Evaluation of cell segmentation methods without reference segmentations

Mapping Intimacies ◽

10.1101/2021.09.17.460800 ◽

2021 ◽

Author(s):

Haoran Chen ◽

Robert F. Murphy

Keyword(s):

Principal Component ◽

Evaluation Process ◽

Cell Segmentation ◽

Reproducible Research ◽

Computational Error ◽

Bioimage Informatics ◽

Cellular Analysis ◽

Segmentation Methods ◽

Segmentation Quality ◽

Unknown Bias

AbstractCell segmentation is a cornerstone of many bioimage informatics studies. Inaccurate segmentation introduces computational error in downstream cellular analysis. Evaluating the segmentation results is thus a necessary step for developing the segmentation methods as well as choosing the most appropriate one for a certain kind of tissue or image. The evaluation process has typically involved comparison of segmentations to those generated by humans, which can be expensive and subject to unknown bias. We present here an approach that seeks to evaluate cell segmentation methods without relying upon comparison to results from humans. For this, we defined a number of segmentation quality metrics that can be applied to multichannel fluorescence images. We calculated these metrics for 11 previously-described segmentation methods applied to datasets from 5 multiplexed microscope modalities covering 5 tissues. Using principal component analysis to combine the metrics we defined an overall cell segmentation quality score and ranked the segmentation methods. A Reproducible Research Archive containing all data and code will be made available upon publication at http://hubmap.scs.cmu.edu.

Download Full-text

Unsupervised Segmentation Evaluation Using Area-Weighted Variance and Jeffries-Matusita Distance for Remote Sensing Images

Remote Sensing ◽

10.3390/rs10081193 ◽

2018 ◽

Vol 10 (8) ◽

pp. 1193 ◽

Cited By ~ 8

Author(s):

Yongji Wang ◽

Qingwen Qi ◽

Ying Liu

Keyword(s):

Distance Measures ◽

Unsupervised Segmentation ◽

Segmentation Evaluation ◽

Object Based Image Analysis ◽

Object Based ◽

Segmentation Methods ◽

Segmentation Quality ◽

Two Measures ◽

Multiresolution Segmentation ◽

Adjacent Segments

Image segmentation is an important process and a prerequisite for object-based image analysis. Thus, evaluating the performance of segmentation algorithms is essential to identify effective segmentation methods and to optimize the scale. In this paper, we propose an unsupervised evaluation (UE) method using the area-weighted variance (WV) and Jeffries-Matusita (JM) distance to compare two image partitions to evaluate segmentation quality. The two measures were calculated based on the local measure criteria, and the JM distance was improved by considering the contribution of the common border between adjacent segments and the area of each segment in the JM distance formula, which makes the heterogeneity measure more effective and objective. Then the two measures were presented as a curve when changing the scale from 8 to 20, which can reflect the segmentation quality in both over- and under-segmentation. Furthermore, the WV and JM distance measures were combined by using three different strategies. The effectiveness of the combined indicators was illustrated through supervised evaluation (SE) methods to clearly reveal the segmentation quality and capture the trade-off between the two measures. In these experiments, the multiresolution segmentation (MRS) method was adopted for evaluation. The proposed UE method was compared with two existing UE methods to further confirm their capabilities. The visual and quantitative SE results demonstrated that the proposed UE method can improve the segmentation quality.

Download Full-text

Algorithms for Cytoplasm Segmentation of Fluorescence Labelled Cells

Analytical Cellular Pathology ◽

10.1155/2002/821782 ◽

2002 ◽

Vol 24 (2-3) ◽

pp. 101-111 ◽

Cited By ~ 101

Author(s):

Carolina Wählby ◽

Joakim Lindblad ◽

Mikael Vondrus ◽

Ewert Bengtsson ◽

Lennart Björkesten

Keyword(s):

Statistical Analysis ◽

Single Cells ◽

Quality Measurement ◽

Feedback System ◽

Cell Segmentation ◽

New Combination ◽

Processing Step ◽

Segmentation Quality ◽

Fully Automatic

Automatic cell segmentation has various applications in cytometry, and while the nucleus is often very distinct and easy to identify, the cytoplasm provides a lot more challenge. A new combination of image analysis algorithms for segmentation of cells imaged by fluorescence microscopy is presented. The algorithm consists of an image pre‐processing step, a general segmentation and merging step followed by a segmentation quality measurement. The quality measurement consists of a statistical analysis of a number of shape descriptive features. Objects that have features that differ to that of correctly segmented single cells can be further processed by a splitting step. By statistical analysis we therefore get a feedback system for separation of clustered cells. After the segmentation is completed, the quality of the final segmentation is evaluated. By training the algorithm on a representative set of training images, the algorithm is made fully automatic for subsequent images created under similar conditions. Automatic cytoplasm segmentation was tested on CHO‐cells stained with calcein. The fully automatic method showed between 89% and 97% correct segmentation as compared to manual segmentation.

