Nearest-neighbor classification for identification of aggressive versus nonaggressive low-grade astrocytic tumors by means of image cytometry-generated variables

Semi-Automatic Oil Spill Detection on X-Band Marine Radar Images Using Texture Analysis, Machine Learning, and Adaptive Thresholding

Oil spills bring great damage to the environment and, in particular, to coastal ecosystems. The ability of identifying them accurately is important to prompt oil spill response. We propose a semi-automatic oil spill detection method, where texture analysis, machine learning, and adaptive thresholding are used to process X-band marine radar images. Coordinate transformation and noise reduction are first applied to the sampled radar images, coarse measurements of oil spills are then subjected to texture analysis and machine learning. To identify the loci of oil spills, a texture index calculated by four textural features of a grey level co-occurrence matrix is proposed. Machine learning methods, namely support vector machine, k-nearest neighbor, linear discriminant analysis, and ensemble learning are adopted to extract the coarse oil spill areas indicated by the texture index. Finally, fine measurements can be obtained by using adaptive thresholding on coarsely extracted oil spill areas. Fine measurements are insensitive to the results of coarse measurement. The proposed oil spill detection method was used on radar images that were sampled after an oil spill accident that occurred in the coastal region of Dalian, China on 21 July 2010. Using our processing method, thresholds do not have to be set manually and oil spills can be extracted semi-automatically. The extracted oil spills are accurate and consistent with visual interpretation.

Brain tumor classification using AFM in combination with data mining techniques

Although classification of astrocytic tumors is standardized by the WHO grading system, which is mainly based on microscopy-derived, histomorphological features, there is great interobserver variability. The main causes are thought to be the complexity of morphological details varying from tumor to tumor and from patient to patient, variations in the technical histopathological procedures like staining protocols, and finally the individual experience of the diagnosing pathologist. Thus, to raise astrocytoma grading to a more objective standard, this paper proposes a methodology based on atomic force microscopy (AFM) derived images made from histopathological samples in combination with data mining techniques. By comparing AFM images with corresponding light microscopy images of the same area, the progressive formation of cavities due to cell necrosis was identified as a typical morphological marker for a computer-assisted analysis. Using genetic programming as a tool for feature analysis, a best model was created that achieved 94.74% classification accuracy in distinguishing grade II tumors from grade IV ones. While utilizing modern image analysis techniques, AFM may become an important tool in astrocytic tumor diagnosis. By this way patients suffering from grade II tumors are identified unambiguously, having a less risk for malignant transformation. They would benefit from early adjuvant therapies.

Improving accuracy in astrocytomas grading by integrating a robust least squares mapping driven support vector machine classifier into a two level grade classification scheme

Grading of astrocytomas is an important task for treatment planning; however, it suffers from significantly great inter-observer variability. Computer-assisted diagnosis systems have been propose to assist towards minimizing subjectivity, however, these systems present either moderate accuracy or utilize specialized staining protocols and grading systems that are difficult to apply in daily clinical practice. The present study proposes a robust mathematical formulation by integrating state-of-art technologies (support vector machines and least squares mapping) in a cascade classification scheme for separating low from high and grade III from grade IV astrocytic tumours. Results have indicated that low from high-grade tumours can be correctly separated with a certainty as high as 97.3%, whereas grade III from grade IV tumours with 97.8%. The overall performance was 95.2%. These high rates have been a result of applying the least squares mapping technique to features prior to classification. A significant byproduct of least squares mapping is that the number of support vectors of the SVM classifiers dropped dramatically from about 80% when no mapping was used to less than 5% when mapping was used. The latter is a clear indication that the SVM classifier has a greater potential to generalize well to new data. In this way, digital image analysis systems for automated grading of astrocytomas are brought closer to clinical practice.

An image-analysis system based on support vector machines for automatic grade diagnosis of brain-tumour astrocytomas in clinical routine

An image-analysis system based on the concept of Support Vector Machines (SVM) was developed to assist in grade diagnosis of brain tumour astrocytomas in clinical routine. One hundred and forty biopsies of astrocytomas were characterized according to the WHO system as grade II, III and IV. Images from biopsies were digitized, and cell nuclei regions were automatically detected by encoding texture variations in a set of wavelet, autocorrelation and parzen estimated descriptors and using an unsupervised SVM clustering methodology. Based on morphological and textural nuclear features, a decision-tree classification scheme distinguished between different grades of tumours employing an SVM classifier. The system was validated for clinical material collected from two different hospitals. On average, the SVM clustering algorithm correctly identified and accurately delineated 95% of all nuclei. Low-grade tumours were distinguished from high-grade tumours with an accuracy of 90.2% and grade III from grade IV with an accuracy of 88.3% The system was tested in a new clinical data set, and the classification rates were 87.5 and 83.8%, respectively. Segmentation and classification results are very encouraging, considering that the method was developed based on every-day clinical standards. The proposed methodology might be used in parallel with conventional grading to support the regular diagnostic procedure and reduce subjectivity in astrocytomas grading.

