Apply Deep Neural Network to Detect Malignant Lung Nodules in CT Scans

Lung cancer has a poor prognosis and a high incidence in low resource settings. Early detection of malignant nodules can enable prompt intervention and improve treatment outcomes. Despite the recognized benefits of early nodule detection, it is clinically and computational challenging. Convolutional neural network (CNN) is a type of machine learning that has been applied to analyzing visual images, and has received increasing interest in facilitating image recognition in medical field. The study will apply CNN to pre-labeled low dosage computed tomography (CT) scans collected in Zhongshan Hospital (Shanghai) to differentiate malignant nodules from benign nodules. A training set will be used to train the model first, followed by fine-tuning the model to increase the diagnosis accuracy and sensitivity. A set number of CT scans will be used as the test set to examine the accuracy rate.

Faculty Supervisor:

Tillie-Louise Hackett

Student:

Partner:

Fudan University

Discipline:

Life Sciences

Sector:

Education

University:

The University of British Columbia

Program: