Automated Detection of P. falciparum Using Machine Learning Algorithms with Quantitative Phase Images of Unstained Cells.

Loading...
Thumbnail Image

Date

2016-01

Journal Title

Journal ISSN

Volume Title

Repository Usage Stats

86
views
25
downloads

Citation Stats

Abstract

Malaria detection through microscopic examination of stained blood smears is a diagnostic challenge that heavily relies on the expertise of trained microscopists. This paper presents an automated analysis method for detection and staging of red blood cells infected by the malaria parasite Plasmodium falciparum at trophozoite or schizont stage. Unlike previous efforts in this area, this study uses quantitative phase images of unstained cells. Erythrocytes are automatically segmented using thresholds of optical phase and refocused to enable quantitative comparison of phase images. Refocused images are analyzed to extract 23 morphological descriptors based on the phase information. While all individual descriptors are highly statistically different between infected and uninfected cells, each descriptor does not enable separation of populations at a level satisfactory for clinical utility. To improve the diagnostic capacity, we applied various machine learning techniques, including linear discriminant classification (LDC), logistic regression (LR), and k-nearest neighbor classification (NNC), to formulate algorithms that combine all of the calculated physical parameters to distinguish cells more effectively. Results show that LDC provides the highest accuracy of up to 99.7% in detecting schizont stage infected cells compared to uninfected RBCs. NNC showed slightly better accuracy (99.5%) than either LDC (99.0%) or LR (99.1%) for discriminating late trophozoites from uninfected RBCs. However, for early trophozoites, LDC produced the best accuracy of 98%. Discrimination of infection stage was less accurate, producing high specificity (99.8%) but only 45.0%-66.8% sensitivity with early trophozoites most often mistaken for late trophozoite or schizont stage and late trophozoite and schizont stage most often confused for each other. Overall, this methodology points to a significant clinical potential of using quantitative phase imaging to detect and stage malaria infection without staining or expert analysis.

Department

Description

Provenance

Citation

Published Version (Please cite this version)

10.1371/journal.pone.0163045

Publication Info

Park, Han Sang, Matthew T Rinehart, Katelyn A Walzer, Jen-Tsan Ashley Chi and Adam Wax (2016). Automated Detection of P. falciparum Using Machine Learning Algorithms with Quantitative Phase Images of Unstained Cells. PloS one, 11(9). p. e0163045. 10.1371/journal.pone.0163045 Retrieved from https://hdl.handle.net/10161/20286.

This is constructed from limited available data and may be imprecise. To cite this article, please review & use the official citation provided by the journal.

Scholars@Duke

Chi

Jen-Tsan Ashley Chi

Professor in Molecular Genetics and Microbiology

We are using functional genomic approaches to investigate the nutrient signaling and stress adaptations of cancer cells when exposed to various nutrient deprivations and microenvironmental stress conditions. Recently, we focus on two areas. First, we are elucidating the genetic determinants and disease relevance of ferroptosis, a newly recognized form of cell death. Second, we have identified the mammalian stringent response pathway which is highly similar to bacterial stringent response, but with some very interesting twists and novel mechanisms.

A. The genetic determinants and disease relevance of ferroptosis

Ferroptosis is a newly recognized form of cell death that is characterized by iron dependency and lipid peroxidation. The importance of ferroptosis is being recognized in many human diseases, including cancers, ischemia injuries, and neurodegeneration. Previously, we have identified the profound cystine addiction of renal cell carcinoma (1), breast cancer cells (2, 3), and ovarian cancer cells (4). Based on the concept that cystine deprivation triggers the ferroptosis due to the unopposed oxidative stresses, we have performed functional genomic screens to identify many novel genetic determinants of ferroptosis. For example, we have found that DNA damage response and ATM kinase regulate ferroptosis via affecting iron metabolism (5). This finding supports the potential of ionizing radiation to trigger DNA damage response and synergize with ferroptosis to treat human cancers. In addition, we found that ferroptosis is highly regulated by cell density. When cells are grown at low density, they are highly susceptible to ferroptosis. In contrast, the same cells become resistant to ferroptosis when grown at high density and confluency. we have found the Hippo pathway effectors TAZ and YAP are responsible for the cell density-dependent ferroptosis (4, 6, 7). Right now, we are pursuing several other novel determinants of ferroptosis that will reveal surprising insights into this new form of cell death.