Supratentorial pilocytic astrocytomas, astrocytomas, anaplastic astrocytomas and glioblastomas are characterized by a differential expression of S100 proteins

The levels of expression of the S100A1, S100A2, S100A3, S100A4, S100A5, S100A6 and S100B proteins were immunohistochemically assayed and quantitatively determined in a series of 95 astrocytic tumors including 26 World Health Organization (WHO) grade I (pilocytic astrocytomas), 23 WHO grade II (astrocytomas), 25 WHO grade III (anaplastic astrocytomas) and 21 WHO grade IV (glioblastomas) cases. The level of the immunohistochemical expression of the S100 proteins was quantitatively determined in the solid tumor tissue (tumor mass). In addition twenty blood vessel walls and their corresponding perivascular tumor astrocytes were also immunohistochemically assayed for 10 cases chosen at random from each of the four histopathological groups. The data showed modifications in the level of S100A3 protein expression; these modifications clearly identified the pilocytic astrocytomas from WHO grade II-IV astrocytic tumors as a distinct biological group. Modifications in the level of S100A6 protein expression enabled a clear distinction to be made between low (WHO grade I and II) and high (WHO grade III and IV) grade astrocytic tumors. Very significant modifications occurred in the level of S100A1 protein expression (and, to a lesser extent, in their of the S100A4 and S100B proteins) in relation to the increasing levels of malignancy. While the S100A5 protein was significantly expressed in all the astrocytic tumors (but without any significant modifications in the levels of malignancy), the S100A2 protein was never expressed in these tumors. These data thus indicate that several S100 proteins play major biological roles in human astrocytic tumors.

Galectins are differentially expressed in supratentorial pilocytic astrocytomas, astrocytomas, anaplastic astrocytomas and glioblastomas, and significantly modulate tumor astrocyte migration

Galectins, a family of mammalian lectins with specificity to beta-galactosides, are involved in growth-regulatory mechanisms and cell adhesion. A relationship is assumed to exist between the levels of expression of galectins and the level of malignancy in human gliomas. A comparative study of this aspect in the same series of clinical samples is required to prove this hypothesis. Using computer-assisted microscopy, we quantitatively characterized by immunohistochemistry the levels of expression of galectins-1, -3 and -8 in 116 human astrocytic tumors of grades I to IV. Extent of transcription of galectins-1, -3, and -8 genes was investigated in 8 human glioblastoma cell lines by means of RT-PCR techniques. Three of these cell lines were grafted into the brains of nude mice in order to characterize in vivo the galectins-1, -3 and -8 expression in relation to the patterns of the tumor invasion of the brain. The role of galectin-1, -3 and -8 in tumor astrocyte migration was quantitatively determined in vitro by means of computer-assisted phase-contrast videomicroscopy. The data indicate that the levels of galectin-1 and galectin-3 expression significantly change during the progression of malignancy in human astrocytic tumors, while that of galectin-8 remains unchanged. These three galectins are involved in tumor astrocyte invasion of the brain parenchyma since their levels of expression are higher in the invasive parts of xenografted glioblastomas than in their less invasive parts. Galectin-3, galectin-1, and to a lesser extent galectin-8, markedly stimulate glioblastoma cell migration in vitro. Since bands for the transcripts of human galectins-2, -4 and -9 were apparently less frequent and intense in the 8 human glioblastoma cell lines, this system provides an excellent model to assign defined roles to individual galectins and delineate overlapping and distinct functional aspects.

