基于活动轮廓模型的彩色白细胞图像自动分割方法研究

为解决血液白细胞显微图像自动识别中的图像分割问题,提出了一种基于活动轮廓的彩色白细胞图像自动分割方法,首先在Hue,Saturation,Intensitv（HSI）彩色空间中运用聚类分割得到细胞核,从而得到细胞所在的位置,然后用流域算法得到细胞大致的轮廓,最后将此轮廓作为初始轮廓,用梯度矢量流（GVF）外力及来自全局信息的区域力驱动,结合彩色信息,使得轮廓收敛于真实的细胞边界。实验结果表明,此方法能精确、有效地分割出单个以及部分重叠白细胞区域。     

基于卷积神经网络的白带中白细胞的自动检测

作为妇科常规的检查项目,白带常规检查有着相当广泛的应用,在临床检验中具有非常重要的地位。鉴于白带中白细胞在临床医学中的重要意义和现行检测方法的诸多不足,对白细胞的自动检测技术进行研究。依托于白带自动检测仪,与本地医院进行协作,采集得到白带显微图像。对滤波增强后的图像进行分割,建立样本库,基于卷积神经网络完成特征提取和分类,最终采用交叉验证证实该方法的有效性。在白细胞的自动检测中,对于由2万个样本组成的数据集,运用该方法实现95%的敏感性、84%的特异性、89.5%的准确率,达到医学临床检验的要求。将数字图像处理技术和卷积神经网络综合应用于医学显微图像中白细胞的检测,解决特征表达的关键问题,验证白细胞自动检测的可行性,实现检测质量和检测效率的提升。     

A region-based segmentation of tumour from brain CT images using nonlinear support vector machine classifier

The proposed system provides new textural information for segmenting tumours, efficiently and accurately and with less computational time, from benign and malignant tumour images, especially in smaller dimensions of tumour regions of computed tomography (CT) images. Region-based segmentation of tumour from brain CT image data is an important but time-consuming task performed manually by medical experts. The objective of this work is to segment brain tumour from CT images using combined grey and texture features with new edge features and nonlinear support vector machine (SVM) classifier. The selected optimal features are used to model and train the nonlinear SVM classifier to segment the tumour from computed tomography images and the segmentation accuracies are evaluated for each slice of the tumour image. The method is applied on real data of 80 benign, malignant tumour images. The results are compared with the radiologist labelled ground truth. Quantitative analysis between ground truth and the segmented tumour is presented in terms of segmentation accuracy and the overlap similarity measure dice metric. From the analysis and performance measures such as segmentation accuracy and dice metric, it is inferred that better segmentation accuracy and higher dice metric are achieved with the normalized cut segmentation method than with the fuzzy c-means clustering method.     

A region-based segmentation of tumour from brain CT images using nonlinear support vector machine classifier

The proposed system provides new textural information for segmenting tumours, efficiently and accurately and with less computational time, from benign and malignant tumour images, especially in smaller dimensions of tumour regions of computed tomography (CT) images. Region-based segmentation of tumour from brain CT image data is an important but time-consuming task performed manually by medical experts. The objective of this work is to segment brain tumour from CT images using combined grey and texture features with new edge features and nonlinear support vector machine (SVM) classifier. The selected optimal features are used to model and train the nonlinear SVM classifier to segment the tumour from computed tomography images and the segmentation accuracies are evaluated for each slice of the tumour image. The method is applied on real data of 80 benign, malignant tumour images. The results are compared with the radiologist labelled ground truth. Quantitative analysis between ground truth and the segmented tumour is presented in terms of segmentation accuracy and the overlap similarity measure dice metric. From the analysis and performance measures such as segmentation accuracy and dice metric, it is inferred that better segmentation accuracy and higher dice metric are achieved with the normalized cut segmentation method than with the fuzzy c-means clustering method.     

