Severity Assessment of Cervical Lymph Nodes using Modified VGG-Net, and Squeeze and Excitation Concept.

Severity assessment of cervical lymph nodes (CLNs) in terms of benign and malignant categories play a significant part in the treatment management of patients having Head and Neck Cancer. Severity assessment through invasive pathological tests like biopsy are painful and time-consuming procedures. Computed tomography (CT) is an extensively used and chosen non-invasive radiological modality for imaging evaluation of all oncological diseases. Manual evaluation of CT images is a time consuming and complex job. Hence, in this paper authors have proposed state-of-the-art deep learning-based automated computer-aided detection (CAD) system for the classification of benign, and malignant CLNs. In the proposed methodology authors have modified the popular VGG-Net, and Squeeze and Excitation (SE) concept. Further, the residual concept is also utilized to enhance the performance without increasing the computation complexity. For this work, the dataset is collected from the Regional Cancer Center (RCC) of Raipur Chhattisgarh, India. The achieved best performance parameters are sensitivity = 96.81%, specificity = 95.51%, accuracy = 96.56%, and area under curve = 96.16%.