Electrical Engineering and Systems Science > Image and Video Processing
[Submitted on 21 Dec 2024]
Title:Patherea: Cell Detection and Classification for the 2020s
View PDF HTML (experimental)Abstract:This paper presents a Patherea, a framework for point-based cell detection and classification that provides a complete solution for developing and evaluating state-of-the-art approaches. We introduce a large-scale dataset collected to directly replicate a clinical workflow for Ki-67 proliferation index estimation and use it to develop an efficient point-based approach that directly predicts point-based predictions, without the need for intermediate representations. The proposed approach effectively utilizes point proposal candidates with the hybrid Hungarian matching strategy and a flexible architecture that enables the usage of various backbones and (pre)training strategies. We report state-of-the-art results on existing public datasets - Lizard, BRCA-M2C, BCData, and the newly proposed Patherea dataset. We show that the performance on existing public datasets is saturated and that the newly proposed Patherea dataset represents a significantly harder challenge for the recently proposed approaches. We also demonstrate the effectiveness of recently proposed pathology foundational models that our proposed approach can natively utilize and benefit from. We also revisit the evaluation protocol that is used in the broader field of cell detection and classification and identify the erroneous calculation of performance metrics. Patherea provides a benchmarking utility that addresses the identified issues and enables a fair comparison of different approaches. The dataset and the code will be publicly released upon acceptance.
Current browse context:
eess.IV
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.