Identification of selected monogeneans using image processing, artificial neural network and K-nearest neighbor

Item Type article

Authors Yousef Kalafi, E.; Wooi Boon, T.; Town, C.; Kaur Dhillon, S.

Download date 02/10/2021 11:21:57

Link to Item http://hdl.handle.net/1834/40338 Iranian Journal of Fisheries Sciences 17(4) 805-820 2018 DOI:10.22092/ijfs.2018.117017 Identification of selected monogeneans using image processing, artificial neural network and K-nearest neighbor

٭Yousef Kalafi E.¹; Wooi Boon T.¹; Town C.²; Kaur Dhillon S.¹

Received: June 2017 Accepted: August 2017

Abstract Over the last two decades, improvements in developing computational tools have made significant contributions to the classification of images of biological specimens to their corresponding species. These days, identification of biological species is much easier for taxonomists and even non-taxonomists due to the development of automated computer techniques and systems. In this study, we developed a fully automated identification model for monogenean images based on the shape characters of the haptoral organs of eight species: Sinodiplectanotrema malayanum, Diplectanum jaculator, Trianchoratus pahangensis, Trianchoratus lonianchoratus, Trianchoratus malayensis, Metahaliotrema ypsilocleithru, Metahaliotrema mizellei and Metahaliotrema similis. Linear Discriminant Analysis (LDA) method was used to reduce the dimension of extracted feature vectors which were then used in the classification with K-Nearest Neighbor (KNN) and Artificial Neural Network (ANN) classifiers for the identification of monogenean specimens of eight species. The need for the discovery of new characters for identification of species has been

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 acknowledged for log by systematic parasitology. Using the overall form of anchors and bars for extraction of features led to acceptable results in automated classification of monogeneans. To date, this is the first fully automated identification model for monogeneans with an accuracy of 86.25% using KNN and 93.1% using ANN.

Keywords: Monogenean, Morphology, Fish parasite, Artificial neural networks, K- nearest neighbor

1-Institute of Biological Sciences, Faculty of Science, University of Malaya, Kuala Lumpur, Malaysia 2-Computer Laboratory, University of Cambridge, Cambridge CB3 0FD, UK *Corresponding author's Email: [email protected] 806 Yousef Kalafi et al., Identification of selected monogeneans using image…

Introduction morphology and ecology and with Monogeneans are platyhelminthes respect to the variation of structural which are characterized by having a designs in the attachment organs proper body and haptor with sizes (Kearn, 1994), which are usually used ranging from 0.5mm to 1-2cm in length for species identification. The haptoral live on lower aquatic invertebrates or attachment organs of monogeneans are the gills, skin or fins of fishes as hosts. sclerotized structures of anchors, bars Their appendage attachments in their and marginal hooks. In particular, the anterior and posterior (haptoral) regions morphology of each of these organsis are used to prevent physical unique to monogenean species (Boeger dislodgement from the host (Fig. 1). and Kritsky, 1993) and is used as a diagnostic feature in their taxonomical classification (Vignon, 2011). Earlier, Active Shape Models (ASM) (Ali et al., 2012) were used to classify several Gyrodactylus species according to attachment hooks. ASM were applied to extract diagnostic information from hook images as features. Extracted features were used as input data to Linear Discriminant Analysis (LDA) , K-Nearest Neighbor (KNN), Multilayer Perceptron (MLP) and Support Vector Machine (SVM) classifiers. According to Khang et al. (2016), data from size

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 and shape of anchors were generated using geometric morphometrics. They used principal components and cluster analysis to classify 13 species of Ligophorus. Innovations in the area of computer vision have significantly contributed to the development of automated taxonomic identification systems such Figure 1: Illustration of a monogenean worm as an automated identification system consisting of three main parts: head, body which estimates densities of whiteflies, and haptor. aphids and thrips in a greenhouse (CHO Monogeneans are a diverse group, with et al., 2008), automatic image several thousand species described in recognition and diagnosis of protozoan the world (Poulin, 2002). The diversity parasites (Castañón et al., 2007), of monogeneans is not only in terms of automatic recognition of biological numbers but also in terms of their particles in microscopic images Iranian Journal of Fisheries Sciences 17(4) 2018 807

