Early Detection of Black Fungus Using Deep Learning Models for Efficient Medical Diagnosis

Mucormycosis, also known as black fungus, is a rare infection caused by mould that can affect the lungs, brain, skin, and sinuses. People with weakened immune systems due to underlying health conditions (e.g., organ transplant, uncontrolled diabetes, cancer, neutropenia, AIDS, and COVID-19) or drug use are more likely to contract the infection. Recent reports have indicated that a small number of COVID-19 survivors with co-morbidities are particularly vulnerable to black fungus. Therefore, recovered COVID-19 patients should seek medical attention immediately if they experience any mucormycosis symptoms. Patients with COVID-19 have numerous concerns regarding this life-threatening infection, particularly those with immunocompromised states such as solid organ transplant (SOT) and hematopoietic stem cell transplant (HSCT) or tumors, who are more likely to be infected. Early diagnosis and treatment are essential to prevent patient mortality. In this paper, we present the Multi-Class Black Fungus (MCBF) dataset, which consists of three compressed folders containing images. The first folder includes 66 images of eye black fungus after data preprocessing, while the second folder contains 63 images of mouth black fungus after data preprocessing. The third contains 62 images of the mouth after applying data preprocessing, and the last one contains 79 images of black skin fungus after applying data preprocessing. The MCBF dataset contains images that aid in training and validation when using deep learning algorithms to recognise and classify black fungus diseases, then we apply a resnet(50) pretrained model by using MCBF accuracy reaches 96.12%.