Download Full-text

Measures of Image and Video Segmentation

Applied Video Processing in Surveillance and Monitoring Systems - Advances in Multimedia and Interactive Technologies ◽

10.4018/978-1-5225-1022-2.ch002 ◽

2017 ◽

pp. 28-53

Author(s):

Pushpajit A. Khaire ◽

Roshan R. Kotkondawar

Keyword(s):

Image Segmentation ◽

Video Segmentation ◽

Evaluation Process ◽

Final Evaluation ◽

Evaluation Metrics ◽

Introductory Section ◽

Segmentation Methods ◽

Segmentation Algorithms ◽

Benchmark Datasets ◽

Traditional Image

Study on Video and Image segmentation is currently limited by the lack of evaluation metrics and benchmark datasets that covers the large variety of sub-problems appearing in image and video segmentation. Proposed chapter provides an analysis of Evaluation Metrics, Datasets for Image and Video Segmentation methods. Importance is on wide-ranging, Datasets robust Metrics which used for evaluation purposes without inducing any bias towards the evaluation results. Introductory Section discusses traditional image and video segmentation methods available, the importance and need of measures, metrics and dataset required to evaluate segmentation algorithms are discussed in next section. Main focus of the chapter explains the measures, metrics and dataset available for evaluation of segmentation techniques of both image and video. The goal is to provide details about a set of impartial datasets and evaluation metrics and to leave the final evaluation of the evaluation process to the understanding of the reader.

Download Full-text

Toward consistent cell segmentation: quality assessment of cell segments via appearance and geometry features

10.1117/12.2082329 ◽

2015 ◽

Cited By ~ 1

Author(s):

Andrew Brinker ◽

Annika Fredrikson ◽

Xiaofan Zhang ◽

Richard Sourvenir ◽

Shaoting Zhang

Keyword(s):

Quality Assessment ◽

Cell Segmentation ◽

Segmentation Quality

Download Full-text

Comparative assessment of the optical-electronic images segmentation quality by the ant colony optimization and the artificial bee colony

Системи обробки інформації ◽

10.30748/soi.2021.164.11 ◽

2021 ◽

pp. 104-113

Author(s):

Г.В. Худов ◽

І.А. Хижняк

Keyword(s):

Ant Colony Optimization ◽

Swarm Intelligence ◽

Artificial Bee Colony ◽

Visual Assessment ◽

Comparative Assessment ◽

Ant Colony ◽

Bee Colony ◽

Segmentation Methods ◽

Segmentation Quality

The article discusses the methods of swarm intelligence, namely, an improved method based on the ant colony optimization and the method of an artificial bee colony. The goal of the work is to carry out a comparative assessment of the optical-electronic images segmentation quality by the ant colony optimization and the artificial bee colony. Segmentation of tonal optical-electronic images was carried out using the proposed methods of swarm intelligence. The results of the segmentation of optical-electronic images obtained from the spacecraft are presented. A visual assessment of the quality of segmentation results was carried out using improved methods. The classical errors of the first and second kind of segmentation of optoelectronic images are calculated for the proposed methods of swarm intelligence and for known segmentation methods. The features of using each of the proposed methods of swarm intelligence are determined. The tasks for which it is better to use each of the proposed methods of swarm intelligence are determined.

Download Full-text

Patch-Wise Semantic Segmentation for Hyperspectral Images via a Cubic Capsule Network with EMAP Features

Remote Sensing ◽

10.3390/rs13173497 ◽

2021 ◽

Vol 13 (17) ◽

pp. 3497

Author(s):

Le Sun ◽

Xiangbo Song ◽

Huxiang Guo ◽

Guangrui Zhao ◽

Jinwei Wang

Keyword(s):

