CaltechAUTHORS
  A Caltech Library Service

J Regularization Improves Imbalanced Multiclass Segmentation

Guerrero Peña, Fidel A. and Marrero Fernandez, Pedro D. and Tarr, Paul T. and Ren, Tsang Ing and Meyerowitz, Elliot M. and Cunha, Alexandre (2020) J Regularization Improves Imbalanced Multiclass Segmentation. In: 2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI). IEEE , Piscataway, NJ, pp. 948-952. ISBN 9781538693308. https://resolver.caltech.edu/CaltechAUTHORS:20200528-143929426

[img] PDF - Submitted Version
Creative Commons Attribution Non-commercial Share Alike.

2335Kb

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20200528-143929426

Abstract

We propose a new loss formulation to further advance the multiclass segmentation of cluttered cells under weakly supervised conditions. When adding a Youden's J statistic regularization term to the cross entropy loss we improve the separation of touching and immediate cells, obtaining sharp segmentation boundaries with high adequacy. This regularization intrinsically supports class imbalance thus eliminating the necessity of explicitly using weights to balance training. Simulations demonstrate this capability and show how the regularization leads to correct results by helping advancing the optimization when cross entropy stagnates. We build upon our previous work on multiclass segmentation by adding yet another training class representing gaps between adjacent cells. This addition helps the classifier identify narrow gaps as background and no longer as touching regions. We present results of our methods for 2D and 3D images, from bright field images to confocal stacks containing different types of cells, and we show that they accurately segment individual cells after training with a limited number of images, some of which are poorly annotated.


Item Type:Book Section
Related URLs:
URLURL TypeDescription
https://doi.org/10.1109/isbi45749.2020.9098550DOIArticle
https://arxiv.org/abs/1910.09783arXivDiscussion Paper
ORCID:
AuthorORCID
Meyerowitz, Elliot M.0000-0003-4798-5153
Cunha, Alexandre0000-0002-2541-6024
Additional Information:© 2020 IEEE. We thank financial support from the Brazilian funding agencies FACEPE, CAPES and CNPq (FAG, PF, TIR), from the Beckman Institute at Caltech to the Center for Advanced Methods in Biological Image Analysis (AC, FAG), from the Howard Hughes Medical Institute (PTT, EMM), and thank the IBM Matching Grants Program for computer donation (AC).
Funders:
Funding AgencyGrant Number
Fundação do Amparo a Ciência e Tecnologia (FACEPE)UNSPECIFIED
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)UNSPECIFIED
Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)UNSPECIFIED
Caltech Beckman InstituteUNSPECIFIED
Howard Hughes Medical Institute (HHMI)UNSPECIFIED
Subject Keywords:Loss modeling, deep learning, instance segmentation, multiclass segmentation, cell segmentation, data imbalance
Record Number:CaltechAUTHORS:20200528-143929426
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20200528-143929426
Official Citation:F. A. Guerrero Peña, P. D. Marrero Fernandez, P. T. Tarr, T. I. Ren, E. M. Meyerowitz and A. Cunha, "J Regularization Improves Imbalanced Multiclass Segmentation," 2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI), Iowa City, IA, USA, 2020, pp. 1-5, doi: 10.1109/ISBI45749.2020.9098550
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:103528
Collection:CaltechAUTHORS
Deposited By: Tony Diaz
Deposited On:28 May 2020 21:49
Last Modified:29 Oct 2020 23:52

Repository Staff Only: item control page