2022
DOI: 10.1111/vru.13069
|View full text |Cite
|
Sign up to set email alerts
|

Comparison of error rates between four pretrained DenseNet convolutional neural network models and 13 board‐certified veterinary radiologists when evaluating 15 labels of canine thoracic radiographs

Abstract: Convolutional neural networks (CNNs) are commonly used as artificial intelligence (AI) tools for evaluating radiographs, but published studies testing their performance in veterinary patients are currently lacking. The purpose of this retrospective, secondary analysis, diagnostic accuracy study was to compare the error rates of four CNNs to the error rates of 13 veterinary radiologists for evaluating canine thoracic radiographs using an independent gold standard. Radiographs acquired at a referral institution … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
4
1

Citation Types

0
9
0

Year Published

2023
2023
2024
2024

Publication Types

Select...
6
2

Relationship

0
8

Authors

Journals

citations
Cited by 14 publications
(9 citation statements)
references
References 29 publications
0
9
0
Order By: Relevance
“…The DICOM les were initially converted to the MHA format, resampled to 224 x 224 and normalized by a Z-normalization speci c to the ResNet-50 network. The ResNet-50 pre-trained on ImageNet was used, since previous research has indicated that it provides the most accurate results for X-ray classi cation with a limited size datasets [10][11][12][13] . The architecture was then ne-tuned on the aforementioned database with a multi-label setting, as the quality classes were not mutually exclusive.…”
Section: Deep Learningmentioning
confidence: 99%
See 1 more Smart Citation
“…The DICOM les were initially converted to the MHA format, resampled to 224 x 224 and normalized by a Z-normalization speci c to the ResNet-50 network. The ResNet-50 pre-trained on ImageNet was used, since previous research has indicated that it provides the most accurate results for X-ray classi cation with a limited size datasets [10][11][12][13] . The architecture was then ne-tuned on the aforementioned database with a multi-label setting, as the quality classes were not mutually exclusive.…”
Section: Deep Learningmentioning
confidence: 99%
“…Particularly in the last few years, studies on the applications of AI in classifying canine meningiomas from MR 7 , in distinguishing between meningiomas and gliomas in MR 8 , and in detecting spinal cord diseases from MR images 9 have been published. To date, the most proli c sector of investigation in this eld is the application of AI for the automatic detection of lesions from thoracic x-rays with an increasing number of publications on this topic [10][11][12][13] .…”
Section: Introductionmentioning
confidence: 99%
“…On the other hand, the reduced availability of trained veterinary radiologists struggles to meet the demand for accurate interpretation of diagnostic images 3 . In such a scenario, the role of AI-assisted tools is gaining popularity also in veterinary medicine, with an increasing number of publications on this topic [4][5][6][7][8][9][10][11] . The ever-increasing popularity of deep learning models due to their high performance in various domains has led to an increased interest in such methods in the field of computer-aided diagnosis.…”
Section: Introductionmentioning
confidence: 99%
“…Multiple recent studies have demonstrated artificial intelligence's (AI) ability to evaluate a range of canine thoracic radiographic features such as cardiac size, pulmonary patterns, and pleural changes. [1][2][3][4][5][6][7][8][9][10][11] Although the practicality of AI in veterinary diagnostic imaging is not yet fully established, several commercial AI platforms are currently available to the practitioner. Further independent scrutiny of these tools is necessary to validate these products, in large part so that practitioners can build confidence in their application.…”
Section: Introductionmentioning
confidence: 99%
“…Multiple recent studies have demonstrated artificial intelligence's (AI) ability to evaluate a range of canine thoracic radiographic features such as cardiac size, pulmonary patterns, and pleural changes 1–11 . Although the practicality of AI in veterinary diagnostic imaging is not yet fully established, several commercial AI platforms are currently available to the practitioner.…”
Section: Introductionmentioning
confidence: 99%