Automated Malaria Parasite Identification and Classification Using Deep Learning Techniques

The parasites which transmit malaria are part of the Plasmodium genus that transmit the disease via the bite of a mosquito. The traditional method of diagnosing this illness involves examining the patient's blood cells under a microscope, which show discoloration Examining a blood sample under...

Full description

Saved in:
Bibliographic Details
Published in2024 International Conference on Science Technology Engineering and Management (ICSTEM) pp. 1 - 6
Main Authors N, Mohana Priya, C, Bennila Thangammal, U, Leelavathi, KPV, Pinkey Roshan, B, Vennila Priya, Radley, Sheryl
Format Conference Proceeding
LanguageEnglish
Published IEEE 26.04.2024
Subjects
Accuracy
Automation
Blood cells
Cells (biology)
CNN
Data models
Dataset
Diagnosis
Geneus Plasmodium
Image Classification
Image processing
Image recognition
Infected cells
Machine learning
Machine learning algorithms
Malaria
Microscope
Microscopy
Mosquito bite
Neural network
Parasites
Supervised Learning
Time measurement
Online AccessGet full text

Cover

More Information
Summary:The parasites which transmit malaria are part of the Plasmodium genus that transmit the disease via the bite of a mosquito. The traditional method of diagnosing this illness involves examining the patient's blood cells under a microscope, which show discoloration Examining a blood sample under a microscope allows for the identification of the quantity of pathogenic red blood cells. A qualified technician looks attentively at the slide, identifying elements that are obviously prominent along with those which are subtle. It requires a lot of time and diligence to recognize and measure Plasmodium parasites in blood cells under a microscope. We have developed a novel image processing technique able to recognize and count those parasites on blood specimen slides to solve this problem Additionally, they developed an algorithm utilising machine learning that has the capability to recognise, detect, and categorise various types of infected cells based on their distinct characteristics. Using a standard malaria dataset containing numerous images of normal and contaminated blood cells, we assessed the algorithm's performance. A diverse range of quantitative specifications, including multi-class comparisons, indicated excellent outcomes from both the proposed and subsequently optimized models. Despite the use of authentic patient-level data, that have mismatched class sizes, this was still accomplished. Therefore, they have created a neural network that efficiently and accurately filters data with the lowest number of mistakes possible.
DOI:10.1109/ICSTEM61137.2024.10560969