B. A new stress pathway – mammalian stress response

All living organisms encounter a wide variety of nutrient deprivations and environmental stresses. Therefore, all organisms have developed various mechanisms to respond and promote survival under stress. In bacteria, the main strategy is “stringent response” triggered by the accumulation of the alarmone (p)ppGpp (shortened to ppGpp below) via regulation of its synthetase RelA and its hydrolase SpoT (8). The ppGpp binds to the transcription factor DksA and RNA polymerase to orchestrate extensive transcriptional changes that repress proliferation and promote stress survival (8, 9). While highly conserved among bacteria, the stringent response had not been reported in metazoans. However, a recent study identified Drosophila and human MESH1 (Metazoan SpoT Homolog 1) as the homologs of the ppGpp hydrolase domain of the bacterial SpoT (10). Both MESH1 proteins exhibit ppGpp hydrolase activity, and the deletion of Mesh1 in Drosophila led to a transcriptional response reminiscent of the bacterial stringent response (10). Recently, we have found that the genetic removal of MESH1 in tumor cells triggers extensive transcriptional changes and confers protection against oxidative stress-induced ferroptosis (11). Importantly, MESH1 removal also triggers proliferative arrest and other robust anti-tumor effects. Therefore, MESH1 knockdown leads to both stress survival and proliferation arrest, two cardinal features highly reminiscent of the bacterial stringent response. Therefore, we termed this pathway as “mammalian stringent response” (12). We have found that NADPH is the relevant MESH1 in the contexts of ferroptosis (13). Now, we are investigating how MESH1 removal leads to proliferation of arrests and anti-tumor phenotypes. Furthermore, we have found several other substrates of MESH1. We are investigating their function using culture cells, MESH1 KO mice, and other model organisms.

 

C. Genomic and single cell RNA analysis of Red Blood Cells

Red blood cells (RBC) are responsible for oxygen delivery to muscles during vigorous exercise. Therefore, many doping efforts focus on increasing RBC number and function to boost athletic performance during competition. For many decades, RBC were thought to be merely identical “sacs of hemoglobin” with no discernable differences due to factors such as age or pre-transfusion storage time. Additionally, because RBC lose their nuclei during terminal differentiation, they were not believed to retain any genetic materials.  These long-held beliefs have now been disproven and the results have significant implications for detecting autologous blood transfusion (ABT) doping in athletes.  We were among the first to discover that RBCs contain abundant and diverse species of RNAs. Using this knowledge, we subsequently optimized protocols and performed genomic analysis of the RBC transcriptome in sickle cell disease; these results revealed that heterogeneous RBCs could be divided into several subpopulations, which had implications for the mechanisms of malaria resistance. As an extension of these studies, we used high resolution Illumina RNA-Seq approaches to identify hundreds of additional known and novel microRNAs, mRNAs, and other RNA species in RBCs. This dynamic RBC transcriptome represents a significant opportunity to assess the impact that environmental factors (such as pre-transfusion refrigerate storage) on the RBC transcriptome. We have now identified a >10-fold change in miR-720 as well as several other RNA transcripts whose levels are significantly altered by RBC storage (14) which gained significant press coverage. We are pursuing the genomic and single cell analysis of RNA transcriptome in the context of blood doping, sickle cell diseases and other red cell diseases.

 

 

 

 

1.         Tang X, Wu J, Ding CK, Lu M, Keenan MM, Lin CC, et al. Cystine Deprivation Triggers Programmed Necrosis in VHL-Deficient Renal Cell Carcinomas. Cancer Res. 2016;76(7):1892-903.

2.         Tang X, Ding CK, Wu J, Sjol J, Wardell S, Spasojevic I, et al. Cystine addiction of triple-negative breast cancer associated with EMT augmented death signaling. Oncogene. 2017;36(30):4379.

