Modified Convolutional Neural Network Architecture with XGBoost for Mucormycosis Detection and compare performance with XGBoost
Panthangi Venkata Sai Charan, G. Ramkumar
Abstract
Patients with coronavirus illness 2019, especially those in India, are more likely to see an increase in rhino-orbital mucormycosis. A well-known risk factor during COVID-19 infection and mucormycosis is diabetes mellitus (DM). With roughly 0.15 instances per 1000 people, mucormycosis is almost 82 times more common in India than it is in Western nations. Additionally, this fungus expanded quickly across numerous states, leading some of them to designate this illness an epidemic. Finding a solution for this potentially fatal fungal infection requires the aid of modern technologies, including artificial intelligence and data learning. In this paper, we combine a modified convolutional learning neural network with an XGBoost classifier to propose a unique black fungus detection method. Under the right circumstances, the CNNXGB model is made simpler by lowering the no of attributes since it is not essential to re-adjust the weight values throughout a back propagation cycle. On testing data, the mean classification performance is 98.26%, far better than current approaches.