基于深度学习的腺体病理图像分割

Glandular Pathological Image Segmentation Based on Deep Learning

腺体病变引起的疾病如结肠腺癌、乳腺癌等的发病率逐年增高,病理检查是临床诊断的“金标准”,从病理图像中准确分割病灶范围对疾病的诊疗至关重要,然而这是一项费时费力的工作,同时与病理医生的水平与经验有关。近年来,计算机辅助诊断系统和深度学习(Deep learning)在医学图像处理领域快速发展并得到广泛应用,为进一步减轻医生的工作负担,采用经典神经网络对腺体病理图像进行区域分割,并使其能够适用于更加广泛的腺体分割,在腺体病理图像中取得较好的分割效果,为辅助早期诊断及减小误诊概率提供可能。

The incidence of diseases caused by glandular lesions such as colon adenocarcinoma and breast cancer is increasing year by year.Pathological examination is the'gold standard'for clinical diagnosis.Accurate segmentation of lesion range from pathological images is essential for the diagnosis and treatment of diseases.However,this is a time-consuming and laborious work,which is related to the level and experience of pathologists.In recent years,computer aided diagnosis system and deep learning(Deep learning)has been developed rapidly and widely used in the field of medical image processing.In order to further reduce the heavy work of doctors,the classical neural network is used to segment the gland pathological images,and it can be applied to more extensive gland segmentation.Effective segmentation results are achieved in the gland pathological images,which provide the possibility for early diagnosis and reducing the probability of misdiagnosis.