Scale Aware Multi-Instance Learning for Early Prognosis of Subjects at Risk of Developing Hepatocellular Carcinoma

Ananya Jana, Ramanathan Arunachalam, Carlos D Minacapelli, Kaitlyn Catalano, Carlos Joaquin Catalano, Vinod K Rustgi, Dimitris N. Metaxas

Keywords: Hepatocellular carcinoma, liver cancer, Deep Learning, HCC prognosis, early detection

TL;DR: Scale-Aware Classification of pre-HCC

Abstract:

Hepatocellular carcinoma (HCC) is a type of primary liver cancer and it can lead to the subject’s death. The subject may not experience symptoms at the onset of HCC. The symptoms usually occur when the HCC has grown significantly and needs more advanced treatment like surgery, extensive chemotherapy, etc. Thus it would be beneficial to classify subjects at risk of developing HCC based on non-invasive CT scans.
In this work, we propose a novel method for pre-cancer stage classification. As per our knowledge, we are the first to propose a Deep Learning based method to classify pre-cancer subjects based on CT images. The subjects at risk of developing HCC may have grown different scales/levels of visual cues on the CT scans. So, we propose a scale-aware MIL method to facilitate the classification with no additional annotation cost. We show the efficacy of our method on a dataset of 60 subjects

Registration: I acknowledge that publication of this at MIDL and in the proceedings 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: methodological development

Primary Subject Area: Detection and Diagnosis

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.