Cristina M. R. Caridade

The use of texture for image classification of black & white air photographs

The use of black & white (B&W) air photographs for the production of historic land cover maps can be done by image classification, using additional texture features. In this paper we evaluate the importance of a number of parameters in the image classification process based on texture, such as the window size, angle and distance used to produce the texture features, the number of features used, the image quantization level and its spatial resolution. The evaluation was performed using five photographs from the 1950s. The influence of the classification method, the number of classes searched for in the images and the post‐processing tasks were also investigated. The effect of each of these parameters for the classification accuracy was evaluated by cross‐validation. The selection of the best parameters was performed based on the validation results, and also on the computation load involved for each case and the end user requirements. The final classification results were good (average accuracy of 85.7%, k = 0.809) and the method has proven to be useful for the production of historic land cover maps from B&W air photographs.

Evolutionary and Experimental Assessment of Novel Markers for Detection of Xanthomonas euvesicatoria in Plant Samples

Background Bacterial spot-causing xanthomonads (BSX) are quarantine phytopathogenic bacteria responsible for heavy losses in tomato and pepper production. Despite the research on improved plant spraying methods and resistant cultivars, the use of healthy plant material is still considered as the most effective bacterial spot control measure. Therefore, rapid and efficient detection methods are crucial for an early detection of these phytopathogens. Methodology In this work, we selected and validated novel DNA markers for reliable detection of the BSX Xanthomonas euvesicatoria (Xeu). Xeu-specific DNA regions were selected using two online applications, CUPID and Insignia. Furthermore, to facilitate the selection of putative DNA markers, a customized C program was designed to retrieve the regions outputted by both databases. The in silico validation was further extended in order to provide an insight on the origin of these Xeu-specific regions by assessing chromosomal location, GC content, codon usage and synteny analyses. Primer-pairs were designed for amplification of those regions and the PCR validation assays showed that most primers allowed for positive amplification with different Xeu strains. The obtained amplicons were labeled and used as probes in dot blot assays, which allowed testing the probes against a collection of 12 non-BSX Xanthomonas and 23 other phytopathogenic bacteria. These assays confirmed the specificity of the selected DNA markers. Finally, we designed and tested a duplex PCR assay and an inverted dot blot platform for culture-independent detection of Xeu in infected plants. Significance This study details a selection strategy able to provide a large number of Xeu-specific DNA markers. As demonstrated, the selected markers can detect Xeu in infected plants both by PCR and by hybridization-based assays coupled with automatic data analysis. Furthermore, this work is a contribution to implement more efficient DNA-based methods of bacterial diagnostics.

Identification of Xanthomonas fragariae, Xanthomonas axonopodis pv. phaseoli, and Xanthomonas fuscans subsp. fuscans with Novel Markers and Using a Dot Blot Platform Coupled with Automatic Data Analysis

ABSTRACT Phytosanitary regulations and the provision of plant health certificates still rely mainly on long and laborious culture-based methods of diagnosis, which are frequently inconclusive. DNA-based methods of detection can circumvent many of the limitations of currently used screening methods, allowing a fast and accurate monitoring of samples. The genus Xanthomonas includes 13 phytopathogenic quarantine organisms for which improved methods of diagnosis are needed. In this work, we propose 21 new Xanthomonas-specific molecular markers, within loci coding for Xanthomonas-specific protein domains, useful for DNA-based methods of identification of xanthomonads. The specificity of these markers was assessed by a dot blot hybridization array using 23 non-Xanthomonas species, mostly soil dwelling and/or phytopathogens for the same host plants. In addition, the validation of these markers on 15 Xanthomonas spp. suggested species-specific hybridization patterns, which allowed discrimination among the different Xanthomonas species. Having in mind that DNA-based methods of diagnosis are particularly hampered for unsequenced species, namely, Xanthomonas fragariae, Xanthomonas axonopodis pv. phaseoli, and Xanthomonas fuscans subsp. fuscans, for which comparative genomics tools to search for DNA signatures are not yet applicable, emphasis was given to the selection of informative markers able to identify X. fragariae, X. axonopodis pv. phaseoli, and X. fuscans subsp. fuscans strains. In order to avoid inconsistencies due to operator-dependent interpretation of dot blot data, an image-processing algorithm was developed to analyze automatically the dot blot patterns. Ultimately, the proposed markers and the dot blot platform, coupled with automatic data analyses, have the potential to foster a thorough monitoring of phytopathogenic xanthomonads.