Improving morphology-based malignancy grading schemes in astrocytic tumors by means of computer-assisted techniques

We propose an original methodology which improves the accuracy of the prognostic values associated with conventional morphologically-based classifications in supratentorial astrocytic tumors in the adult. This methodology may well help neuropathologists, who must determine the aggressiveness of astrocytic tumors on the basis of morphological criteria. The proposed methodology comprises two distinct steps, i.e. i) the production of descriptive quantitative variables (related to DNA ploidy level and morphonuclear aspects) by means of computer-assisted microscopy and ii) data analysis based on an artificial intelligence-related method, i.e. the decision tree approach. Three prognostic problems were considered on a series of 250 astrocytic tumors including 39 astrocytomas (AST), 47 anaplastic astrocytomas (ANA) and 164 glioblastomas (GBM) identified in accordance with the WHO classification. These three problems concern i) variations in the aggressiveness level of the high-grade tumors (ANA and GBM), ii) the detection of the aggressive as opposed to the less aggressive low-grade astrocytomas (AST), and iii) the detection of the aggressive as opposed to the less aggressive anaplastic astrocytomas (ANA). Our results show that the proposed computer-aided methodology improves conventional prognosis based on conventional morphologically-based classifications. In particular, this methodology enables some reference points to be established on the biological continuum according to the sequence AST-->ANA-->GBM.

Automated diagnosis of brain tumours astrocytomas using probabilistic neural network clustering and support vector machines

A computer-aided diagnosis system was developed for assisting brain astrocytomas malignancy grading. Microscopy images from 140 astrocytic biopsies were digitized and cell nuclei were automatically segmented using a Probabilistic Neural Network pixel-based clustering algorithm. A decision tree classification scheme was constructed to discriminate low, intermediate and high-grade tumours by analyzing nuclear features extracted from segmented nuclei with a Support Vector Machine classifier. Nuclei were segmented with an average accuracy of 86.5%. Low, intermediate, and high-grade tumours were identified with 95%, 88.3%, and 91% accuracies respectively. The proposed algorithm could be used as a second opinion tool for the histopathologists.

Molecular characterization of cell substratum attachments in human glial tumors relates to prognostic features

Glioma cell attachments to substratum play crucial roles in the invasion by glioma cells of normal brain tissue. These attachments are mediated through interactions between extracellular matrix (ECM) components, integrins, focal adhesion-linked molecules, and the actin cytoskeleton. In the present study, we investigate the molecular elements involved in cell substratum attachments in human glial tumors and their potential relationships to prognostic features. We used 10 human glioma cell lines, for which we characterized glial differentiation by means of quantitative RT-PCR for nestin, vimentin, and GFAP mRNA. We quantitatively determined the amounts of laminin, fibronectin, vitronectin, and thrombospondin secreted by these glioma cell lines in vitro, as well as the amount of each of the eight beta integrin subunits and the adhesion complex-related molecules, including talin, vinculin, profilin, zyxin, alpha-actinin, paxillin, and VASP. After quantification of the levels of migration and invasion of these 10 cell lines in vitro and, through grafts into the brains of nude mice, of their biological aggressiveness in vivo, it appeared that the levels of the beta 5 integrin subunit and alpha-actinin were directly related to biological aggressiveness. These experimental data were clinically confirmed because increasing immunohistochemical amounts of the beta 5 integrin subunit and alpha-actinin were directly related to dismal prognoses in the case of astrocytic tumors. In addition, we show that the beta 4 integrin subunit are expressed significantly more in oligodendrogliomas than in astrocytic tumors. A potential role for the beta 8 integrin subunit in glioma cell substratum attachments is also emphasized.

A comparison of machine learning methods for the diagnosis of pigmented skin lesions

We analyze the discriminatory power of k-nearest neighbors, logistic regression, artificial neural networks (ANNs), decision tress, and support vector machines (SVMs) on the task of classifying pigmented skin lesions as common nevi, dysplastic nevi, or melanoma. Three different classification tasks were used as benchmarks: the dichotomous problem of distinguishing common nevi from dysplastic nevi and melanoma, the dichotomous problem of distinguishing melanoma from common and dysplastic nevi, and the trichotomous problem of correctly distinguishing all three classes. Using ROC analysis to measure the discriminatory power of the methods shows that excellent results for specific classification problems in the domain of pigmented skin lesions can be achieved with machine-learning methods. On both dichotomous and trichotomous tasks, logistic regression, ANNs, and SVMs performed on about the same level, with k-nearest neighbors and decision trees performing worse.

