- Keywords: PseudoEdgeNet, YOLO, PD-L1, lung cancer, histopathology
- TL;DR: We extend PseudoEdgeNet for the multi-class task of PD-L1 stained cell detection in non-small cell lung cancer histopathology images and benchmark it against YoloV5.
- Abstract: In this paper, we take the recently presented PseudoEdgeNet model to the level of multi-class cell segmentation in histopathology images solely trained with point annotations. We tailor its loss function to the challenges of multi-class segmentation and equip it with an additional false positive loss term. We evaluate it on the assessment of tumor and immune cells in PD-L1 stained lung cancer histopathology images, and compare it with YOLOv5.
- Paper Type: validation/application paper
- Source Latex: zip
- Primary Subject Area: Detection and Diagnosis
- Secondary Subject Area: Application: Histopathology
- Paper Status: original work, not submitted yet
- Source Code Url: All code used for the YoloV5 model was directly taken from the open-source Github repo for Yolov5 by Ultralytics: https://github.com/ultralytics/yolov5 The code used for the PseudoEdgeNet model is currently a work in progress and contains multiple novel contributions that we do not fully elaborate on in this paper.
- Data Set Url: The work in this paper is fully based on a dataset collected in-house at our hospital - we currently do not have permission to share this data in an open-source way.
- 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.