Introduction and Background: Ankle injury is one of the most frequent presenting injuries to the emergency room and ankle fractures are common fracture in the lower limbs injuries that may require operative treatment ...Introduction and Background: Ankle injury is one of the most frequent presenting injuries to the emergency room and ankle fractures are common fracture in the lower limbs injuries that may require operative treatment with variable outcomes. Materials and Methods: Sixty-three patients were included in my retrospective study, and all with a displaced fracture of the ankle caused by high energy trauma were treated by open reduction and rigid internal fixation. Results: After follow-up at six weeks and twelve weeks, the results were satisfactory in fifty-five percent out of all the sixty-three patients. Conclusions and Recommendations: Ankle fractures occur mainly in young males of the age group between 26 and 35 years, mostly caused by fall down and motor vehicle accident.展开更多
目的以股骨转子间骨折数据集为研究对象,探究医用X线图像输入分辨率的改变对YOLO系列目标检测网络识别效果的影响。方法收集陆军特色医学中心2017-2022年间患转子间骨折患者的X线正位数据,按排除标准排除后共保留426例患者,847幅图像。...目的以股骨转子间骨折数据集为研究对象,探究医用X线图像输入分辨率的改变对YOLO系列目标检测网络识别效果的影响。方法收集陆军特色医学中心2017-2022年间患转子间骨折患者的X线正位数据,按排除标准排除后共保留426例患者,847幅图像。结合2018版Arbeitsgemeinschaft für Osteosynthesefragen/Orthopaedic Trauma Association(AO/OTA)分型指南和临床实际发病率,将转子间骨折重新整理划分为A1.2、A1.3、A2.2、A2.3、A3等5个亚组,并按8∶1∶1的比例划分为训练集(678幅)、验证集(84幅)和测试集(85幅),每次试验中保持严格一致。将YOLOX-Swin-Transformer、YOLOX、YOLOv5、YOLOv4目标识别网络图像输入大小分别设置为8种常用的分辨率,使用从头训练和迁移学习方法分别训练转子间骨折数据集,记录训练耗时,利用测试集进行模型测试并记录评价指标。采用SPSS20.0统计软件回归分析对训练耗时和均值平均精度(mean average precision,mAP)做曲线拟合,利用频数统计功能对各输入分辨率下评价指标评级为优的频数进行统计,以此判断最优区间。结果图像输入分辨率与各网络训练耗时呈正相关,所有P<0.05,线性回归分析有统计学意义。图像输入分辨率与网络mAP均值二次曲线拟合R^(2)=0.834,P=0.011,说明曲线拟合优度较好,回归分析有统计学意义。当图像输入分辨率在480×480、576×576、640×640范围内时,数据评级为优出现的频率最高(占42.86%)。结论训练耗时随分辨率的增加而增加。在不改变网络结构的前提下采用YOLO系列网络做医学图像识别的下游任务时,要取得最佳的识别效果,图像输入分辨率应控制在480×480、576×576、640×640范围之内。展开更多
文摘Introduction and Background: Ankle injury is one of the most frequent presenting injuries to the emergency room and ankle fractures are common fracture in the lower limbs injuries that may require operative treatment with variable outcomes. Materials and Methods: Sixty-three patients were included in my retrospective study, and all with a displaced fracture of the ankle caused by high energy trauma were treated by open reduction and rigid internal fixation. Results: After follow-up at six weeks and twelve weeks, the results were satisfactory in fifty-five percent out of all the sixty-three patients. Conclusions and Recommendations: Ankle fractures occur mainly in young males of the age group between 26 and 35 years, mostly caused by fall down and motor vehicle accident.
文摘目的以股骨转子间骨折数据集为研究对象,探究医用X线图像输入分辨率的改变对YOLO系列目标检测网络识别效果的影响。方法收集陆军特色医学中心2017-2022年间患转子间骨折患者的X线正位数据,按排除标准排除后共保留426例患者,847幅图像。结合2018版Arbeitsgemeinschaft für Osteosynthesefragen/Orthopaedic Trauma Association(AO/OTA)分型指南和临床实际发病率,将转子间骨折重新整理划分为A1.2、A1.3、A2.2、A2.3、A3等5个亚组,并按8∶1∶1的比例划分为训练集(678幅)、验证集(84幅)和测试集(85幅),每次试验中保持严格一致。将YOLOX-Swin-Transformer、YOLOX、YOLOv5、YOLOv4目标识别网络图像输入大小分别设置为8种常用的分辨率,使用从头训练和迁移学习方法分别训练转子间骨折数据集,记录训练耗时,利用测试集进行模型测试并记录评价指标。采用SPSS20.0统计软件回归分析对训练耗时和均值平均精度(mean average precision,mAP)做曲线拟合,利用频数统计功能对各输入分辨率下评价指标评级为优的频数进行统计,以此判断最优区间。结果图像输入分辨率与各网络训练耗时呈正相关,所有P<0.05,线性回归分析有统计学意义。图像输入分辨率与网络mAP均值二次曲线拟合R^(2)=0.834,P=0.011,说明曲线拟合优度较好,回归分析有统计学意义。当图像输入分辨率在480×480、576×576、640×640范围内时,数据评级为优出现的频率最高(占42.86%)。结论训练耗时随分辨率的增加而增加。在不改变网络结构的前提下采用YOLO系列网络做医学图像识别的下游任务时,要取得最佳的识别效果,图像输入分辨率应控制在480×480、576×576、640×640范围之内。
