Adipose tissue estimation in green turtles (Chelonia mydas) using deep learning, diagnostic imaging and bioelectrical impedance analysis

Conference Contribution ResearchOnline@JCU
Kophamel, Sara;Ward, Leigh;Konovalov, Dmitry;Ariel, Ellen;Mendez, Diana;Bell, Ian;Cassidy, Nathan;Balastegui Martinez, Maria T.;Munns, Suzanne
Abstract

Sea turtle populations are endangered worldwide due to anthropogenic impacts. The loss of fat reserves (adipose tissue) in individual turtles is linked to environmental stress and diseases, and is considered an indicator of health status. Bioelectrical Impedance Analysis (BIA) has emerged as a portable technology that accurately estimates adipose tissue in fishes and humans, showing potential for field-based studies. We determined device reliability, quantified the impacts of potential confounders, and calibrated BIA measurements using computed tomography (CT) scans in green turtles (Chelonia mydas). The effects of temperature (20-30ᵒC), digestive state, and dry-docking time on impedance parameters (R0, R50 and Rinf) were measured on 35 juvenile turtles using multi-frequency BIA. Impedance measurements were used to calculate adipose tissue mass. Adipose tissue calculations (mass) were compared to those determined from CT scans on 50 additional turtles. A deep learning model, based on convolutional neural networks, was trained for CT image processing. Results showed that BIA measurements were reproducible. The impedance parameters did not significantly alter with digestive state, but were inversely correlated with body temperatures. A temperature correction algorithm was therefore developed to enable field BIA use between 20-30ºC. Calculations of adipose tissue correlated well between BIA and CT techniques. We present the first BIA calibration study in sea turtles, and our standardised protocol provides an accurate estimation of adipose tissue in green turtles. This valuable information on individual health status can inform conservation management decisions on a population level.

Journal

N/A

Publication Name

CUENCA 2021: 69th Wildlife Disease Association (WDA)/14th European Wildlife Disease Association (EWDA) Joint Virtual Conference

Volume

N/A

ISBN/ISSN

N/A

Edition

N/A

Issue

N/A

Pages Count

1

Location

Online

Publisher

Wildlife Disease Association (WDA) and European Wildlife Disease Association (EWDA)

Publisher Url

N/A

Publisher Location

Online

Publish Date

N/A

Url

N/A

Date

N/A

EISSN

N/A

DOI

N/A