(Ranzato et al., 2007), automatic methods are performed on every single detection of malaria parasites for image (specimen) which substantially estimating parasitemia (Savkare and slows down the process of Narote, 2011), automated identification identification and classification. Hence, of copepods using digital image we propose a fully automated processing and artificial neural identification model for monogeneans networks (Leow et al., 2015), which is robust with respect to variable automated identification of fish species imaging conditions and damaged based on otolith contour using short- specimens. time Fourier transform and discriminant analysis (STFT-DA) (Salimi et al., Materials and methods 2016), and other systems (Larios et al., Recognition of monogeneans is based 2008; Vogt et al., 2009; Mansoor et al., on morphometric features of their hard 2011; Feng et al., 2016; Perre et al., parts (Lim and Gibson, 2010). For this 2016). study, images of the hard haptoral Many classification methods such as organs such as anchors and bars were Artificial Neural Network (ANN) captured using a Leica digital camera (Yang et al., 2001; Cho et al., 2008; DFC 320 attached to Leica DMRB Mansoor et al., 2011), KNN (Keller et microscope at 40× magnification. The al., 1985; Parisi-Baradad et al., 2010), resolution of the images was 1044×772 SVM (Thiel et al., 1996; Pronobis et pixels and they were saved in Tagged al., 2010), Discriminant Analysis (DA) Image File format (TIF). (Thiel et al., 1996; Salimi et al., 2016), Our database consists of 160 images Decision trees (Jalba et al., 2005), from 8 species (20 images of each Semantically-Related Visual (SRV) species): Sinodiplectanotrema

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 (Feng and Bhanu, 2013), and malayanum, Diplectanum jaculator, Convolutional Neural Networks Trianchoratus pahangensis, (Gomez et al., 2016), etc. have been Trianchoratus lonianchoratus, utilized for developing automated Trianchoratus malayensis, identification systems. Metahaliotrema ypsilocleithru, Automated classification of images Metahaliotrema mizellei and of specimens requires development of Metahaliotrema similis. Fig. 2 models and methods that are able to illustrates the flowchart for the characterize species images based on development of automated the texture or shape of objects to extract identification for monogeneans. important visual information for classification. Current approaches in monogenean identification rely heavily on manual input during image processing and feature extraction such as specifying morphological landmark features. These manual identification 808 Yousef Kalafi et al., Identification of selected monogeneans using image…

features for the next process in feature extraction. Preprocessing started with converting RGB images to intensity images. Then intensity images were filtered and edges of anchors and bars were detected (Fig. 4).

Figure 4: The process of detecting edges from intensity image. Figure 2: Flowchart for development of proposed model for monogenean identification. Since the segmented images contained negative and positive values, the images Preprocessing were binarized with a threshold of zero. Monogenean specimen images are very Then the borders were cleared and complex due to their messy background objects smaller than 1000 pixels were and overlapping of anchors and bars. removed (Fig. 5). Coordinates of Despite consistent efforts to acquire contour pixels for species` anchors clear images, some overlapping and were calculated. Features were clutter were unavoidable (Fig. 3). extracted either from all anchors and Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 bars as a consolidated object or from individual anchors.

Figure 3: The illustration of anchors and bars of Metahaliotrema ypsilocleithrum. a) The illustration of dorsal and ventral Figure 5: The process of converting binary anchors and bars. b) The microscopic image image to segmented image. of anchors and bars and their overlapping.

