Automatic classification of canine thoracic radiographs using deep learning

The interpretation of thoracic radiographs is a challenging and error-prone task for veterinarians. Despite recent advancements in machine learning and computer vision, the development of computer-aided diagnostic systems for radiographs remains a challenging and unsolved problem, particularly in the context of veterinary medicine.

In this study, a novel method, based on multi-label deep convolutional neural network (CNN), for the classification of thoracic radiographs in dogs was developed. All the thoracic radiographs of dogs performed between 2010 and 2020 in the institution were retrospectively collected. Radiographs were taken with two different radiograph acquisition systems and were divided into two data sets accordingly. One data set (Data Set 1) was used for training and testing and another data set (Data Set 2) was used to test the generalization ability of the CNNs. Radiographic findings used as non mutually exclusive labels to train the CNNs were: unremarkable, cardiomegaly, alveolar pattern, bronchial pattern, interstitial pattern, mass, pleural effusion, pneumothorax, and megaesophagus.

Two different CNNs, based on ResNet-50 and DenseNet-121 architectures respectively, were developed and tested. The CNN based on ResNet-50 had an Area Under the Receive-Operator Curve (AUC) above 0.8 for all the included radiographic findings except for bronchial and interstitial patterns both on Data Set 1 and Data Set 2. The CNN based on DenseNet-121 had a lower overall performance.

Statistically significant differences in the generalization ability between the two CNNs were evident, with the CNN based on ResNet-50 showing better performance for alveolar pattern, interstitial pattern, megaesophagus, and pneumothorax.

Authors: Tommaso Banzato, Marek Wodzinski, Silvia Burti, Valentina Longhin Osti, Valentina Rossoni, Manfredo Atzori, Alessandro Zotti

Source: https://www.nature.com/

Like0

Dislike0

List
Add

Please enter a comment

Name *

Email address *

Comment *

* Required fields

Information on the processing of your personal data
We inform you that, in compliance with the provisions of current national and European regulations for the Protection of Personal Data and Services of the Information Society and Electronic Commerce, by sending us this form you are expressly giving your consent to Grupo Asís Biomedia , SL, (hereinafter, "ASIS GROUP") so that, as the person in charge, it may process your personal data in order to respond to your request for contact and information by electronic means.

Likewise, when you expressly consent, we will process your personal data to send you specialized information, newsletters, offers and exclusive promotions from GRUPO ASIS and related companies.

For the aforementioned purpose, GRUPO ASIS may transfer your data to other companies linked to GRUPO ASIS or to third party service providers for the management of electronic communications and other security services, even in cases where they are outside of the European Union, provided that they legally guarantee the adequate level of protection required by European regulations.

At any time you can withdraw the consent given and exercise the rights of access, rectification, deletion, portability of your data and limitation or opposition to its treatment by contacting GRUPO ASIS by sending an email to protecciondatos @ grupoasis.com, or by written communication to address at Centro Empresarial El Trovador, 8th floor, office I, Plaza Antonio Beltrán Martínez 1, 50002, Zaragoza (Spain), indicating in either case the Ref. Personal data and the right you exercise, as well as attaching a copy of your ID or replacement identification document.

I have read and accept the treatment of my data according to the informed purpose and according Legal notes and the Privacy Policy
I wish to receive commercial information from GRUPO ASIS and related companies