YOLOv8-AM: YOLOv8 Based on Effective Attention Mechanisms for Pediatric Wrist Fracture Detection (2024-02-14T00:00:00.000000Z)

TL;DR

This work proposes YOLOv8-AM, which incorporates the attention modules into the YOLOv8 architecture, and employs four different attention modules, ResBlock with Convolutional Block Attention Module (ResCBAM), Shuffle Attention, Efficient Channel Attention, and ResBlock with Global Attention Mechanism (ResGAM), to improve the model architecture.

Abstract

Wrist trauma and even fractures occur frequently in daily life, particularly among children who account for a significant proportion of fracture cases. Before performing surgery, surgeons often request patients to undergo X-ray imaging first and prepare for surgery based on the analysis of the radiologists. With the development of neural networks, You Only Look Once (YOLO) series models have been widely used in fracture detection as computer-assisted diagnosis (CAD) tools. Ultralytics presented the latest version of the YOLO models in 2023, which has been employed for detecting fractures across various parts of the body. Attention mechanism is one of the most popular methods to improve the model performance. This work proposes YOLOv8-AM, which incorporates the attention modules into the YOLOv8 architecture. Specifically, we respectively employ four different attention modules, ResBlock with Convolutional Block Attention Module (ResCBAM), Shuffle Attention (SA), Efficient Channel Attention (ECA), and ResBlock with Global Attention Mechanism (ResGAM), to improve the model architecture, and train these models on the GRAZPEDWRI-DX dataset. Experimental results demonstrate that the mean Average Precision at IoU 50 (mAP@50) of one of the variants of YOLOv8-AM model (i.e., YOLOv8+ResCBAM) increased from 63.6% to 65.8%, which achieves the state-of-the-art (SOTA) performance. In addition, the proposed YOLOv8-AM models can detect the single-label category “fracture” with mAP@50 value of 95.7% in pediatric wrist trauma X-ray images. The implementation code for this work is available on GitHub at https://github.com/RuiyangJu/Fracture_Detection_Improved_YOLOv8.

Authors

References82 items

Skeletal Fracture Detection with Deep Learning: A Comprehensive Review

Fracture detection in pediatric wrist trauma X-ray images using YOLOv8 algorithm

Resolution enhancement processing on low quality images using swin transformer based on interval dense connection strategy

Maxillofacial fracture detection and classification in computed tomography images using convolutional neural network-based models

Automatic Rib Fracture Detection and Localization from Frontal and Oblique Chest X-Rays

Designing Network Design Strategies Through Gradient Path Analysis

Fracture Recognition in Paediatric Wrist Radiographs: An Object Detection Approach

An Improved YOLOv5s Algorithm for Object Detection with an Attention Mechanism

YOLOv7: Trainable Bag-of-Freebies Sets New State-of-the-Art for Real-Time Object Detectors

The Evaluation of Bone Fracture Detection of YOLO Series

A pediatric wrist trauma X-ray dataset (GRAZPEDWRI-DX) for machine learning

Avalanche decision schemes to improve pediatric rib fracture detection

Assessment of deep convolutional neural network models for mandibular fracture detection in panoramic radiographs.

Localization and Edge-Based Segmentation of Lumbar Spine Vertebrae to Identify the Deformities Using Deep Learning Models

Global Attention Mechanism: Retain Information to Enhance Channel-Spatial Interactions

SA-Net: Shuffle Attention for Deep Convolutional Neural Networks

Hierarchical fracture classification of proximal femur X-Ray images using a multistage Deep Learning approach.

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

Artificial Intelligence Solutions for Analysis of X-ray Images

Object detection based on an adaptive attention mechanism

Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection

Enhancing Geometric Factors in Model Learning and Inference for Object Detection and Instance Segmentation

Detection and localization of distal radius fractures: Deep learning system versus radiologists.

CSPNet: A New Backbone that can Enhance Learning Capability of CNN

Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression

Using a Dual-Input Convolutional Neural Network for Automated Detection of Pediatric Supracondylar Fracture on Conventional Radiography.

ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks

Asymmetric Non-Local Neural Networks for Semantic Segmentation

Expectation-Maximization Attention Networks for Semantic Segmentation

GCNet: Non-Local Networks Meet Squeeze-Excitation Networks and Beyond

CenterNet: Keypoint Triplets for Object Detection

Artificial intelligence detection of distal radius fractures: a comparison between the convolutional neural network and professional assessments

SRM: A Style-Based Recalibration Module for Convolutional Neural Networks

Selective Kernel Networks

Object Detection Network Based on Feature Fusion and Attention Mechanism

Detection of distal radius fractures trained by a small set of X-ray images and Faster R-CNN

Global Second-Order Pooling Convolutional Networks

CCNet: Criss-Cross Attention for Semantic Segmentation

A2-Nets: Double Attention Networks

Deep neural network improves fracture detection by clinicians

Shortfall of consultant clinical radiologists in the UK.

Dual Attention Network for Scene Segmentation

PSANet: Point-wise Spatial Attention Network for Scene Parsing

OCNet: Object Context Network for Scene Parsing

ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design

BAM: Bottleneck Attention Module

CBAM: Convolutional Block Attention Module

Automated detection and classification of the proximal humerus fracture by using deep learning algorithm

Group Normalization

Path Aggregation Network for Instance Segmentation

Artificial intelligence in fracture detection: transfer learning from deep convolutional neural networks.

Radiologist shortage leaves patient care at risk, warns royal college

Squeeze-and-Excitation Networks

Attention is All you Need

Feature Pyramid Networks for Object Detection

Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization

An overview of gradient descent optimization algorithms

Learning Deep Features for Discriminative Localization

Deep Residual Learning for Image Recognition

You Only Look Once: Unified, Real-Time Object Detection

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Adam: A Method for Stochastic Optimization

Visualizing and Understanding Convolutional Networks

Volume 4

Fractures in children: epidemiology and activity-specific fracture rates.

Most Frequently Missed Fractures in the Emergency Department

The treatment of upper limb fractures in children and adolescents.

A qualitative investigation into the rehabilitation experience of patients following wrist fracture

ImageNet: A large-scale hierarchical image database

HIS/RIS/PACS integration: getting to the gold standard.

Information Engineering

An Improved YOLOv5 for Skull Fracture Detection

Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

“Cbam: Convolutional block attentionmodule,”inProceedingsoftheEuropeanconferenceoncomputer

Looking Within How X Ray Ct Mri Ultrasound And Other Medical Images Are Created And How They Help Physicians Save Lives

Imaging in the Land of 1000 Hills: Rwanda Radiology Country Report

Overlooked extremity fractures in the emergency department.

Increased incidence over the past decade: a population-based study from northern Sweden

Learning Multiple Layers of Features from Tiny Images

‘‘YOLO by ultralytics,’’

the B.S. degree from the Department of Computer Science and

YOLOv8-AM