Feature extraction Hence, preprocessing played an Binary images were used two times for important role to omit redundant feature extraction; (i) firstly using all information and to highlight reliable Iranian Journal of Fisheries Sciences 17(4) 2018 809

anchors and bars as a consolidated of 160×7 feature vector. object and (ii) secondly by calculating coordinates of one anchor and then Classification extracting the features from that anchor. The features that were extracted and Features extracted consisted of: length selected in the previous stage were used of bounding box, width of bounding as inputs to KNN and ANN classifiers box, center of bounding box, to train the system based on a training orientation of bounding box, perimeter, set and evaluate performance of each perimeter density, area, area density, trained model on a testing dataset. Euler number, entropy, major axis Commonly, practical and theoretical length and number of white pixels in data do not follow the same white area. A feature vector with 24 assumptions and since our dataset was elements was computed from these from the real world, it was an advantage features. to use KNN. Therefore, no hypothesis was made on the fundamental data Feature selection distribution. In addition, based on Feature selection is a technique for previous approaches in the reducing the dimensionality of feature classification of monogenean samples vectors. In this study, the informative (McHugh et al., 2000; Ali et al., 2011; and independent feature vectors were Ali et al., 2012), the performance of transformed to smaller dimensions by KNN was as reliable to be used in our using the LDA (Park and Park, 2008) investigation. In this study, the trained method (Song et al., 2010). The goal of model from the KNN classifier was LDA is to distinguish multiple classes constructed using 80 images (10 for by maximizing the inter-class variance each species) and tested with 80 images

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 and minimizing the intra-class variance. of monogeneans. According to Therefore, besides projecting a feature successful experiments in (Sang-Hee, space to a smaller subspace, the class- 2010; Jin et al., 2015), we had decided discriminatory information is also to use half (10 images) of each species maintained. In this approach, firstly, 24 images to train the classifier and the dimensional mean vectors for each of other half as a testing set and the best the 8 classes were calculated. After results were achieved with 9 nearest computing the in-between class and neighbors. The ANN classifier structure within-class scatter matrix, the was a two layer feed-forward network corresponding eigenvalues and with ten sigmoid hidden nodes and eigenvectors were calculated. Then eight output neurons (Fig. 6). The eigenvectors were sorted according to number of nodes in the hidden layer is descending eigenvalues and seven important since they may cause eigenvectors with largest eigenvalues overfitting (if there are so many were selected to form 24×7 dimensional neurons in hidden layer) or underfitting matrix W. The matrix W will be used to (if the number of neurons are few), transform the samples to new subspace therefore, we used trial and error 810 Yousef Kalafi et al., Identification of selected monogeneans using image…

method to select the number of hidden For evaluating the trained network we neurons. This approach began by used confusion matrices and Mean selecting two to fifteen nodes. The best Square Error (MSE). result was achieved with ten nodes. We In both KNN and ANN the best divided the entire dataset (160 images) results were achieved after 20 into three random subsets: training iterations. The best results in KNN (70%), testing (15%) and validation classification were obtained using 50% (15%) set. The training dataset was of samples in testing and 50% in used for training ANN, the testing training data. On the other hand the best dataset for performance measurement results in ANN were accomplished by of the networks and the validation set to using 70% of samples in training, 15% measure generalization of network and in testing and 15% in validation set. terminate training before overfitting.

Figure 6: Neural Network with 10 sigmoid hidden nodes and eight output neurons.

Results with 24 elements was transformed to a The results of this study were achieved lower dimensional feature space with Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 by using QWin Plus software package seven distinct elements. First, 24 as imaging modular and MATLAB dimensional mean vectors for each of R2013a [28] as simulation and image the 8 classes were calculated. After processing and classification tool, computing the in-between class and installed on Intel(R) Xeon (R) CPU E5- within-class scatter matrix, the 1620 v2 @ 3.70GHz, 16 GB RAM, corresponding eigenvalues and Windows 7 Professional (64-bit). eigenvectors were calculated. Then eigenvectors were sorted according to Feature selection descending eigenvalues and seven The feature space was defined eigenvectors with largest eigenvalues according to shape characteristics of were selected to form 24×7 dimensional anchors and bars of monogeneans. A matrix W. The matrix W will be used to total of 24 feature elements were transform the samples to a new initially extracted (Table 1). By subspace of 160×7 feature vector. adopting LDA (dimensionality reduction method), the feature vector

