跟骨骨折占t骨区域所有成人骨折的60％以上。建议确定患者是否患有跟骨骨折可能非常具有挑战性。医生通常会进行X光和CT检查，并对患者进行额外的检查。为了给予最佳治疗，整形外科需要知道跟骨骨折的类型。然而，即使在CT图像的帮助下，它们仍然成为确定骨折类型的问题。 除此之外，机器学习和深度学习已经显着增长，它能够解决许多不同领域的复杂任务。在生物医学领域，它已被用于从医学图像中检测和分类癌组织。在这项研究中，我们将PCANet应用于深度神经网络架构。 此外，由于患有跟骨骨折CT图像的患者数量较少，影响数据集的数量，我们采用数据增强技术来为训练阶段生成新的数据集。我们的观察表明，为训练添加增强数据可以显着提高检测准确性。 最后，我们提出了一个初步研究，内容涉及建立健壮的跟骨骨折检测系统所涉及的可能性，机遇和挑战。我们的实验结果表明，我们提出的方法可以达到高达70％的检测精度。 ;Calcaneus fracture has contributed to more than 60% of all adult fractures on the tarsal bones area. It is suggested that determining if a patient is having a calcaneal fracture can be very challenging. Doctors usually will perform an X-Ray and CT examination and doing additional tests with the patients. In order to give the best treatment, orthopedics need to know the type of the calcaneal fracture. However, it is still becoming a problem for them to determine the fracture type even with the help of CT images. Apart from that, machine learning and deep learning has grown significantly that it is able to solve complex tasks in many different fields. In biomedical fields, it has been used to detect and classify cancer tissues from medical images. In this research, we apply PCANet for the deep neural network architecture. Also, due to the small number of patients having the calcaneal fracture CT images which affect on the number of dataset, we employ a data augmentation technique to generate new dataset for the training phase. Our observation shows that the addition of augmented data for training can increase the detection accuracy significantly. In the end, we present a preliminary study regarding the possibility, opportunities, and challenges which are involved in building a robust calcaneal fracture detection system. Our experimental results show that our proposed approach can reach up to 70% of detection accuracy.
