Morphological Pattern Recognition Research Articles

Leveraging mid-infrared spectroscopic imaging and deep learning for tissue subtype classification in ovarian cancer.

Mid-infrared spectroscopic imaging (MIRSI) is an emerging class of label-free techniques being leveraged for digital histopathology. Modern histopathologic identification of ovarian cancer involves tissue staining followed by morphological pattern recognition. This process is time-consuming and subjective and requires extensive expertise. This paper presents the first label-free, quantitative, and automated histological recognition of ovarian tissue subtypes using a new MIRSI technique. This optical photothermal infrared (O-PTIR) imaging technique provides a 10× enhancement in spatial resolution relative to prior instruments. It enables sub-cellular spectroscopic investigation of tissue at biochemically important fingerprint wavelengths. We demonstrate that the enhanced resolution of sub-cellular features, combined with spectroscopic information, enables reliable classification of ovarian cell subtypes achieving a classification accuracy of 0.98. Moreover, we present a statistically robust analysis from 78 patient samples with over 60 million data points. We show that sub-cellular resolution from five wavenumbers is sufficient to outperform state-of-the-art diffraction-limited techniques with up to 235 wavenumbers. We also propose two quantitative biomarkers based on the relative quantities of epithelia and stroma that exhibit efficacy in early cancer diagnosis. This paper demonstrates that combining deep learning with intrinsic biochemical MIRSI measurements enables quantitative evaluation of cancerous tissue, improving the rigor and reproducibility of histopathology.

Epithelioid, biphasic and mixed tumors of soft tissue

Epithelioid soft tissue tumors consist exclusively of epithelioid tumor cells. Biphasic tumors are composed of both aspindle-cell and an epithelioid component. The rare mixed tumors of soft tissue show abroader variety of cellular and stromal differentiation but also include at least one, possibly several, epithelioid portions.The close morphological similarity of some of these entities with each other, as well as with the more frequent soft tissue metastases of carcinomas, carcinosarcomas, and melanomas, to malignant mesothelioma and certain lymphomas, can often make the correct diagnosis extremely difficult. Recent advances in the detection of certain molecular alterations (mostly chromosomal translocations) have contributed to changes in tumor classification but also to improved pathological diagnostics (e.g. through the development of potent diagnostic antibodies) and biological understanding.The present overview should help the pathologist in the diagnosis of these rare tumors through the classical approach of morphological pattern recognition. The most important entities are discussed and illustrated in more detail, with the incorporation of the latest immunohistochemical and molecular aspects and the differential diagnosis of similar tumors.

An Adaptive Homomorphic Aperture Photometry Algorithm for Merging Galaxies

We present a novel automatic adaptive aperture photometry algorithm for measuring the total magnitudes of merging galaxies with irregular shapes. First, we use a morphological pattern recognition routine for identifying the shape of an irregular source in a background-subtracted image. Then, we extend the shape of the source by using the Dilation image operation to obtain an aperture that is quasi-homomorphic to the shape of the irregular source. The magnitude measured from the homomorphic aperture would thus have minimal contamination from the nearby background. As a test of our algorithm, we applied our technique to the merging galaxies observed by the Sloan Digital Sky Survey and the Canada–France–Hawaii Telescope. Our results suggest that the adaptive homomorphic aperture algorithm can be very useful for investigating extended sources with irregular shapes and sources in crowded regions.

Myeloproliferative neoplasms: Morphology and clinical practice.

In myeloproliferative neoplasms (MPNs), controversy persists regarding the usefulness and reproducibility of bone marrow (BM) features. Disagreements concerning the WHO classification are mainly focused on the discrimination between essential thrombocythemia (ET) and prefibrotic/early primary myelofibrosis (prePMF) and prodromal polycythemia vera (PV). Criticism mostly refers to lack of standardization of distinctive BM features precluding correct morphological pattern recognition. The distinction between WHO-defined ET and prePMF is not trivial because outcome is significantly worse in prePMF. Morphology was generally considered to be non-specific for the diagnosis of PV. Recent studies have revealed under-diagnosis of morphologically and biologically consistent PV.

Role of Gray Scale, Color Doppler and Spectral Doppler in Differentiation Between Malignant and Benign Thyroid Nodules.