Iranian Journal of Fisheries Sciences 17(4) 2018 811

Table 1: Extracted features from shape characteristics of monogenean. Anchor Anchors and bars together Features × × Area × × Area density × × Perimeter × × Perimeter density × × Length of bounding box × × Width of bounding box × × Center of bounding box × × Orientation of bounding box × × Euler number × × Entropy × × Major axis length × Number of white pixels in white area

The distribution of some of the feature LDA feature transformation, the values are illustrated in Fig. 7 as a 3D features are transformed to a discrete scatter plot. It is notable that in Fig. feature space in which the eight species 7(a,b), the features are not distinct are more easily separable for enough to separate and classify eight classification. classes. Instead, in Fig. 7 (c,d), after

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018

Figure 7: Distribution of feature values in 3D scatter plot. a) Illustration of distribution of features values which were extracted from area density, area and Euler number of anchors and bars of monogenean. b) Illustration of distribution of features values which were extracted from entropy, perimeter and Euler number of anchors and bars of monogenean. c & d) Illustration of distribution of features values which were selected by LDA.

Classification classified using KNN and ANN. In The experiment was conducted on eight KNN we used 80 images for training species of four monogenean families, and 80 images for testing the trained 812 Yousef Kalafi et al., Identification of selected monogeneans using image…

model. In ANN, 112 images were used K-Nearest Neighbor (KNN) for training, 24 images for testing the In KNN classification, we tried the network and 24 images for system classification in 15 iterations and the validation. We found that ANN best score was achieved with 9 nearest accuracy of 93.1% on the test set neighbors (Fig. 8). outperformed KNN classifier accuracy of 86.25%.

Figure 8: Accuracy of KNN with different K values. It is notable that when k is 9 and 10 KNN classifier yielded the best performance.

According to the confusion matrix one misclassification with M. (Table 2), the overall classification ypsilocleithru. T. malayensis has one score for 8 species was 86.25%. In misclassification with T. pahangensis. Table 2, S. malayanum and M. similis M. ypsilocleithru has one were identified correctly. T. misclassification with samples of T. pahangensis has one misclassification pahangensis, T. malayensis, M. similis with M. similis. M. mizellei has one and two misclassifications with M.

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 misclassification with T. pahangensis. mizellei. Finally, D. jaculator has only T. lonianchoratus has one one misclassification with M. mizellei. misclassification with M. mizellei and

Table 2: Confusion matrix of KNN classification for 8 species of Sinodiplectanotrema malayanum (Smm), Diplectanum jaculator (Dj), Trianchoratus pahangensis (Tp), Trianchoratus lonianchoratus (Tl), Trianchoratus malayensis (Tm), Metahaliotrema ypsilocleithru (My), Metahaliotrema mizellei (Mmi) and Metahaliotrema similis (Mma).

Accuracy Predicted (%)

Species Smm Tp Mmi Mma Tl Tm My Dj Smm 10 0 0 0 0 0 0 0 100 Tp 0 9 0 1 0 0 0 0 90 Mmi 0 1 9 0 0 0 0 0 90 Mma 0 0 0 10 0 0 0 0 100 Tl 0 0 1 0 8 0 1 0 80 Tm 0 1 0 0 0 9 0 0 90 Actual My 0 1 2 1 0 1 5 0 50 Dj 0 0 1 0 0 0 0 9 90 Overall 86.25

Iranian Journal of Fisheries Sciences 17(4) 2018 813

Artificial Neural Network (ANN) MSE of 0.026168 on the validated set at The ANN classification structure was a epoch 40 (Fig. 9). According to two layer feed-forward network which confusion matrix in Table 3, it is was trained with back propagation and notable that the best overall with respect to ten hidden neurons in accomplished classification was 93.1% the hidden layer and eight neurons in of all 160 images in the training, the output layer. After 46 iterations, the validation and testing sets. best trained network was selected with

Figure 9: Illustration of performance evaluation of trained network by MSE. Best trained network was constructed at epoch 40.

