Abstract

Detection of mitotic tumor cells per tissue area is one of the critical markers of breast cancer prognosis. The aim of this paper is to develop a method for the automatic detection of mitotic figures from breast cancer histological slides using a partially supervised deep learning framework. Unlike the previous literature, which has focused on solving the problem of mitosis detection in the weakly annotated datasets using centroid pixel labels (weak labels) only without taking advantage of the available pixel-level labels (strong labels) of other datasets, in this paper, we design a novel partially supervised framework based on two parallel deep fully convolutional networks. One of them is trained using weak labels and the other is trained using strong labels, together with a weight transfer function. In the detection phase, we fuse the segmentation maps produced by the two networks to obtain the final mitosis detections. Our system exploits the available large sets of mitosis detection samples with mitosis centroid annotation, such as the 2014 ICPR dataset and the AMIDA13 dataset, and only a small set of samples with the annotation of all mitosis pixels, such as the 2012 ICPR dataset, to perform a more accurate mitosis detection on weakly labeled data. This enables us to outperform all previous mitosis detection systems by achieving F-scores of 0.575 and 0.698 on the 2014 ICPR dataset and the AMIDA13 dataset respectively.

Highlights

  • The most recommended breast cancer grading system by the World Health Organization (WHO), is the Nottingham grading system [1]

  • With the availability of two different types of annotations in mitosis benchmarks, centroid-pixel annotations and pixellevel annotations, a key question can be raised: Is it possible to train a mitosis detection model using both strong and weak labels? With this motivation, in the present paper we introduce a novel method based on partially supervised semantic segmentation for mitosis detection from histopathological slide images

  • A partially supervised deep learning framework for accurate and reliable mitosis detection from Hematoxylin and Eosin (H&E) stained histopathological images has been introduced in this paper

Read more

Summary

Introduction

The most recommended breast cancer grading system by the World Health Organization (WHO), is the Nottingham grading system [1]. It involves three biomarkers: tubule formation, nuclear pleomorphism score, and mitosis (i.e., cell in the process of nuclear division) counting. Because the spread of cancer is highly related to cellular divisions, detecting the mitotic cells in histopathology images and counting them is the most important indicator for assessing the risk of metastasis. The breast biopsied tissue specimens are fixed by paraffin and stained with Hematoxylin and Eosin (H&E) dyes. Pathology experts manually mark the mitotic cells on High Power Fields (HPFs), which are microscopic observations with 40x magnification.

Objectives
Methods
Findings
Conclusion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call