High resolution ultrasound is the most sensitive imaging test available for the examination of the thyroid gland and due to increase in use of ultrasound more incidental thyroid nodules are diagnosed. In this study we try to establish the specific grayscale, color and spectral Doppler characteristics of malignant and benign thyroid nodules. To determine the specific gray scale characteristics, angioarchitecture and cut-off values of Doppler indices of malignant and benign thyroid nodules. To assess the efficacy of grayscale, Doppler and combined conventional and Doppler using defined criteria in differentiating malignant from benign nodules. We prospectively examined 194 thyroid nodules which were confirmed on FNAC. Each nodule was described according to size, number, contents, echogenicity, margins, halo, shape, calcification, local infiltration and lymphnode enlargement. Vascularity, RI and PI values of each nodule were assessed on Doppler. Each nodule was characterized as benign, indeterminate or malignant based on grayscale and Doppler characteristics. Cut-off RI and PI values for malignant thyroid nodules were obtained by ROC. Out of 194 nodules, 151 nodules were benign and 43 nodules were malignant. Significant relationship was observed between malignancy and hypoechogenicity, irregular margins, taller than wide, thick incomplete halo, micro calcifications, lymphnode enlargement and local infiltration. Intranodular vascularity was a significant criterion to suggest malignancy in thyroid nodules on color Doppler. Malignant nodules had a mean RI of 0.73 and mean PI of 1.3 which were significantly higher than the benign nodules. Accuracy of detecting malignant thyroid nodules by combining gray scale and Doppler is higher than either of them alone. Using specific morphological pattern recognition features like microcalcifications, hypoechogenicity, taller than wide, irregular thick halo, lymphadenopathy and local extra thyroidal invasion has helped in differentiating malignant from benign features. Intranodular vascularity and high RI indices were the specific Doppler signs for malignant thyroid nodules. Since, Gray scale and Doppler have their own strengths and weaknesses, they were complementary rather than competitive modalities in diagnosing benign from malignant thyroid nodules.

Automated prognostic pattern detection shows favourable diffuse pattern of FOXP3(+) Tregs in follicular lymphoma.

Background:Histopathological prognostication relies on morphological pattern recognition, but as numbers of biomarkers increase, human prognostic pattern-recognition ability decreases. Follicular lymphoma (FL) has a variable outcome, partly determined by FOXP3 Tregs. We have developed an automated method, hypothesised interaction distribution (HID) analysis, to analyse spatial patterns of multiple biomarkers which we have applied to tumour-infiltrating lymphocytes in FL.Methods:A tissue microarray of 40 patient samples was used in triplex immunohistochemistry for FOXP3, CD3 and CD69, and multispectral imaging used to determine the numbers and locations of CD3+, FOXP3/CD3+ and CD69/CD3+ T cells. HID analysis was used to identify associations between cellular pattern and outcome.Results:Higher numbers of CD3+ (P=0.0001), FOXP3/CD3+ (P=0.0031) and CD69/CD3+ (P=0.0006) cells were favourable. Cross-validated HID analysis of cell pattern identified patient subgroups with statistically significantly different survival (35.5 vs 142 months, P=0.00255), a more diffuse pattern associated with favourable outcome and an aggregated pattern with unfavourable outcome.Conclusions:A diffuse pattern of FOXP3 and CD69 positivity was favourable, demonstrating ability of HID analysis to automatically identify prognostic cellular patterns. It is applicable to large numbers of biomarkers, representing an unsupervised, automated method for identification of undiscovered prognostic cellular patterns in cancer tissue samples.

Automated recognition of urban objects for morphological urban analysis

Abstract It has often been suggested that the physical structure of traditional urban settlements represents a time-tested response to the climatic constraints of a given location, particularly in hot-arid or other regions that are characterized by thermally harsh conditions. To test this premise, vernacular urban forms have been analyzed to identify patterns of built morphology which might have developed as a response to prevailing climatic conditions. This approach is often based on visual analysis and subjective and qualitative recognition of climate-related patterns. The current paper presents an alternative approach, using a quantitative methodology to support the systematic and automated recognition of urban morphological patterns. A parametric GIS model was developed for automated recognition of urban objects, employing the ability of GIS to recognize contextual relations among objects and perform a context-based recognition using basic rules that apply to the organization of urban form. The model was statistically tested for its accuracy with satisfying results. A case study is presented to demonstrate the workability of the model and its potential as a quantitative and systematic tool for quantitatively describing the morphology of urban surfaces for urban climate analysis. As a parametric model, it can be modified and applied to a large number of case studies for developing and updating databases of urban morphology, and these can serve as a basis for the recognition of morphological patterns in urban form.

Automatic determination of L/H transition times in DIII-D through a collaborative distributed environment

An automatic predictor of L/H transition times has been implemented for the DIII-D tokamak. The system predicts the transition combining two techniques: A morphological pattern recognition algorithm, which estimates the transition based on the waveform of a Dα emission signal, and a support vector machines multi-layer model, which predicts the L/H transition using a non-parametric model. The predictor is employed within a collaborative distributed computing environment. The system is trained remotely in the Ciemat computer cluster and operated on the DIII-D site.

