ANHIR: Automatic Non-Rigid Histological Image Registration Challenge - UTU Research Portal

A1 Refereed original research article in a scientific journal

ANHIR: Automatic Non-Rigid Histological Image Registration Challenge

Authors: Borovec J, Kybic J, Arganda-Carreras I, Sorokin DV, Bueno G, Khvostikov AV, Bakas S, Chang EIC, Heldmann S, Kartasalo K, Latonen L, Lotz J, Noga M, Pati S, Punithakumar K, Ruusuvuori P, Skalski A, Tahmasebi N, Valkonen M, Venet L, Wang YZ, Weiss N, Wodzinski M, Xiang Y, Xu Y, Yan Y, Yushkevich P, Zhao SY, Munoz-Barrutia A, Munoz-Barrutia A

Publisher: IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC

Publication year: 2020

Journal: IEEE Transactions on Medical Imaging

Journal name in source: IEEE TRANSACTIONS ON MEDICAL IMAGING

Journal acronym: IEEE T MED IMAGING

Volume: 39

Issue: 10

First page : 3042

Last page: 3052

Number of pages: 11

ISSN: 0278-0062

eISSN: 1558-254X

DOI: https://doi.org/10.1109/TMI.2020.2986331

Self-archived copy’s web address: https://europepmc.org/articles/pmc7584382?pdf=render

Abstract

Automatic Non-rigid Histological Image Registration (ANHIR) challenge was organized to compare the performance of image registration algorithms on several kinds of microscopy histology images in a fair and independent manner. We have assembled 8 datasets, containing 355 images with 18 different stains, resulting in 481 image pairs to be registered. Registration accuracy was evaluated using manually placed landmarks. In total, 256 teams registered for the challenge, 10 submitted the results, and 6 participated in the workshop. Here, we present the results of 7 well-performing methods from the challenge together with 6 well-known existing methods. The best methods used coarse but robust initial alignment, followed by non-rigid registration, used multiresolution, and were carefully tuned for the data at hand. They outperformed off-the-shelf methods, mostly by being more robust. The best methods could successfully register over 98% of all landmarks and their mean landmark registration accuracy (TRE) was 0.44% of the image diagonal. The challenge remains open to submissions and all images are available for download.