基于改进LBP特征的白细胞识别

采用先进的计算机图像处理与分析技术完成白细胞分类计数是辅助诊断血液疾病的重要方法。各种白细胞间的纹理差异较大，纹理是区分白细胞的重要特征之一。局部二进制模式（local binary pattern，LBP）是一种有效的纹理描述算子。本研究提出了一种提取细胞的改进LBP特征用于白细胞分类识别的算法。首先，用小波变换对图像进行分解并重构，获得四幅不同频率的分量图，对其中的低频信号采用离散余弦变换。此后，用可变大小的子窗口对变换后的图像扫描，并根据不同区域赋以权值，获取改进的LBP特征，构成加权直方图。这种特征既能反映细胞局部特征又能反映整体特征。根据测试样本和模板的LBP特征直方图之间的马氏距离构建分类器。根据这种改进LBP特征有效地实现了白细胞的5种分类，得到了令人满意的分类正确率。实验结果表明：本研究提出的算法能有效地区分不同类的白细胞；与其他一些算法相比；提高了分类的精确度。     

A New Blood Vessel Extraction Technique Using Edge Enhancement and Object Classification

Diabetic retinopathy (DR) is increasing progressively pushing the demand of automatic extraction and classification of severity of diseases. Blood vessel extraction from the fundus image is a vital and challenging task. Therefore, this paper presents a new, computationally simple, and automatic method to extract the retinal blood vessel. The proposed method comprises several basic image processing techniques, namely edge enhancement by standard template, noise removal, thresholding, morphological operation, and object classification. The proposed method has been tested on a set of retinal images. The retinal images were collected from the DRIVE database and we have employed robust performance analysis to evaluate the accuracy. The results obtained from this study reveal that the proposed method offers an average accuracy of about 97 %, sensitivity of 99 %, specificity of 86 %, and predictive value of 98 %, which is superior to various well-known techniques.     

Leukocyte concentration and the yields of metaphase cells in cultured peripheral blood of leukemic patients with hyperleukocytosis

The optimal concentration of leukocytes for maximizing the yields of metaphase cells was examined using PHA-stimulated and unstimulated peripheral blood cells of various leukemic patients with hyperleukocytosis, including 5 acute nonlymphocytic leukemias, 4 Ph-positive chronic myelocytic leukemias, one acute lymphoblastic leukemia, and one malignant lymphoma at a leukemic phase. The yields of metaphases reached maximum at leukocyte concentrations of either 1 × 10⁶ or 2 × 10⁶/ml in both stimulated and unstimulated blood cultures of each patient, except for one patient with malignant lymphoma who exhibited no metaphases at any cell concentration level in the unstimulated culture. Metaphase yields in the stimulated and unstimulated cultures correlated with neither the percentages of blast cells and lymphocytes in the peripheral blood of the leukemia patients nor with the type of leukemia studied.     

Segmentation of nerve fibers using multi-level gradient watershed and fuzzy systems

ObjectiveThis paper presents an algorithm based on multi-level watershed segmentation combined with three fuzzy systems to segment a large number of myelinated nerve fibers in microscope images. The method can estimate various geometrical parameters of myelinated nerve fibers in peripheral nerves. It is expected to be a promising tool for the quantitative assessment of myelinated nerve fibers in related research.Materials and methodsA novel multi-level watershed scheme iteratively detects pre-candidate nerve fibers. At each immersion level, watershed segmentation extracts the initial axon locations and obtains meaningful myelinated nerve fiber features. Thereafter, according to a priori characteristics of the myelinated nerve fibers, fuzzy rules reject unlikely pre-candidates and collect a set of candidates. Initial candidate boundaries are then refined by a fuzzy active contour model, which flexibly deforms contours according to the observed features of each nerve fiber. A final scan with a different set of fuzzy rules based on the a priori properties of the myelinated nerve fibers removes false detections. A particle swarm optimization method is employed to efficiently train the large number of parameters in the proposed fuzzy systems.ResultsThe proposed method can automatically segment the transverse cross-sections of nerve fibers obtained from optical microscope images. Although the microscope image is usually noisy with weak or variable levels of contrast, the proposed system can handle images with a large number of myelinated nerve fibers and achieve a high fiber detection ratio. As compared to manual segmentation by experts, the proposed system achieved an average accuracy of 91% across different data sets.ConclusionWe developed an image segmentation system that automatically handles myelinated nerve fibers in microscope images. Experimental results showed the efficacy of this system and its superiority to other nerve fiber segmentation approaches. Moreover, the proposed method can be extended to other applications of automatic segmentation of microscopic images.     

