Deep Scale-Variant Network for Femur Trochanteric Fracture Classification with HP Loss

Abstract

Achieving automatic classification of femur trochanteric fracture from the edge computing device is of great importance and value for remote diagnosis and treatment. Nevertheless, designing a highly accurate classification model on 31A1/31A2/31A3 fractures from the X-ray is still limited due to the failure of capturing the scale-variant and contextual information. As a result, this paper proposes a deep scale-variant (DSV) network with a hybrid and progressive (HP) loss function to aggregate more influential representations of the fracture regions. More specifically, the DSV network is based on the ResNet and integrated with the designed scale-variant (SV) layer and HP loss, where the SV layer aims to enhance the representation ability to extract the scale-variant features, and HP loss is intended to force the network to condense more contextual clues. Furthermore, to evaluate the effect of the proposed DSV network, we carry out a series of experiments on the real X-ray images for comparison and evaluation, and the experimental results demonstrate that the proposed DSV network could outperform other classification methods on this classification task.

Skeletal Fracture Detection with Deep Learning: A Comprehensive Review

“…[46] designed the DDA layer (dense dilated attention module) to give scope to the advantages of attention mechanisms. In classification tasks, most of the studies [41,44,46,47] utilized state-of-the-art deep learning algorithms, which have a high accuracy, sensitivity, and specificity of near or more than 90% and have a high AUC of more than 0.9. These experimental results show that deep learning algorithms can improve performance and accelerate application in clinical practice.…”

“…The CNN also works on the classification task, with a bit of tuning. The variations in CNNs applied for classification include DenseNet [43], Dense Dilated attentive network [46], and ResNet [41,47]. Their secret key technology is utilizing the convolutional filter.…”

“…Their secret key technology is utilizing the convolutional filter. Therefore, the classification task has multiple output values and the Softmax activation function can normalize these output values; these models always select the Softmax activation function rather than Sigmoid [41][42][43]46,47]. Compared with the recognition task, bone fractures in the classification task have more categories.…”

“…[42] exploits another interactive pipeline to ask the user to re-orient the femur bones to obtain better features. Additionally, Kang added a scaled variant layer and a hybrid and progressive loss function to ResNet to better classify fractures in 2020 [47]. In conclusion, CNNs for classification often utilize the Softmax function due to varied outputs, and as bone fracture classification demands richer features, innovations like interactive pipelines and enhanced layers have been introduced for superior results.…”

Retracted: Deep Scale‐Variant Network for Femur Trochanteric Fracture Classification with HP Loss

“…This article has been retracted by Hindawi, as publisher, following an investigation undertaken by the publisher [ 1 ]. This investigation has uncovered evidence of systematic manipulation of the publication and peer-review process.…”