A CATALOG OF MORPHOLOGICALLY IDENTIFIED MERGING GALAXIES

We present a new catalog and images of interacting and merging galaxies obtained with morphological identification from Canada-France-Hawaii Telescope observations. We first used a morphological pattern recognition searching routine to select possible candidates, and then visually inspected the images of these candidates to identify true merging/interacting sources. We discover 15,147 new pairs of merging galaxies within 422 deg{sup 2} of Red Sequence Cluster Survey 2. Furthermore, we also find nine new candidates of galaxy clusters by searching for regions with significant density enhancements of merging galaxies. This catalog has the largest number of morphologically identified interacting and merging galaxies based on consistent searching criteria. These sources provide a uniform sample of merging galaxies for further photometric and spectroscopic studies of galaxy evolutio000.

A CATALOG OF MORPHOLOGICALLY IDENTIFIED MERGING GALAXIES

We present a new catalog and images of interacting and merging galaxies obtained with morphological identification from Canada-France-Hawaii Telescope observations. We first used a morphological pattern recognition searching routine to select possible candidates, and then visually inspected the images of these candidates to identify true merging/interacting sources. We discover 15,147 new pairs of merging galaxies within 422 deg2 of Red Sequence Cluster Survey 2. Furthermore, we also find nine new candidates of galaxy clusters by searching for regions with significant density enhancements of merging galaxies. This catalog has the largest number of morphologically identified interacting and merging galaxies based on consistent searching criteria. These sources provide a uniform sample of merging galaxies for further photometric and spectroscopic studies of galaxy evolution.

Intelligent technique to search for patterns within images in massive databases

An image retrieval system for JET has been developed. The image database contains the images of the JET high speed visible camera. The system input is a pattern selected inside an image and the output is the group of frames (defined by their discharge numbers and time slices) that show patterns similar to the selected one. This approach is based on morphological pattern recognition and it should be emphasized that the pattern is found independently of its location in the frame. The technique encodes images into characters and, therefore, it transforms the pattern search into a character-matching problem.

Infrared spectroscopic imaging for histopathologic recognition

The process of histopathology, comprising tissue staining and morphological pattern recognition, has remained largely unchanged for over 140 years. Although it is integral to clinical and research activities, histopathologic recognition remains a time-consuming, subjective process to which only limited statistical confidence can be assigned because of inherent operator variability. Although immunohistochemical approaches allow limited molecular detection, significant challenges remain in using them for quantitative, automated pathology. Vibrational spectroscopic approaches, by contrast, directly provide nonperturbing molecular descriptors, but a practical spectroscopic protocol for histopathology is lacking. Here we couple high-throughput Fourier transform infrared (FTIR) spectroscopic imaging of tissue microarrays with statistical pattern recognition of spectra indicative of endogenous molecular composition and demonstrate histopathologic characterization of prostatic tissue. This automated histologic segmentation is applied to routine archival tissue samples, incorporates well-defined tests of statistical significance and eliminates any requirement for dyes or molecular probes. Finally, we differentiate benign from malignant prostatic epithelium by spectroscopic analyses.

On the relation of order-statistics filters and template matching: optimal morphological pattern recognition

In this paper, we investigate methods for optimal morphological pattern recognition. The task of optimal pattern recognition is posed as a solution to a hypothesis testing problem. A minimum probability of error decision rule-maximum a posteriori filter-is sought. The classical solution to the minimum probability of error hypothesis testing problem, in the presence of independent and identically distributed noise degradation, is provided by template matching (TM). A modification of this task, seeking a solution to the minimum probability of error hypothesis testing problem, in the presence of composite (mixed) independent and identically distributed noise degradation, is demonstrated to be given by weighted composite template matching (WCTM). As a consequence of our investigation, the relationship of the order-statistics filter (OSF) and TM-in both the standard as well as the weighted and composite implementations-is established. This relationship is based on the thresholded cross-correlation representation of the OSF. The optimal order and weights of the OSF for pattern recognition are subsequently derived. An additional outcome of this representation is a fast method for the implementation of the OSF.

Intradural extramedullary causes of myelopathy

This article emphasizes the utility of contrast-enhanced MRI in the detection of intradural extramedullary causes of myelopathy. Recognition of morphological patterns, signal intensities, and enhancement patterns frequently permits a specific diagnosis or a limited differential diagnosis.