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 Table 3: Confusion matrix of overall performance of ANN classification for 8 species of Sinodiplectanotrema malayanum (Smm), Diplectanum jaculator (Dj), Trianchoratus pahangensis (Tp), Trianchoratus lonianchoratus (Tl), Trianchoratus malayensis (Tm), Metahaliotrema ypsilocleithru (My), Metahaliotrema mizellei (Mmi) and Metahaliotrema similis (Mma). Accuracy Predicted (%) Species Smm Tp Mmi Mma Tl Tm My Dj

Smm 20 0 0 0 0 0 0 0 100 Tp 0 19 0 0 0 1 0 0 95 Mmi 0 0 19 1 1 0 0 1 86.4 Mma 0 1 0 19 0 0 0 0 95

Tl 0 0 0 0 18 0 0 0 100 Actual Tm 0 0 0 0 1 18 1 0 90 My 1 0 1 0 0 0 17 0 89.5 Dj 1 0 0 0 0 1 0 19 90.5 Overall 93.1

814 Yousef Kalafi et al., Identification of selected monogeneans using image…

Discussion identification technique for The proposed automated identification monogeneans developed in this study method in this study was able to could identify the species with an classify monogeneans at the species acceptable accuracy. This shows the level with an overall accuracy of technique is robust with respect to 86.25% using K Nearest Neighbor variable imaging conditions and classification and 93.1% using damaged specimens (Figs. 10 and 11). Artificial Neural Network classification for eight species of monogenean. The images focused on anchors and bars of specimens since these organs contain diagnostic shape features which are used for classification of monogenean species.

Image processing In this experiment the database also Figure 10: The broken tail of anchor in contained some poor quality images Metahaliotrema ypsilocleithru which we had to use in this study as they were rare collections which were only available in small numbers. Despite that, the automated

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018

Figure 11: Variable imaging condition in terms of lighting source.

Additionally, extracting the anchors and to improved identification accuracy. bars as a single organ in poor quality images was also challenging since the Feature selection bars and anchors were overlapping. The extracted features from anchors Such samples are difficult even for and bars were affected by microscopic experts to distinguish. Increasing the clutters, some broken specimens and quality of images in the future will lead overlapping of anchors and bars in Iranian Journal of Fisheries Sciences 17(4) 2018 815

images. LDA was used for transforming after using LDA (Fig. 12 (b)) the feature vectors to distinct feature space dimensionality of the feature vector of seven elements. As in Fig. 12 (a), the decreased and feature vectors also 24 features were close in values and transformed into a separable feature using them as input to classifiers, made space. many misclassifications as a result, but

Figure 12: Feature vector comparison after and before feature transformation. a) Illustration of 24 dimensional extracted feature vector for 80 samples. Except one of the features, the rest contain close values. b) Illustration of 7 dimensional feature vector which is the result of LDA feature transformation.

Classification to the similar shape of their tails. The In both KNN and ANN, S. malayanum classification of T. lonianchoratus in was classified correctly due to its ANN was 100% correct while in KNN Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 distinct shape and size of anchors and there were two misclassifications with bars in this species. There was one M. mizellei and M. ypsilocleithru. misclassification of T. pahangensis Comparing the classification with M. similis in KNN mainly because accuracy in ANN (93.1%) and the the shapes of their tails were similar accuracy in KNN (86.26%), we can and one misclassification with T. declare that ANN classifiers were more malayensis in ANN since both of them powerful in classifying samples than have three anchors. There was one KNN classifiers. This means in ANN misclassification of M. mizellei with M. the features were more distinct for similis in KNN since both are from the training the network but the distance same genus, as such the overall shape distinction in KNN was not sufficient of all anchors and bars as an object is for classification as much as ANN. similar. In KNN, the classification of M. similis was 100% correct while in Future works ANN there was one misclassification One of the critical issues affecting with T. pahangensis. This could be due performance of identification systems is 816 Yousef Kalafi et al., Identification of selected monogeneans using image…

