帕金森定律_通过图像分析对帕金森病进行分类

应用计算机视觉 (Applied Computer Vision)

简介 (Introduction)

Parkinson’s disease is often associated with movement disorder symptoms such as tremors and rigidity. These can have a noticeable effect on the handwriting and sketching (drawing)of a person suffering from early stages of the disease [1]. Micrographia, are abnormally small undulations in a persons handwriting, however, have claimed to be difficult to interpret due to the variability in one’s developed handwriting, language, proficiency and education etc [1]. As such, a study conducted in 2017 aimed to improve the diagnosis through a standardized analysis using spirals and waves. In this series of posts, we will analyze the raw images collected in that study and see if we can create a classifier for a patient having Parkinson’s, and draw some conclusions along the way. The data we will be using is hosted on Kaggle [2] with special thanks to Kevin Mader for sharing the dataset upload.

P arkinson病常与运动障碍症状，如震颤和刚性有关。 这些可以对患有该疾病早期阶段的人的笔迹和草图(绘画)产生显着影响[1]。 显微照相术是人类笔迹中异常小的波动，然而，由于人的笔迹，语言，熟练程度和受教育程度等方面的差异，据称难以解释。 因此，2017年进行的一项研究旨在通过使用螺旋和波浪的标准化分析来改善诊断。 在这一系列文章中，我们将分析该研究收集的原始图像，看看是否可以为患有帕金森氏症的患者创建分类器，并一路得出结论。 我们将使用的数据托管在Kaggle [2]上，特别感谢Kevin Mader分享了数据集上传。

In this part 1, we will be conducting some exploratory data analysis and pre-processing the images to create some features that will hopefully be helpful in classification. I am choosing to NOT use a convolutional neural network (CNN) to simply classify the images as this will be black box — without any metric into the underlying differences between the curves/sketches. Instead, we are not simply performing a task of classifying but trying to use image processing to understand and quantify the differences. In a subsequent post, I will compare with a CNN.

在第1部分中，我们将进行一些探索性数据分析并对图像进行预处理，以创建一些有望对分类有所帮助的功能。 我选择不使用卷积神经网络(CNN)来对图像进行简单分类，因为这将是黑盒-曲线/草图之间的潜在差异没有任何度量标准。 相反，我们不仅仅是执行分类任务，而是尝试使用图像处理来理解和量化差异。 在后续文章中，我将与CNN进行比较。

Before we begin, disclaimer that this is not meant to be any kind of medical study or test. Please refer to the original paper for details on the actual experiment, which I was not a part of.Zham P, Kumar DK, Dabnichki P, Poosapadi Arjunan S, Raghav S. Distinguishing Different Stages of Parkinson’s Disease Using Composite Index of Speed and Pen-Pressure of Sketching a Spiral. Front Neurol. 2017;8:435. Published 2017 Sep 6. doi:10.3389/fneur.2017.00435

在我们开始之前，请声明这并不意味着要进行任何医学研究或测试。 请参考原始文件的详细信息，实际的实验，我是不是部分of.Zham P，库马尔DK，Dabnichki P，Poosapadi阿晶南S， 帕金森氏病的使用速度和笔的综合指数拉哈夫S. 区分不同阶段-绘制螺旋线的压力 。 前神经元。 2017; 8：435。 2017年9月6日发布。doi：10.3389 / fneur.2017.00435

探索性数据分析 (Exploratory Data Analysis)

First, let us take a look at the images, perform some basic segmentation and start poking around with some potential features of interest. We will be using pandas throughout to store the images and information. For those of you questioning whether you will read this section here is what we will get into: - Thresholding and cleaning- Thi