george-chou/MIC

Medical image computing (MIC) is an interdisciplinary field, there are three essential medical image analysis techniques: medical image enhancement (MIE) and medical image segmentation (MIS).|医学图像计算 (MIC) 是计算机科学、信息工程、电气工程、物理学、数学和医学交叉的交叉学科领域。在 MIC 领域内，存在三种基本的医学图像分析技术：医学图像增强 (MIE) 和医学图像分割（MIS）。

MIC

Medical Image Computing Module Development

Medical Image Enhancement (MIE)

The source code of MIE is in MedicalImageEnhancement.py.

Table 1 compares the three filters, including smoothing, sharpening, edge detection in the aspects of the kernel, experiment demo, and time cost of running. The way of achieving a filter is to use the filter to perform convolution operations on 3D images.

Table 1: Comparison of different filters

Medical Image Segmentation (MIS)

The source code of MIS is in MedicalImageSegmentation.py.

Figure 1 shows the demonstration of 3D segmentation results achieved using three multiple viewing angles.

Figure 1: 3D segmentation result of the tumor

Table 2 shows the demonstration of the experiments on different global and local parameter combinations.

Table 2: Part experiments on different global and local parameters

Based on experiment results, the best global and local parameters are among experiments with ID from 2 to 3 considering whether the classification is true or false, positive or negative. The best way to find them is to define an indicator of the best result, then define a distance between the result indicator in current parameters and the best outcome, and finally search for the settings of minimum distance by deep learning. In that way, we do not need to search for the best parameters manually.