Automated recognition of urinary microscopic solid particles

Urine analysis reveals the presence of many problems and diseases in the human body. Manual microscopic urine analysis is time-consuming, subjective to human observation and causes mistakes. Computer aided automatic microscopic analysis can help to overcome these problems. This paper introduces a comprehensive approach for automating procedures for detecting and recognition of microscopic urine particles. Samples of red blood cells (RBC), white blood cells (WBC), calcium oxalate, triple phosphate and other undefined images were used in experiments. Image processing functions and segmentation were applied, shape and textural features were extracted and five classifiers were tested to get the best results. Repeated experiments were done for adjusting factors to produce the best evaluation results. A good performance was achieved compared with many related works.     

Robust feature-based registration using a Gaussian-weighted distance map and brain feature points for brain PET/CT images

Feature-based registration is an effective technique for clinical use, because it can greatly reduce computational costs. However, this technique, which estimates the transformation by using feature points extracted from two images may cause misalignments, particularly in brain PET and CT images that have low correspondence rates between features due to differences in image characteristics. To cope with this limitation, we propose a robust feature-based registration technique using a Gaussian-weighted distance map (GWDM) that finds the best alignment of feature points even when features of two images are mismatched. A GWDM is generated by propagating the value of the Gaussian-weighted mask from feature points of CT images and leads the feature points of PET images to be aligned on an optimal location even though there is a localization error between feature points extracted from PET and CT images. Feature points are extracted from two images by our automatic brain segmentation method. In our experiments, simulated and clinical data sets were used to compare our method with conventional methods such as normalized mutual information (NMI)-based registration and chamfer matching in accuracy, robustness, and computational time. Experimental results showed that our method aligned the images robustly even in cases where conventional methods failed to find optimal locations. In addition, the accuracy of our method was comparable to that of the NMI-based registration method.     

基于Local Jet变换空间纹理特征的肺结节分类方法

为了在纹理特征下改善肺结节良、恶性的模式识别,提出一种基于local jet变换空间的纹理特征提取方法。首先利用二维高斯函数的前三阶偏微分函数将结节原图像变换到local jet纹理图像空间,然后利用纹理描述子在该空间提取特征参数。以灰度值的前四阶矩和基于灰度共生矩阵的特征参数作为纹理描述子,分别提取结节原图像和变换后纹理图像的特征参数,以BP神经网络作为分类器,对同一纹理描述子下的2个不同图像空间的经核主成分分析优化后的特征参数集进行结节良、恶性分类。以157个肺结节(51个良性,106个恶性)作为实验数据进行对比实验,结果显示:两种纹理描述子基于local jet变换空间提取的特征参数分别获得82.69%和86.54%的分类正确率,较原图像空间提高6%～8%,同时AUC值提高约10%。实验结果表明,基于local jet变换空间提取的纹理特征可以有效地改善肺结节良、恶性的模式识别。     

Computer-assisted lip diagnosis on traditional Chinese medicine using multi-class support vector machines

