用于医疗诊断的一种新型轮廓提取算法

A New Contour Extraction Algorithm for Medical Diagnosis

在辅助医学诊疗方面,机器学习方法由于操作简单、识别准确率高而被广泛应用在医学细胞图像识别中。细胞图像处理的关键是细胞边缘的提取,而细胞图像中存在重叠、细小颗粒杂质的复杂现象,因此选择一个合适的轮廓提取算法十分必要。提出了一种新型的轮廓提取算法,该轮廓提取方法可以获得细胞的精确轮廓,从而更有利于细胞的边缘提取。并在公开的数据集中进行了仿真比较,通过观察正确率、灵敏度、特异度、阳性预测率和阴性预测率等5个评价指标,发现该算法得到了较高的准确率,从而有效降低了人工识别的难度,对于人工智能识别医疗细胞具有较为重要的推广意义。

In the field of assisted medical diagnosis and treatment,the machine learning method is widely used in medical cell images due to its simple operation and high recognition accuracy.The key to cell image processing is the extraction of cell edges,and there are complex phenomena of overlapping and small particle impurities in cell images.Therefore,it is necessary to choose a suitable contour extraction algorithm.This paper proposes a new contour extraction algorithm,through the contour extraction method,the precise contour of the cells can be obtained,thereby facilitating the edge extraction of the cells.In addition,this paper compares the algorithm in an unified open-source dataset.By observing the evaluation indexes of each algorithm,it can be found that the accuracy of the proposed algorithm is the highest,which effectively reduces the difficulty of manual recognition.It has an important promotion significance for artificial intelligence to identify medical cells.