Neural Network ◽

Hyperspectral Image ◽

Principal Component ◽

Semantic Segmentation ◽

Hyperspectral Images ◽

Experimental Comparison ◽

Spatial Features ◽

Segmentation Methods ◽

High Layer ◽

Morphological Attribute

In order to overcome the disadvantages of convolution neural network (CNN) in the current hyperspectral image (HSI) classification/segmentation methods, such as the inability to recognize the rotation of spatial objects, the difficulty to capture the fine spatial features and the problem that principal component analysis (PCA) ignores some important information when it retains few components, in this paper, an HSI segmentation model based on extended multi-morphological attribute profile (EMAP) features and cubic capsule network (EMAP–Cubic-Caps) was proposed. EMAP features can effectively extract various attributes profile features of entities in HSI, and the cubic capsule neural network can effectively capture complex spatial features with more details. Firstly, EMAP algorithm is introduced to extract the morphological attribute profile features of the principal components extracted by PCA, and the EMAP feature map is used as the input of the network. Then, the spectral and spatial low-layer information of the HSI is extracted by a cubic convolution network, and the high-layer information of HSI is extracted by the capsule module, which consists of an initial capsule layer and a digital capsule layer. Through the experimental comparison on three well-known HSI datasets, the superiority of the proposed algorithm in semantic segmentation is validated.

Download Full-text

Screening Quality Evaluation Factors of Freeze-Dried Peach (Prunus PersicaL. Batsch) Powders from Different Ripening Time Cultivars

Journal of Food Quality ◽

10.1155/2017/7213694 ◽

2017 ◽

Vol 2017 ◽

pp. 1-12 ◽

Cited By ~ 3

Author(s):

Chun-ju Liu ◽

Hai-ou Wang ◽

You-lin Xue ◽

Zhong-yuan Zhang ◽

Li-ying Niu ◽

...

Keyword(s):

Quality Evaluation ◽

Principal Component ◽

Evaluation Process ◽

Analytic Hierarchy ◽

Ripening Time ◽

Titratable Acid ◽

Freeze Dried ◽

Clear Differentiation ◽

Processed Products ◽

Hierarchy Process

The quality evaluation of processed products is complex. To simplify the quality evaluation process and improve the efficiency, fourteen evaluation factors of freeze-dried powders of seventeen cultivars of peach at different ripening times were analyzed. The most important evaluation indicators and criteria were obtained by analysis of variance (ANOVA), correlation analysis (CA), principal component analysis (PCA), system cluster analysis (SCA), and analytic hierarchy process (AHP). Results showed that the peach powders had the significant differences in quality (P<0.05), and some processing factors were related with some physicochemical and nutritional factors. Five principle components were extracted by PCA and the cumulative contribution achieved was 84.46%. Through the score plot of the first two principal components, a clear differentiation among ripening times was found and three distinct groups were separated according to ripening time. Five characteristic factors were obtained as titratable acid, browning index, hemicellulose, hygroscopicity, and vitamin C by SCA. Their weights of 0.1249, 0.3007, 0.0514, 0.4916, and 0.0315 were obtained by AHP, respectively. The peach cultivars were divided into four evaluation grades by the comprehensive quality score.

Download Full-text

Segmentation of the Haematoxylin and Eosin Stained Muscle Cell Images—A Comparative Study

Journal of Medical Imaging and Health Informatics ◽

10.1166/jmihi.2021.3449 ◽

2021 ◽

Vol 11 (6) ◽

pp. 1517-1526

Author(s):

Zihao Wang ◽

Liju Yin ◽

Shuai Mao ◽

Zhenzhou Wang

Keyword(s):

Comparative Study ◽

Muscle Cell ◽

Biomedical Research ◽

Muscle Cells ◽

Morphological Features ◽

Cell Segmentation ◽

Cross Sectional ◽

Segmentation Methods ◽

Automatic Methods ◽

Automated Methods

The effective detection of muscle cells, the accurate counting of their numbers and the analysis of their morphological features have great importance in biomedical research. At present, the quantification of muscle cell and the computation of their cross-sectional areas (CSA) are still manual or semi-automated, and with the increase of the image number, the manual or semi-automated methods might become intractable. Hence, the automatic methods are very desirable, which motivated the developments of many muscle cell segmentation methods. In this paper, three methods, SDDM, CELLSEGM and SMASH are compared and evaluated with 100 images with over 6000 cells. The Dices computed by SDDM, CELLSEGM and SMASH are 97.38%, 89.85% and 90.08% respectively. The average differences between the calculated cross-sectional areas and the ground truths by SDDM, CELLSEGM and SMASH are 5.14%, 10.76% and 7.97% respectively.

Download Full-text