Automatic detection of molecular markers in digital images

Labeled molecular markers are an important tool in molecular biology. This work presents a method for the automatic identification of molecular markers in dot blot images. The method detects the location of markers in the image and their size. An experiment was made with 6 test images, which were used to produce an additional set of 222 images with various rotation, translation, contrast and noise levels. Over 7500 markers were identified automatically and compared with reference values obtained manually. The RMS error for the marker positioning in the original test images were between 1.1 and 3.8 pixels, which is about 1/10 of the typical radius (26 pixels). The method proposed was found to be almost insensitive to grid rotation and translation, and reasonably robust to image contrast changes and presence of noise.

An automatic Method to identify and extract information of DNA bands in Gel Electrophoresis images

This paper presents a system for the automatic processing of Digital Images obtained from Gel Electrophoresis. The system identifies automatically the number and the location of lanes in the digital image, as well as the location of bands on each lane, without any intervention from the user. A reference lane with a know substance is used to compute the molecular weight of the observed (unknown) bands. The system performance was tested using 12 images, obtained from 4 gels with 3 different exposures. A total of 5443 bands were tested in 12 images, 672 reference / observed lane pairs. The average error in the estimation of molecular weight of 9.2%.

A system for automatic counting the number of collembola individuals on petri disk images

This paper describes an image processing system developed for automatic counting the number of collembola individuals on petri disks images. The system uses image segmentation and mathematical morphology techniques to identify and count the number of collembolans. The main challenges are the specular reflections at the edges of the circular samples and the foam present in a number of samples. The specular reflections are efficiently identified and removed by performing a two-stage segmentation. The foam is considered to be noise, as it is at cases difficult to discriminate between the foam and the collembola individuals. Morphological image processing tools are used both for noise reduction and for the identification of the collembolans. A total of 38 samples (divided in 3 groups according to their noise level) were tested and the results produced from the automatic system compared to the values available from manual counting. The relative error was on average 5.0% (3.4% for good quality samples, 4.6% for medium quality and 7.5% for poor quality samples).

Automatic Analysis of Dot Blot Images

This paper presents a method for the automatic analysis of macroarray (dot blot) images. The system developed receives as input a dot blot image, corrects it for grid rotation, identifies the visible markers and provides an evaluation of the status of each marker (ON/OFF). Two experiments were carried out to evaluate the detection and classification stages. A total of 222 test images were produced from 6 original dot blot images, with various rotations, translations, contrast and noise level. Over 7500 markers were identified automatically and compared to manual reference. The RMS error in positioning the molecular marker center was between 1.1 and 3.8 pixels and the marker radius error less than 4%. The automatic classification of markers (ON/OFF) was compared to the classification by 3 human experts, using 10 test images. The overall accuracy evaluated on 5118 markers was 94.0%. For those markers that had the same evaluation by all 3 experts, the classification accuracies were 96.6% (ON) and 95.9% (OFF).

Automatic information extraction from gel electrophoresis images using GEIAS

This paper presents a method (GEIAS) for the automatic processing of digital images obtained from Gel Electrophoresis. The performance of GEIAS was tested using 12 images, obtained from 4 gels with 3 different exposures with a total of 1082 bands, comparing the results provided by GEIAS and 3 other software tools. The GEIAS is able to fully automatically detect DNA lanes while the other 3 software tools tested can only do this in a semi-automatic or manual way. For the correct location of DNA bands, GEIAS required a manual correction of the location in 10.0% of the bands, and the other software tools 13.0%, 15.0% and 25.4%. The average error in the estimation of molecular weight was tested using a total of 5443 bands in 12 image using 672 reference/observed lane pairs. The average error was found to be 9.2% for GEIAS and 11.2%, 14.4% and 13.1% for the other software tools tested.

Automatic analysis of macroarrays images

The analysis of dot blot (macroarray) images is currently based on the human identification of positive/negative dots, which is a subjective and time consuming process. This paper presents a system for the automatic analysis of dot blot images, using a pre-defined grid of markers, including a number of ON and OFF controls. The geometric deformations of the input image are corrected, and the individual markers detected, both tasks fully automatically. Based on a previous training stage, the probability for each marker to be ON is established. This information is provided together with quality parameters for training, noise and classification, allowing for a fully automatic evaluation of a dot blot image.

Automatic counting the number of Collembola in digital images

Counting the number of Collembola in digital images is a routine task in laboratories of soil ecotoxicology. This process is based on a direct visual identification of Collembola, and is consequently a time consuming task. This paper present a fully automatic system for counting the number of Collembola in digital images. The system selects the interest area of the image, detects and removes the specular reflection of the incident light, as well as the foam developed during laboratory experiment and finally identifies and counts the number of Collembola. The system performance was tested using 5 treatments with 9 or 10 replicates and 13 treatments with 4 or 5 replicates. A total of 111 images were tested and the results were compared with those obtained by manual identification. The average relative error between automatic and manual counts from multiple observations of the same treatment was 2:1%, which can be considered a good result, given that this value is below the standard deviation between multiple replicate counts.