stained light microscopy images [747 from Ref. 39 and 45 recorded by us]. Label-free light microscopy images were taken in a limited number to help an unambiguous stage-specific labeling of RBCs in the corresponding fluorescence images and were not used in the NN analysis.…”

Automated detection and staging of malaria parasites from cytological smears using convolutional neural networks

“…Only a few studies have also combined parasite detection with the classification of the different stages of the intraerythrocytic development cycle (IDC). Furthermore, treating parasite development as a classification problem disregards information on progression within and between the individual stages; progression through the IDC is a continuous process and experts disagree on the boundaries between the different life stages (10, 21, 25, 26). Automation has the potential to save time and add to number of RBCs sampled for both diagnosis and lab usage, however, its usage in diagnosis would require confidence in the analysis process; if results are accessible for review post-analysis by a microscopist, such a system is more likely to be implemented as a robust decision support tool.…”