ABSTRACT: BACKGROUND: In Traditional Chinese Medicine (TCM), the lip diagnosis is an important diagnostic method which has a long history and is applied widely. The lip color of a person is considered as a symptom to reflect the physical conditions of organs in the body. However, the traditional diagnostic approach is mainly based on observation by doctor's nude eyes, which is non-quantitative and subjective. The non-quantitative approach largely depends on the doctor's experience and influences accurate the diagnosis and treatment in TCM. Developing new quantification methods to identify the exact syndrome based on the lip diagnosis of TCM becomes urgent and important. In this paper, we design a computer-assisted classification model to provide an automatic and quantitative approach for the diagnosis of TCM based on the lip images. METHODS: A computer-assisted classification method is designed and applied for syndrome diagnosis based on the lip images. Our purpose is to classify the lip images into four groups: deep-red, red, purple and pale. The proposed scheme consists of four steps including the lip image preprocessing, image feature extraction, feature selection and classification. The extracted 84 features contain the lip color space component, texture and moment features. Feature subset selection is performed by using SVM-RFE (Support Vector Machine with recursive feature elimination), mRMR (minimum Redundancy Maximum Relevance) and IG (information gain). Classification model is constructed based on the collected lip image features using multi-class SVM and Weighted multi-class SVM (WSVM). In addition, we compare SVM with k-nearest neighbor (kNN) algorithm, Multiple Asymmetric Partial Least Squares Classifier (MAPLSC) and Naive Bayes for the diagnosis performance comparison. All displayed faces image have obtained consent from the participants. RESULTS: A total of 257 lip images are collected for the modeling of lip diagnosis in TCM. The feature selection method SVM-RFE selects 9 important features which are composed of 5 color component features, 3 texture features and 1 moment feature. SVM, MAPLSC, Naive Bayes, kNN showed better classification results based on the 9 selected features than the results obtained from all the 84 features. The total classification accuracy of the five methods is 84%, 81%, 79% and 81%, 77%, respectively. So SVM achieves the best classification accuracy. The classification accuracy of SVM is 81%, 71%, 89% and 86% on Deep-red, Pale Purple, Red and lip image models, respectively. While with the feature selection algorithm mRMR and IG, the total classification accuracy of WSVM achieves the best classification accuracy. Therefore, the results show that the system can achieve best classification accuracy combined with SVM classifiers and SVM-REF feature selection algorithm. CONCLUSIONS: A diagnostic system is proposed, which firstly segments the lip from the original facial image based on the Chan-Vese level set model and Otsu method, then extracts three kinds of features (color space features, Haralick co-occurrence features and Zernike moment features) on the lip image. Meanwhile, SVM-REF is adopted to select the optimal features. Finally, SVM is applied to classify the four classes. Besides, we also compare different feature selection algorithms and classifiers to verify our system. So the developed automatic and quantitative diagnosis system of TCM is effective to distinguish four lip image classes: Deep-red, Purple, Red and Pale. This study puts forward a new method and idea for the quantitative examination on lip diagnosis of TCM, as well as provides a template for objective diagnosis in TCM.     

一种白血细胞图像训练集扩充方法

针对因训练集较小导致的白血细胞图像识别精度低以及传统的扩充训练集方法需要人工介入的问题,提出一种白血细胞图像训练集扩充方法,将图像旋转任意角度后,提取因旋转产生的黑色区域边缘,然后对黑色区域进行填充,并弱化边缘特征,得到扩充训练集。实验结果表明,使用本文方法扩充训练集对ResNet50、MobileNet与ShuffleNet 3种模型进行训练后,对比原始数据集,模型的识别精度分别提高220.18%、140.84%与88.99%,且不需要人工介入。     