the quality of the images. Improving the systematic parasitology (Vignon, 2011) quality of the data would lead to an and because of the lack of increase in identification accuracy. discrimination of traditional methods, The acquired images were two several researchers have used additional dimensional (2D) and due to the loss of points to take into account the some information in 2D imaging, it is maximum amount of shape information suggested that in future, the model be (Murith and Beverley-Burton, 1985; based on three dimensional (3D) Řehulková and Gelnar, 2005). Using images. As the solution to loss of overall form of anchors and bars for information in 2D imaging, in the study extraction of features led us to create by Leow et al. (2015), they used a built new characters in the morphological in function in their imaging software, classification of monogeneans which called Extended Depth of Focus had never been used before. Imaging (EFI) to create a single plane This study proposes a model for image with increased in-focus details. automated identification of images of In this study, two classification eight selected monogenean species and techniques were adopted for developing it works by running the commands in an automated identification system for the MATLAB workspace which means monogeneans. However, other there is no dedicated user interface. As classification techniques such as SVM, future work, a stand-alone Graphical DA, and decision tree may improve the User Interface (GUI) could be deployed performance of the system. The as an executable application for ease of classification performance in some of use by taxonomists. To increase the these methods is also dependent on the number of species in the proposed size of the database. The dataset size model, further enhancements are

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 can be increased with new collections required. of monogenean specimens through field work which we plan to conduct in the Acknowledgments future. The specimens used in this study The authors are thankful to late Prof. were archive specimens from Lim Susan Lim Lee Hong from whom all 1998; Lim, 2006 and Lim et al., 2010. the specimens were obtained. This In this paper, we presented a fully project was supported by University of automated identification technique that Malaya Postgraduate Research Fund classifies monogeneans at the species (PG092-2013B) to the first author and level based on microscopic images of the University of Malaya Research haptoral anchors and bars with an Grant (UMRG) Program Based Grant overall accuracy of 86.25% using K (RP008-2012A) and the University of Nearest Neighbors and 93.1% using Malaya`s Living Lab Grant Program – Artificial Neural Network. The need for Sustainability Science (LL020-16SUS) the discovery of new characters for to the fourth author. identification of species has been acknowledged for a long time by Iranian Journal of Fisheries Sciences 17(4) 2018 817

References Cho, J., Choi, J., Qiao, M., Ji, C.W. Ali, R., Hussain, A., Bron, J.E. and and Kim, H.Y., 2008. Automatic Shinn, A.P., 2011. Multi-stage identification of tobacco whiteflies, classification of Gyrodactylus aphids and thrips in greenhouse species using machine learning and using image processing techniques. feature selection techniques. In 2011 In 4th WSEAS International 11th International Conference on Conference on MATHEMATICAL Intelligent Systems Design and BIOLOGY and ECOLOGY. Applications. 2011 11th Mexico. pp. 46–53. International Conference on https://www.researchgate.net/profile/ Intelligent Systems Design and Chang_Woo_Ji2/publication/264853 Applications. pp. 457–462. 945_Automatic_Identification_of_T http://ieeexplore.ieee.org/document/ obacco_Whiteflies_Aphids_and_Thr 6121698/ 30 September 2016. ips_in_Greenhouse_Using_Image_P Ali, R., Hussain, A., Bron, J.E. and rocessing_Techniques/links/55ca8f7 Shinn, A.P., 2012. The use of ASM 108aebc967dfbe60e.pdf 30 feature extraction and machine September 2016. learning for the discrimination of Feng, L. and Bhanu, B., 2013. members of the fish ectoparasite Automated identification and genus Gyrodactylus. In T. Huang, Z. retrieval of moth images with Zeng, C. Li, & C. S. Leung, eds. semantically related visual attributes Neural Information Processing. on the wings. In 20th IEEE Lecture Notes in Computer Science. International Conference on Image Springer Berlin Heidelberg. pp. 256– Processing (ICIP). pp. 2577–2581. 263. http://ieeexplore.ieee.org/document/

