In a recent study published in Medicine in Novel Technology and Devices Journal, researchers used a large dataset including images of skin lesions from monkeypox (mpox) patients to develop a machine learning-based detection tool for detect mpox.
Study: Deep learning-based detection of monkeypox virus using skin lesion imaging. Image credit: sulit.photos/Shutterstock.com
Background
Mpox is a zoonotic systemic disease caused by the monkeypox virus (MPV), which belongs to the Orthox virus genus of the family Poxviridae.
Until early 2022, the disease was endemic in West and Central Africa, but cases of monkeypox have been reported from more than one hundred countries outside the endemic region by the end of 2022, and this recent spread of MPV in North America and Europe is considered a global outbreak.
Along with fever, headache, body aches, and swollen lymph nodes, the disease also causes rashes and lesions on the palms of the hands, soles of the feet, and face, and on the mucous membranes of the mouth and genital regions.
The rashes begin on the soles of the feet and palms of the hands, spread to the eyes, genitals, and mouth, and usually progress from a flat or mottled shape to firm, raised lesions called papules that eventually fill with pus to form pustules .
The current standard method for detecting monkeypox uses polymerase chain reaction (PCR) tests, which can often be inconclusive due to the short duration the virus remains in the body and/or inaccessible in rural and remote areas .
However, artificial intelligence and machine learning methods provide faster and more accessible methods of diagnosing diseases.
About the studio
The present study developed a model based on deep learning methods to detect mpox using images of skin lesions taken with ordinary smartphone cameras. The study aimed to use various deep learning methods, including AlexNet and GoogLeNet, to accurately detect mpox.
They also compared the performance metrics of other machine learning models used to diagnose mpox in terms of accuracy, recall, precision, and f1 score.
The training dataset included 228 images, of which 102 were of mpox and the remaining 126 were of measles and varicella lesions. Various augmentation methods such as translation, rotation, cropping, reflection, hue, contrast, brightness, saturation, and scaling were used to augment the data set, which consisted of 1,428 mpox lesion images and 1,764 photos of other lesions.
Deep neural networks were trained using the training image dataset in Deep Network Designer running on MATLAB 2022. Pilot runs were conducted for several neural networks including Places365-GoogleNet, GoogLeNet, AlexNet, SqueezeNet, Vision Transformer and ResNet-18.
Results
The results reported that of all the neural networks tested, the ResNet-18 results had the highest accuracy (99.49%).
The researchers believe that ResNet-18 performed with better accuracy than Places365-GoogleNet, Squeezenet and GoogLeNet due to its effective and simple architecture, which allowed them to learn the complex functionality of the sensing method without numerous inputs. ResNet-18 also has fewer convolutional layers than other neural networks and requires less computer memory.
The Vision Transformer model was used as an alternative to conventional neural network models and was found to perform poorly compared to neural network models when using similar training and validation hyperparameters.
This performance difference could be due to vision transformation models that require a large training data set due to their many parameters.
Deep learning methods in medicine provide faster and more accurate testing options. They can efficiently filter large amounts of patient data without compromising accuracy and time.
Furthermore, resource efficiency and lack of heavy or expensive equipment make it an ideal mpox detection method in various healthcare facilities and clinics in various regions.
Conclusions
To summarize, the researchers used a large dataset of mpox lesions and measles and chickenpox lesions to train various neural networks to detect mpox cases from images taken on easily accessible smartphone cameras.
Overall, the results indicated that the ResNet-18 neural network model performed best, with an accuracy of 99.49%.
Additionally, with other techniques, such as locally interpretable model agnostic explanations (LIME), healthcare professionals can potentially use this method to detect mpox and visually interpret predictions based on neural network model results.
#Deep #learning #models #crucial #fight #monkeypox #virus