一种快速的全自动超声子宫图像分割算法

使用超声成像进行子宫节育环检查工作已在我国广泛地开展,利用图像识别技术进行计算机辅助诊断对于减轻检查人员工作负担意义十分明显,其中图像分割部分的主要目标则是快速地全自动分割开图中的几个主要器官及节育环。本研究提出了一种快速的全自动子宫图像分割算法。该算法包括以下三个主要步骤:首先运用BP神经网络处理图像整体灰度分布获取基准分割阈值;其后使用超声图像斑点噪声统计特征进行同质区域判别,并根据局部灰度分布自适应调整分割阈值;最后使用数学形态学算子对分割效果做进一步的改善。基于由1200幅超声子宫图像组成的图像库,对所提算法与最大类别方差法、SNAKE活动轮廓模型等数种常用分割算法进行了性能比较,实验结果表明所提算法在速度与准确程度两方面均表现良好,平均耗时为0.93s/幅,准确程度达到了94%。本算法无需人工干预,分割速度快,分割准确程度能够被临床医生所接受,可以用作超声子宫图像辅助诊断系统的图像分割部分,具有很好的应用前景。     

基于W-Net的肝静脉和肝门静脉全自动分割

基于术前CT影像的肝静脉和肝门静脉分割对于进行肝脏分段具有重要的临床价值。但在肝脏的静脉期CT影像中,肝静脉和肝门静脉的灰度差异很小,血管结构也错综复杂,因此自动提取三维的肝静脉和肝门静脉一直是个难题。为解决此难题,提出一种基于卷积神经网络(convolutional neural network,CNN)的网络结构W-Net。该结构利用肝静脉和肝门静脉在三维结构上的差异,为全部肝血管和门静脉的提取分别设置损失函数,通过优化两个损失函数的加权和,自动学习全部肝血管和门静脉的三维结构特征,使全部肝血管和肝门静脉均达到优化的提取效果,两者相减即可得到肝静脉。采用公开数据集3Dircadb01中的10组延迟期腹部CT影像用于网络模型构建,另外10组用于测试。结果显示,肝区全部血管Dice系数达到0.715,准确率达到0.970;肝静脉Dice系数达到0.597,准确率达到0.984;肝门静脉Dice系数达到0.608,准确率达到0.970。通过10组临床数据进行测试,所构建的网络均能将肝静脉和肝门静脉有效地分割开。实验结果表明,所提出的方法具有较好的特征提取能力及泛化能力,在公开数据和临床数据中都有较好的表现。     

Feature selection and classification of leukocytes using random forest

In automatic segmentation of leukocytes from the complex morphological background of tissue section images, a vast number of artifacts/noise are also extracted causing large amount of multivariate data generation. This multivariate data degrades the performance of a classifier to discriminate between leukocytes and artifacts/noise. However, the selection of prominent features plays an important role in reducing the computational complexity and increasing the performance of the classifier as compared to a high-dimensional features space. Therefore, this paper introduces a novel Gini importance-based binary random forest feature selection method. Moreover, the random forest classifier is used to classify the extracted objects into artifacts, mononuclear cells, and polymorphonuclear cells. The experimental results establish that the proposed method effectively eliminates the irrelevant features, maintaining the high classification accuracy as compared to other feature reduction methods.     

Electrophoretic properties of esterases in chronic granulocytic leukemia.

Granulocytes were obtained from samples of peripheral blood of five patients who had untreated chronic granulocytic leukemia, and polymorphonuclear leukocytes were isolated from peripheral blood of three normal persons. Specific and nonspecific esterases were extracted from leukocyte preparations with cetyltrimethylammonium bromide and with lysolecithin, and subjected to polyacrylamide disk electrophoresis. In samples from both patients and normal persons, electrophoretic patterns of nonspecific esterase activity using alpha-naphthyl acetate and alpha-naphthyl butyrate were similar, and the esterase bands were weakly inhibited by fluoride. Lysolecithin extracts of specific esterase showed similar electrophoretic patterns for patients and normal subjects. However in cetyltrimethylammonium bromide extracts of specific esterase, 11 bands were seen in preparations from all of the patients with chronic granulocytic leukemia. In preparations of normal polymorphonuclear leukocytes, only eight bands were visualized. The results are consistent with an interpretation that these fast-moving components of specific esterase in chronic granulocytic leukemia granulocytes are present in normal polymorphonuclear leukocytes, but in quantities too small to be visualized with the technics used. Alternatively, the apparent "additional" bands of specific esterase may reflect abnormal metabolism of malignant granulocytes in chronic granulocytic leukemia.     

