38th Annual Student Research Conference
Lung and Colon Cancer Detection Using Deep AI Model
Ritika Lama
Dr. Nazmul Shahadat, Faculty Mentor
Lung and colon cancers rank among the top causes of cancer-related deaths globally, underscoring the need for early and precise detection to enhance treatment success and patient survival. Inaccurate detection can lead to detrimental outcomes, making reliable diagnosis critical. Traditional tissue analysis is complex and slow, but deep learning has improved efficiency and accuracy, enabling faster, cost-effective studies of larger patient cohorts. Despite extensive research, no model has achieved perfect detection rates for these deadly cancers, as misclassification carries severe risks. This study introduces a novel, lightweight, and mobile-compatible deep learning model using a 1D convolutional neural network with squeeze-and-excitation layers. Tested on the LC25000 dataset, it achieves 100% accuracy in detecting lung squamous cell carcinoma, lung adenocarcinoma, and colon cancer from histopathological images. With just 0.35 million parameters and 6.4 million flops, this model outperforms existing architectures, setting a new benchmark in cancer detection.
Keywords: Deep learning, Convolutional Neural Network, Cancer Detection, Image Classification
Topic(s):Computer Science
Biology
Presentation Type: Poster Presentation
Session: 400-27
Location: SUB Activities Room
Time: 9:00