3.         Lin CC, Mabe NW, Lin YT, Yang WH, Tang X, Hong L, et al. RIPK3 upregulation confers robust proliferation and collateral cystine-dependence on breast cancer recurrence. Cell Death Differ. 2020.

4.         Yang WH, Huang Z, Wu J, Ding C-KC, Murphy SK, Chi J-T. A TAZ-ANGPTL4-NOX2 axis regulates ferroptotic cell death and chemoresistance in epithelial ovarian cancer. Molecular Cancer Research. 2019: molcanres.0691.2019.

5.         Chen PH, Wu J, Ding CC, Lin CC, Pan S, Bossa N, et al. Kinome screen of ferroptosis reveals a novel role of ATM in regulating iron metabolism. Cell Death Differ. 2019.

6.         Yang W-H, Chi J-T. Hippo pathway effectors YAP/TAZ as novel determinants of ferroptosis. Molecular & Cellular Oncology. 2019:1699375.

7.         Yang WH, Ding CKC, Sun T, Hsu DS, Chi JT. The Hippo Pathway Effector TAZ Regulates Ferroptosis in Renal Cell Carcinoma Cell Reports. 2019;28(10):2501-8.e4.

8.         Potrykus K, Cashel M. (p)ppGpp: still magical? Annu Rev Microbiol. 2008;62:35-51.

9.         Kriel A, Bittner AN, Kim SH, Liu K, Tehranchi AK, Zou WY, et al. Direct regulation of GTP homeostasis by (p)ppGpp: a critical component of viability and stress resistance. Mol Cell. 2012;48(2):231-41.

10.       Sun D, Lee G, Lee JH, Kim HY, Rhee HW, Park SY, et al. A metazoan ortholog of SpoT hydrolyzes ppGpp and functions in starvation responses. Nat Struct Mol Biol. 2010;17(10):1188-94.

11.       Dixon SJ, Lemberg KM, Lamprecht MR, Skouta R, Zaitsev EM, Gleason CE, et al. Ferroptosis: an iron-dependent form of nonapoptotic cell death. Cell. 2012;149(5):1060-72.

12.       Ding C-KC, Rose J, Wu J, Sun T, Chen K-Y, Chen P-H, et al. Mammalian stringent-like response mediated by the cytosolic NADPH phosphatase MESH1. bioRxiv. 2018.

13.       Ding C-KC, Rose J, Sun T, Wu J, Chen P-H, Lin C-C, et al. MESH1 is a cytosolic NADPH phosphatase that regulates ferroptosis. Nature Metabolism. 2020.

14.       Yang WH, Doss JF, Walzer KA, McNulty SM, Wu J, Roback JD, et al. Angiogenin-mediated tRNA cleavage as a novel feature of stored red blood cells. Br J Haematol. 2018.

 

 

Wax

Adam P. Wax

Professor of Biomedical Engineering

Dr. Wax's research interests include optical spectroscopy for early cancer detection, novel microscopy and interferometry techniques.

The study of intact, living cells with optical spectroscopy offers the opportunity to observe cellular structure, organization and dynamics in a way that is not possible with traditional methods. We have developed a set of novel spectroscopic techniques for measuring spatial, temporal and refractive structure on sub-hertz and sub-wavelength scales based on using low-coherence interferometry (LCI) to detect scattered light. We have applied these techniques in different types of cell biology experiments. In one experiment, LCI measurements of the angular pattern of backscattered light are used to determine non-invasively the structure of sub-cellular organelles in cell monolayers, and the components of epithelial tissue from freshly excised rat esophagus. This work has potential as a diagnostic method for early cancer detection. In another experiment, LCI phase measurements are used to examine volume changes of epithelial cells in a monolayer in response to environmental osmolarity changes. Although cell volume changes have been measured previously, this work demonstrates for the first time the volume of just a few cells (2 or 3) tracked continuously and in situ.


Unless otherwise indicated, scholarly articles published by Duke faculty members are made available here with a CC-BY-NC (Creative Commons Attribution Non-Commercial) license, as enabled by the Duke Open Access Policy. If you wish to use the materials in ways not already permitted under CC-BY-NC, please consult the copyright owner. Other materials are made available here through the author’s grant of a non-exclusive license to make their work openly accessible.