Auxiliary Diagnosis for COVID-19 with Deep Transfer Learning

To assist physicians identify COVID-19 and its manifestations through the automatic COVID-19 recognition and classification in chest CT images with deep transfer learning. In this retrospective study, the used chest CT image dataset covered 422 subjects, including 72 confirmed COVID-19 subjects (260 studies, 30,171 images), 252 other pneumonia subjects (252 studies, 26,534 images) that contained 158 viral pneumonia subjects and 94 pulmonary tuberculosis subjects, and 98 normal subjects (98 studies, 29,838 images). In the experiment, subjects were split into training (70%), validation (15%) and testing (15%) sets. We utilized the convolutional blocks of ResNets pretrained on the public social image collections and modified the top fully connected layer to suit our task (the COVID-19 recognition). In addition, we tested the proposed method on a finegrained classification task; that is, the images of COVID-19 were further split into 3 main manifestations (ground-glass opacity with 12,924 images, consolidation with 7418 images and fibrotic streaks with 7338 images). Similarly, the data partitioning strategy of 70%-15%-15% was adopted. The best performance obtained by the pretrained ResNet50 model is 94.87% sensitivity, 88.46% specificity, 91.21% accuracy for COVID-19 versus all other groups, and an overall accuracy of 89.01% for the three-category classification in the testing set. Consistent performance was observed from the COVID-19 manifestation classification task on images basis, where the best overall accuracy of 94.08% and AUC of 0.993 were obtained by the pretrained ResNet18 (P < 0.05). All the proposed models have achieved much satisfying performance and were thus very promising in both the practical application and statistics. Transfer learning is worth for exploring to be applied in recognition and classification of COVID-19 on CT images with limited training data. It not only achieved higher sensitivity (COVID-19 vs the rest) but also took far less time than radiologists, which is expected to give the auxiliary diagnosis and reduce the workload for the radiologists.     

基于深度学习结合解剖学注意力机制的肺结节良恶性分类

肺结节作为肺癌的初期表现,及时的发现和准确的良恶性诊断对于疾病的治疗具有重要的意义。为了提高肺部CT图像中肺结节良恶性的诊断率,提出一种基于3D ResNet的卷积神经网络,并通过加入解剖学注意力模块有效地提高了肺结节良恶性的分类精度。此外,该方法通过自动分割以获取注意力机制所需的感兴趣区域,实现整个流程的全自动化。解剖学注意力的添加能更好地捕捉图像中的局部纹理信息,进一步提取对于肺结节良恶性诊断有用的特征。本文方法在LIDC-IDRI数据集上进行验证。实验结果表明与传统的3D ResNet及其他现有的方法相比,本文方法在分类精度上有显著的提高,在独立测试集上的最终分类的AUC达到0.973,准确率为0.940。由此可见,本文方法能在辅助医生对肺结节的诊断中起到重要作用。     

革兰氏染色细菌显微图像深度学习分类与计数

菌种和数量是研究菌群失调和疾病预测的重要参数,然而细菌分类和计数工作主要由人工完成,过程繁琐,极易出错,并且耗时费力。本研究提出一种基于图像深度学习的方法对显微图像中的革兰氏阳性杆菌、革兰氏阴性杆菌、革兰氏阳性球菌和革兰氏阴性球菌进行分类。整个算法过程包括分割和分类识别两部分,首先采用U-Net“渐进分割法”对细菌部分和背景部分进行分割;然后将分割后的细菌分别投入ResNet50模型和VGG19模型进行识别和计数。将经过再训练ResNet50模型和VGG19模型的计数结果与人工分类计数标准的结果进行比较,实验结果表明ResNet50模型可以达到人工分类和计数的准确率。