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 http://link.springer.com/chapter/10.1 6738531/ 30 September 2016. 007/978-3-642-34478-7_32 30 Feng, L., Bhanu, B. and Heraty, J., September 2016. 2016. A software system for Boeger, W.A. and Kritsky, D.C., automated identification and 1993. Phylogeny and a revised retrieval of moth images based on classification of the Monogenoidea wing attributes. Pattern Recognition, Bychowsky, 1937 (Platyhelminthes). 51, 225–241. Systematic Parasitology, 26(1), 1– Gomez, A., Salazar, A. and Vargas, 32. F., 2016. Towards automatic wild Castañón, C.A.B., Fraga, J.S., animal monitoring: Identification of Fernandez, S., Gruber, A. and da animal species in camera-trap F. Costa, L., 2007. Biological shape images using very deep characterization for automatic image convolutional neural networks. recognition and diagnosis of arXiv:1603.06169 [cs]. protozoan parasites of the genus http://arxiv.org/abs/1603.06169 28 Eimeria. Pattern Recognition, 40(7), December 2016. 1899–1910. Jalba, A.C., Wilkinson, M.H.F., 818 Yousef Kalafi et al., Identification of selected monogeneans using image…

Roerdink, J.B.T.M., Bayer, M.M. digital image processing and and Juggins, S., 2005. Automatic artificial neural network. BMC diatom identification using contour Bioinformatics, 16(18), 1. analysis by morphological curvature Lim, L.H.S., 2006. Diplectanids scale spaces. Machine Vision and (Monogenea) on the archerfish Applications, 16(4), 217–228. Toxotes jaculatrix (Pallas) Jin, T., Hou, X., Li, P. and Zhou, F., (Toxotidae) off Peninsular Malaysia. 2015. A novel method of automatic Systematic Parasitology, 64(1), 13– plant species identification using 25. sparse representation of leaf tooth Lim, L.H.S., 1998. Diversity of features. PloS ONE, 10(10), monogeneans in Southeast Asia. e0139482. International Journal for Kearn, G.C., 1994. Evolutionary Parasitology, 28(10), 1495–1515. expansion of the monogenea. Lim, L.H.S. and Gibson, D.I., 2010. International Journal for Taxonomy, taxonomists and Parasitology, 24(8), 1227–1271. biodiversity. Sarawak Biodiversity Keller, J.M., Gray, M.R. and Givens, Centre, pp. 33–43. J.A., 1985. A fuzzy k-nearest Lim, L.H.S., Tan, W.B. and Gibson, neighbor algorithm. IEEE D.I., 2010. Description of Transactions on Systems, Man and Sinodiplectanotre mamalayanum n. Cybernetics, 4, 580–585. sp. (Monogenea: Diplectanidae), Khang, T.F., Soo, O.Y.M., Tan, W.B. with comments on the taxonomic and Lim, L.H.S., 2016. position of the genus. Systematic Monogenean anchor morphometry: Parasitology, 76(2), 145–157. systematic value, phylogenetic Mansoor, H., Sorayya, M., Aishah, S.,

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 signal, and evolution. PeerJ, 4, Mogeeb, A. and Mosleh, A., 2011. e1668. Automatic recognition system for Larios, N., Deng, H., Zhang, W., some cyanobacteria using image Sarpola, M., Yuen, J., Paasch, R., processing techniques and ANN Moldenke, A., Lytle, D.A., Correa, approach. In Int Conf on Envir and S.R., Mortensen, E.N., Shapiro, Comp Science IPCBEE. pp. 73–78. L.G. and Dietterich, T.G., 2008. McHugh, E.S., Shinn, A.P. and Kay, Automated insect identification J.W., 2000. Discrimination of the through concatenated histograms of notifiable pathogen Gyrodactylus local appearance features: feature salaris from G. thymalli vector generation and region (Monogenea) using statistical detection for deformable objects. classifiers applied to morphometric Machine Vision and Applications, data. Parasitology, 121(Pt 3), 315– 19(2), 105–123. 323. Leow, L.K., Chew, L.L., Chong, V.C. Murith, D. and Beverley-Burton, M. and Dhillon, S.K., 2015. Automated 1985. Salsuginus Beverley-Burton, identification of copepods using 1984 (Monogenea: Iranian Journal of Fisheries Sciences 17(4) 2018 819

