Form Follows Function: Smart Network Design Enables Zero-shot Network Reuse

Fu W, Husvogt L, Breininger K, Schaffert R, Abu-Qamar O, Fujimoto JG, Maier A (2022)

Publication Type: Conference contribution

Publication year: 2022

Journal

Publisher: Springer Science and Business Media Deutschland GmbH

Pages Range: 121-126

Conference Proceedings Title: Informatik aktuell

Event location: Heidelberg, DEU

ISBN: 9783658369316

DOI: 10.1007/978-3-658-36932-3_26

Abstract

In this work, we construct and train a learning-based neural network pipeline for retinal vessel segmentation on fundus images and transfer the preprocessing module directly onto OCTA en face projections. Without additional fine-tuning, the transferred module retains the edge-preserving denoising functionality, as our smart network design enables zero-shot domain adaptation. Compared to a denoising network trained on OCTA data, the transferred preprocessing module is superior with regard to performance, generalization ability, and numerical stability. In addition, Frangi’s vessel segmentation of the preprocessed images outperforms the predictions using pretrained full segmentation networks. The selection of fitting Gaussian scales of the Frangi filter in the target domain can bypass the over/under-segmentation problem caused by the vessel diameter mismatch between the two domains. The contribution of this work is two-fold: we provide an exemplary evidence that smart design of network pipelines allows for flexible functional module reuse across different imaging modalities, i.e. zero-shot domain adaptation; and we discover a deep learning-based vessel segmentation pipeline for OCTA en face images which reaches an AUC score of 0.9418 without additional training.

Authors with CRIS profile

Involved external institutions

Tufts Medical Center United States (USA) (US) Massachusetts Institute of Technology (MIT) United States (USA) (US)

How to cite

APA:

Fu, W., Husvogt, L., Breininger, K., Schaffert, R., Abu-Qamar, O., Fujimoto, J.G., & Maier, A. (2022). Form Follows Function: Smart Network Design Enables Zero-shot Network Reuse. In Klaus Maier-Hein, Thomas M. Deserno, Heinz Handels, Andreas Maier, Christoph Palm, Thomas Tolxdorff (Eds.), Informatik aktuell (pp. 121-126). Heidelberg, DEU: Springer Science and Business Media Deutschland GmbH.

MLA:

Fu, Weilin, et al. "Form Follows Function: Smart Network Design Enables Zero-shot Network Reuse." Proceedings of the German Workshop on Medical Image Computing, 2022, Heidelberg, DEU Ed. Klaus Maier-Hein, Thomas M. Deserno, Heinz Handels, Andreas Maier, Christoph Palm, Thomas Tolxdorff, Springer Science and Business Media Deutschland GmbH, 2022. 121-126.

BibTeX: Download