Towards a portable mid-infrared tool for analysis of mosquito vectors of malaria

Pazmino Betancourth, Mauro (2023) Towards a portable mid-infrared tool for analysis of mosquito vectors of malaria. PhD thesis, University of Glasgow.

Full text available as:
[thumbnail of 2022PazminoBetancourthPhD.pdf] PDF
Download (17MB)

Abstract

Mid-infrared spectroscopy (MIRS) has emerged as a potential tool to predict species, age and infection in the Anopheles malaria mosquitoes as well as in other disease vectors. The main advantages of optical methods in general are their speed, little or no sample preparation, label-free, lower cost and already established protocols and analysis pipelines. New rapid, low cost, high-throughput tools for vector surveillance are urgently needed to develop and optimise new vector control strategies, as vector borne diseases (VBD) are spreading around the globe due to climate change and globalisation, and endemic countries are suffering resurgence of malaria cases following weakening of control tools. However, the current commercially available FTIR spectrometers have limitations. They are expensive, bulky and low power that hider its implementation in the field. Quantum cascade lasers (QCL) have become an alternative to FTIR light sources due to their unique characteristics (i.e. coherence, high power in a smaller spot size, small chip size), which allows easier implementation for the field due to its lower cost, practicality, and accuracy. These characteristics can expand the possibilities to develop new ways to measure spectral information from disease vectors. This thesis is aimed at developing a QCL-based spectrometer, understanding the most informative infrared region for VBD surveillance and the use of legs for surveillance and prediction of key traits from mosquitoes using MIRS.

In this project, micro-diffuse reflectance spectroscopy (µDRIFT) was used on mosquito legs to predict age, species and cuticular insecticide resistance. Indeed, spectra from legs led to high accuracy ML models for age prediction (overall model accuracy: 77.1% (± 6.5%) with a mean accuracy of 82% for 3 days old and 74% for 10 days old) and moderate accuracy for species identification (overall model accuracy: 69.1% (± 7.9%) with a mean accuracy of 68% for An. gambiae and 71% for An. coluzzii). Finally, cuticular resistance in three strains of Anopheles mosquitoes was identified with high accuracy when grouped into susceptible and resistant classes (overall model accuracy: 71.3% (± 8%) with a mean accuracy of 73% for susceptible and 71% for resistant class). However, these preliminary findings need to further be confirming by ruling out confounding factors such as the use of different strains of Anopheles by using a single strain with various degrees of insecticide resistant. I found that Partial Least Squares Discriminant Analysis (PLS-DA) and can be used for high accuracy prediction between An. gambiae and An. coluzzii when tested on laboratory samples from the same origin (mean accuracy: 87%). However, species prediction decreases when the model is tested on samples from different laboratories (mean accuracy: 62%) and in semi-field samples (mean accuracy: 46.5%). For age prediction, PLS regression was able to predict different group ages (3, 5, 7, 9, 12, 15 days old) when tested with laboratory samples from the same origin (R2 = 0.68, RMSE = 2.24) and with samples from other laboratories (R2 = 0.78, RMSE = 1.89). Nevertheless, the model cannot predict the age of semi-field samples (R2= -1.84, RMSE = 7.99). Also, I found narrower spectral windows of ≈ 300 cm−1 in length located on the Amide I and Amide II region are sufficient to predict mosquito species using machine learning (accuracy from 88% to 98%). This can help for a more efficient way of collecting spectral data. Future work should focus on how to improve model calibration by adding samples with diverse origin (different laboratories, different rearing conditions) to improve model generalisation. Finally, I have developed a QCL-based spectrometer in the range of 8-11 µm with scan speeds up to 500 Hz, with a maximum tuning rate of 400 µm/s. The system can collect spectra from polymers (polypropylene, polyethylene terephthalate and polyethylene) and biological samples (mosquitoes) in transmission mode. When compared to commercial FTIRs, MIRS measurements of whole mosquito bodies in KBr discs through the QCL-based spectrometer were in high agreement at bands 988, 1029 and 1056 cm−1 showing that the newly developed device works in mosquitoes. This study has made the first step towards the use of QCL-based system for spectroscopy of insect disease vectors, opening new opportunities for the implementation and use of midinfrared spectroscopy for vector-borne disease surveillance.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Subjects: T Technology > T Technology (General)
Colleges/Schools: College of Science and Engineering > School of Engineering
Supervisor's Name: Hogg, Professor Richard
Date of Award: 2023
Depositing User: Theses Team
Unique ID: glathesis:2023-83455
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 27 Feb 2023 15:00
Last Modified: 28 Feb 2023 09:02
Thesis DOI: 10.5525/gla.thesis.83455
URI: https://theses.gla.ac.uk/id/eprint/83455
Related URLs:

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year