Ancyrocephalidae) from Perona, P., 2007. Automatic Cyprinodontoidei (Atheriniformes) recognition of biological particles in in North America with descriptions microscopic images. Pattern of Salsuginus angularis (Mueller, Recognition Letters, 28(1), 31–39. 1934) Beverley-Burton, 1984 from Řehulková, E. and Gelnar, M., 2005. Fundulus diaphanus and Salsuginus A revised diagnosis of thylacicleidus heterocliti n. sp. from F. heteroclitus. (monogenea: dactylogyridae) with a Canadian Journal of Zoology, 63(3), redescription of the type species, 703–714. Thylacicleidus serendipitus, and Parisi-Baradad, V., Manjabacas, A., descriptions of two new species from Lombarte, A., Olivella, R., Chic, Southeast Asian pufferfishes ò., Piera, J. and García-Ladona, (tetraodontiformes: tetraodontidae). E., 2010. Automated taxon Journal of Parasitology, 91(4), 794– identification of teleost fishes using 807. an otolith online database—AFORO. Salimi, N., Loh, K.H., Dhillon, S.K. Fisheries Research, 105(1), 13–20. and Chong, V.C., 2016. Fully- Park, C.H. and Park, H., 2008. A automated identification of fish comparison of generalized linear species based on otolith contour: discriminant analysis algorithms. using short-time Fourier transform Pattern Recognition, 41(3), 1083– and discriminant analysis (STFT- 1097. DA). PeerJ, 4, e1664. Perre, P., Faria, F.A., Jorge, L.R., Sang-Hee, L., 2010. A novel approach Rocha, A., Torres, R.S., Souza- to shape recognition using shape Filho, M.F., Lewinsohn, T.M. and outline. Journal of the Korean Zucchi, R.A., 2016. Toward an Physical Society, 56(31), 1016.

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 automated identification of Savkare, S.S. and Narote, S.P., 2011. anastrepha fruit flies in the Automatic detection of malaria fraterculus group (Diptera, parasites for estimating parasitemia. Tephritidae). Neotropical International Journal of Computer Entomology. pp. 1–5. Science and Security (IJCSS), 5(3): Poulin, R., 2002. The evolution of 310. monogenean diversity. International Song, F., Mei, D. and Li, H., 2010. Journal for Parasitology, 32(3): Feature selection based on linear 245–254. discriminant analysis. In Intelligent Pronobis, A., Jie, L. and Caputo, B., System Design and Engineering 2010. The more you learn, the less Application (ISDEA), 2010 you store: Memory-controlled International Conference on. pp. incremental SVM for visual place 746–749. recognition. Image and Vision http://ieeexplore.ieee.org/xpls/abs_al Computing, 28(7), 1080–1097. l.jsp?arnumber=5743287 10 Ranzato, M., Taylor, P.E., House, November 2016. J.M., Flagan, R.C., LeCun, Y. and Thiel, S.U., Wiltshire, R.J. and 820 Yousef Kalafi et al., Identification of selected monogeneans using image…

Davies, L.J., 1996. Automated object recognition of blue-green algae for measuring water quality-a preliminary study. Oceanographic Literature Review, 1(43), 85. Vignon, M., 2011. Putting in shape – towards a unified approach for the taxonomic description of monogenean haptoral hard parts. Systematic Parasitology, 79(3), 161– 174. Vogt, A., Cholewinski, A., Shen, X., Nelson, S.G., Lazo, J.S., Tsang, M. and Hukriede, N.A., 2009. Automated image-based phenotypic analysis in zebrafish embryos. Developmental Dynamics, 238(3), 656–663. Yang, Y.S., Park, D.K., Kim, H.C., Choi, M.H. and Chai, J.Y., 2001. Automatic identification of human helminth eggs on microscopic fecal specimens using digital image processing and an artificial neural network. IEEE Transactions on

Downloaded from jifro.ir at 12:07 +0330 on Saturday October 6th 2018 Biomedical Engineering, 48(6), 718– 730.