NoduleNet: Deep Learning-Based Computational Model for Thyroid Cancer Detection
Year: 2026
Type: conference-paper
Abstract: A thyroid is gland which is placed in the neck, which controls the functions of the body like, heart-rate, bloodpressure, and body-weight. The suggested models are developed in TensorFlow using pre-trained CNN architectures and evaluated with respect to classification performance, robustness, and potential use in real scenarios clinical usage. Early, and accurate detection is vital to reduce morbidity and improve treatment outcomes. We developed a framework using deep-learning to classify the nodules as benign and malignant. TensorFlow is used to design CNN architectures and calculated the robustness and accuracies. Variations in model depth, convolutional and pooling strategies, activation functions, and regularization techniques are explored to identify an optimal configuration for ultrasound imaging data. Preprocessing steps such as data augmentation are applied to ensure the model is generalized. The results from the experiments shows the model achieves high accuracy maintaining the latency, making it best for deployment. This work highlights the potential of lightweight, end-to-end CNN models to support radiologists with fast, objective, and reproducible thyroid cancer diagnosis.
Institution: ABES Engineering College
Language: English
Topic: AI in cancer detection
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7204189750