Automated Detection of Necrotizing Fasciitis in Patient Affected Area Images using YOLO v9

Abstract

Necrotizing fasciitis is often regarded as a clinical and surgical emergency characterized by rapid onset, swift progression, and a significant mortality rate. Often because of atypical clinical presentation, the disease evades early diagnosis and subjectively gives way to delayed treatment with an increased risk for severe complications from septic shock and multi-organ failure. There is no absolute diagnostic indicator that may offer clear and consistent early detection of NF, hence the applications are at the same time based on clinical judgment and ancillary tests, but most importantly on the experience of the physician. This study looks into the possible use of a deep learning model utilizing YOLO v9, which automatically detects NF in images of the affected areas of the patient's body obtained from patients suspected to be infected. Analysis of annotated images dataset, therefore, is primarily targeted at early improvement in detection accuracy with a view to facilitating prompt diagnosis and treatment. Results thus obtained indicate a model boosting the diagnostic precision which would eventually decrease morbidity and mortality rates on matters related to necrotizing fasciitis.