English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Deep learning has emerged as a powerful artificial intelligence tool tointerpret medical images for a growing variety of applications. However, thepaucity of medical imaging data with high-quality annotations that is necessaryfor training such methods ultimately limits their performance. Medical data ischallenging to acquire due to privacy issues, shortage of experts available forannotation, limited representation of rare conditions and cost. This problemhas previously been addressed by using synthetically generated data. However,networks trained on synthetic data often fail to generalize to real data.Cinematic rendering simulates the propagation and interaction of light passingthrough tissue models reconstructed from CT data, enabling the generation ofphotorealistic images. In this paper, we present one of the first applicationsof cinematic rendering in deep learning, in which we propose to fine-tunesynthetic data-driven networks using cinematically rendered CT data for thetask of monocular depth estimation in endoscopy. Our experiments demonstratethat: (a) Convolutional Neural Networks (CNNs) trained on synthetic data andfine-tuned on photorealistic cinematically rendered data adapt better to realmedical images and demonstrate more robust performance when compared tonetworks with no fine-tuning, (b) these fine-tuned networks require lesstraining data to converge to an optimal solution, and (c) fine-tuning with datafrom a variety of photorealistic rendering conditions of the same sceneprevents the network from learning patient-specific information and aids ingeneralizability of the model. Our empirical evaluation demonstrates thatnetworks fine-tuned with cinematically rendered data predict depth with 56.87%less error for rendered endoscopy images and 27.49% less error for real porcinecolon endoscopy images.

Deep Learning with Cinematic Rendering: Fine-Tuning Deep Neural Networks Using Photorealistic Medical Images