Grading of diffusely infiltrating astrocytomas by quantitative histopathology, cell proliferation and image cytometric DNA analysis. Comparison of 133 tumours in the context of the WHO 1979 and WHO 1993 grading schemes

The aim of the study was to evaluate the applicability of quantitative histopathology as an aid for grading diffusely infiltrating astrocytomas. Primary astrocytomas were analysed for parameters (mean nuclear size, mitosis count, area fraction of endothelial cells and tumour necrosis, area fraction of nuclei, and Ki-67 (MIB-1) labelling index), which are closely related to the World Health Organization (WHO) 1979 and WHO 1993 grading criteria. All estimates correlated with the WHO histopathological grade and patient outcome. According to the receiver-operating characteristics curve, the presence of tumour necrosis and mitosis count (cut-off at 3 mitoses/mm2 of neoplastic tissue) showed the best sensitivity and specificity in separating patients with different survival. The multivariate survival analyses confirmed this result. A decision-tree model was constructed based on these two variables: twig I with less than 3 mitoses/mm2, twig II with equal or more than 3 mitoses/mm2 but no necrosis, and twig III with tumour necrosis. This model was found to be more strongly associated with survival than the WHO 1979 or WHO 1993 grading schemes. Low-malignancy astrocytomas (WHO grade II or twig I tumours) could be further divided into two prognostic categories by the image cytometric DNA analysis. The results put an emphasis on astrocytoma grading on mitosis counts (grade II vs. III) and tumour necrosis (grade III vs. IV). To standardize the sampling for mitosis counting, it is suggested that a parallel Ki-67 immunostaining be used for the identification of the most proliferative areas.

Differential expression of S100 calcium-binding proteins characterizes distinct clinical entities in both WHO grade II and III astrocytic tumours

The computer-assisted microscopic analysis of Feulgen-stained nuclei enabled us to identify two subgroups of astrocytomas (WHO grade II) and two subgroups of anaplastic astrocytomas (WHO grade III) with significantly distinct clinical outcomes (Decaestecker et al. Brain Pathol 1998; 8: 29-38). The astrocytomas labelled in the present study as typical (TYP-ASTs) behaved clinically like real astrocytomas while atypical astrocytomas (ATYP-ASTs) behaved similarly to anaplastic astrocytomas. The anaplastic astrocytomas that we labelled as typical (TYP-ANAs) behaved clinically like anaplastic astrocytomas while atypical ones (ATYP-ANAs) behaved like glioblastomas. In the present study, we investigate whether some biological characteristics could be evidenced across these four groups of TYP- and ATYP-ASTs and TYP- and ATYP-ANAs. The data show that the levels of expression (immunohistochemically assayed and quantitatively determined by means of computer-assisted microscopy) of vimentin, the glial fibrillary acidic protein and the platelet-derived growth factor-alpha did not differ significantly across these four groups of astrocytic tumours. The level of cell proliferation (determined by means of both the anti-proliferating cell nuclear antigen and the anti-MIB-1 antibodies; P < 0.001 to P < 0.0001) differed very significantly between the astrocytomas and anaplastic astrocytomas, but not between the typical and atypical variants identified in each group. In sharp contrast, the levels of expression of the S100A3 and S100A5 proteins differed markedly in the solid tumour tissue in relation to the astrocytic tumour types and grades. In addition, while the levels of expression of S100A6 did not change in the astrocytic tumour tissue in relation to histopathological grade, the levels of expression of this S100 protein (but not those of S100A3 and S100A5) differed markedly in the blood vessel walls according to whether these vessels originated from low- or high-grade astrocytic tumours.

Pediatric low-grade gliomas: prognosis with proton magnetic resonance spectroscopic imaging

OBJECTIVE

Our aim was to assess the correlation between the low-grade glioma (LGG) metabolic profile and tumor progression. Using in vivo proton magnetic resonance spectroscopic imaging, we specifically asked whether and which metabolic features are associated with tumor regrowth or recurrence.

METHODS

Eleven pediatric patients with histologically proven partially resected (<20% resection) midline LGG were treated and followed up for a period of 2 years. All patients underwent proton magnetic resonance spectroscopic imaging studies before any management was determined. Tumor progression was defined as radiological evidence of mass enlargement (>25%) during the follow-up period. Proton magnetic resonance spectroscopic imaging was performed using a PRESS-CSI sequence on a General Electric 1.5-tesla scanner (General Electric Medical System, Waukesha, WI). The signal intensities of N-acetylaspartate, choline (CHO), and creatine from the tumor and the normal brain were used to calculate normalized metabolite intensities and metabolite ratios.

RESULTS

Tumors that progressed during a 2-year period displayed higher normalized CHO than those that remained stable (Mann-Whitney test, P < 0.03). The majority (five of six) of the rapidly growing LGG showed values of normalized CHO of at least 1, whereas the nonprogressors had a normalized CHO value of less than 1.

CONCLUSION

In association with pediatric LGG, high normalized CHO values seem to herald the potential for rapid tumor growth. These observations may be valuable for defining subsets of patients with LGG who may benefit from early therapeutic interventions.