Keywords: Colorectal cancer surgery, Ensemble learning, Data scarcity, Ultrasound
TL;DR: Ensemble of models pre-trained on breast ultrasound images for colorectal tumor detection in intraoperative ultrasound images
Abstract: Cancer surgery is characterized by a delicate balance between radical tumor resection and sparing healthy tissue and critical anatomical structures. The trouble of recognizing tissue structures during surgery may either lead to resection too close to the tumor resulting in tumor-positive resection margins or too wide resection around the tumor with potential damage to vital anatomical structures. Ultrasound is a widely available and non-invasive imaging technique which can be used for surgical guidance by continuous real-time tissue recognition during surgery, however, interpretation of US images requires training and experience. One of the notorious challenges in medical image analysis is the scarcity of labeled data. To address this issue, we introduce a deep ensemble learning framework for colorectal tumor detection in ultrasound images using models which are pre-trained for tumor segmentation in breast ultrasound images.
Registration: I acknowledge that acceptance of this work at MIDL requires at least one of the authors to register and present the work during the conference.
Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.
Paper Type: novel methodological ideas without extensive validation
Primary Subject Area: Segmentation
Secondary Subject Area: Learning with Noisy Labels and Limited Data
Confidentiality And Author Instructions: I read the call for papers and author instructions. I acknowledge that exceeding the page limit and/or altering the latex template can result in desk rejection.