An Automatic Colorectal Polyps Detection Approach for Ct Colonography
An Automatic Colorectal Polyps Detection Approach for Ct Colonography
复制标题
DOI：
10.1109/icip49359.2023.10221981
复制
发表时间：
2023-10
期刊：
影响因子：
--
通讯作者：
Mohamed Yousuf;Islam Alkabbany;Asem A. Ali;Salwa Elshazley;Albert Seow;Gerald W. Dryden;Aly A. Farag
中科院分区：
文献类型：
--
作者：
Mohamed Yousuf;Islam Alkabbany;Asem A. Ali;Salwa Elshazley;Albert Seow;Gerald W. Dryden;Aly A. Farag
在这项工作中，我们提出了一种自动结直肠息肉检测方法，包括两个级联阶段。在第一阶段中，训练CNN模型以检测轴向CT切片中的息肉。CNN模型由分割的结肠壁CT切片而不是原始CT切片馈送。使用分割图像作为CNN模型的输入极大地改善了检测和定位结果，例如，mAP增加了36%。为了减少由检测器生成的假阳性，部署第二阶段分类器以利用CT扫描的不同视图而不是仅轴向视图。因此，使用轴向视图的2D图像来训练分类器，即，由探测器生成的候选息肉，以及它们相应的矢状和冠状视图的2D图像。该方法的实验结果由3名放射科医生验证，该方法在分类阶段后成功识别出息肉，AUC为98.6%。
In this work, we propose an automatic colorectal polyps detection approach that consists of two cascade stages. In the first stage, a CNN model is trained to detect polyps in axial CT slices, The CNN model has been fed by the segmented colon wall CT slices instead of the original CT slices. Using the segmented images as an input to the CNN model has drastically improved the detection and localization results, e.g., the mAP is increased by 36%. To reduce the false positives generated by the detector, the second stage classifier is deployed to exploit the different views of the CT scans instead of the axial view only. So, the classifier is trained using the 2D images of axial views, i.e., the candidate polyps generated by the detector, as well as their corresponding 2D images of sagittal and coronal views. The experimental results of this approach were validated by 3 radiologists and the approach successfully identified polyps after the classification stage with an AUC ∼ 98.6%.