The University of Oslo is closed. The PhD defence and trial lecture will therefore be fully digital and streamed directly using Zoom. The host of the session will moderate the technicalities while the chair of the defence will moderate the disputation.
Ex auditorio questions: the chair of the defence will invite the audience to ask ex auditorio questions either written or oral. This can be requested by clicking 'Participants -> Raise hand'.
June 9, 10.00 AM, Zoom
"Encoder and decoder networks for medical segmentation"
Main research findings
Colorectal cancer is the third most common cancer diagnosed in both men and women worldwide. It starts from small non-cancerous growths of tissues called polyps. Some polyps may become malignant if left untreated. Colonoscopy is the standard tool for the early diagnosis of the colon, but it is time consuming and an operator-dependent procedure, around 22%-28% of polyps may be missed during colonoscopy.
This thesis investigates the potential of deep learning in the form of convolutional neural networks to automate the detection of colonic polyps. Different algorithms are developed to overcome the challenges associated with the complexity of the colon and the limitations in the available data. The developed algorithms are designed to understand the complex environment of the inner lining of the colon and distinguish various polyp-like structures mimicking real polyps. The algorithms are invirent to inter-class variation in polyps appearances in terms of size, shape, color, and texture, and light condition. Using the knowledge gathered from the experiments, a real-time polyp detection system was developed, achieving up to 91% of sensitivity and 89% of precision.
The achieved results have proven the capability and potential of the proposed methods which can be further improved and used for automatic review of videos of Wireless Capsule Endoscopy, thereby limiting the excessive use of manpower, and saving the lives of millions of patients suffering from colorectal cancer.
Contact information to Department: Pernille Adine Nordby