Machine Learning-based Pattern Recognition Research Articles

Imaging with Raman photons: a novel use of mixed-mode spectroscopy

Surface enhanced Raman spectroscopy is today an established technique used for chemical fingerprinting. Here, we showcase an engineered hierarchical substrate, in which the plasmonically active regions, restricted to a micron scale, two dimensional hexagonal pattern are examined. Spatial variation of the enhanced Raman signal from any analyte, uniformly coating the substrate, consequently bears a high registry with the underlying pattern. This spatially contrasted enhancement allows optical imaging of the 2D pattern solely using the Raman scattered photons from the analyte. While the pattern brightness and contrast determine analyte identification and detection sensitivity, hyperspectral imaging can be exploited for increasing specificity. Proof of concept demonstration of the technique is carried out via the acquisition of Raman images with rhodamine and fluorescein dyes and then applied to detect glucose in 40 mM concentration. The large area optical imaging and the requirement of long-range uniformity in the detected patterns for positive analyte detection, is implemented using a machine learning based pattern recognition protocol which also increases the statistical confidence of detection. This simultaneous, large area signal detection sacrifices continuous spectral information at the cost of speed, reproducibility and minimising human error via automation of detection in the hyperspectral imaging technique presented here.

Pixel-Based Machine Learning and Image Reconstitution for Dot-ELISA Pathogen Diagnosis in Biological Samples.

Serological methods serve as a direct or indirect means of pathogen infection diagnosis in plant and animal species, including humans. Dot-ELISA (DE) is an inexpensive and sensitive, solid-state version of the microplate enzyme-linked immunosorbent assay, with a broad range of applications in epidemiology. Yet, its applicability is limited by uncertainties in the qualitative output of the assay due to overlapping dot colorations of positive and negative samples, stemming mainly from the inherent color discrimination thresholds of the human eye. Here, we report a novel approach for unambiguous DE output evaluation by applying machine learning-based pattern recognition of image pixels of the blot using an impartial predictive model rather than human judgment. Supervised machine learning was used to train a classifier algorithm through a built multivariate logistic regression model based on the RGB (“Red,” “Green,” “Blue”) pixel attributes of a scanned DE output of samples of known infection status to a model pathogen (Lettuce big-vein associated virus). Based on the trained and cross-validated algorithm, pixel probabilities of unknown samples could be predicted in scanned DE output images, which would then be reconstituted by pixels having probabilities above a cutoff. The cutoff may be selected at will to yield desirable false positive and false negative rates depending on the question at hand, thus allowing for proper dot classification of positive and negative samples and, hence, accurate diagnosis. Potential improvements and diagnostic applications of the proposed versatile method that translates unique pathogen antigens to the universal basic color language are discussed.

Cellular cartography of the organ of Corti based on optical tissue clearing and machine learning.

The highly organized spatial arrangement of sensory hair cells in the organ of Corti is essential for inner ear function. Here, we report a new analytical pipeline, based on optical clearing of tissue, for the construction of a single-cell resolution map of the organ of Corti. A sorbitol-based optical clearing method enabled imaging of the entire cochlea at subcellular resolution. High-fidelity detection and analysis of all hair cell positions along the entire longitudinal axis of the organ of Corti were performed automatically by machine learning-based pattern recognition. Application of this method to samples from young, adult, and noise-exposed mice extracted essential information regarding cellular pathology, including longitudinal and radial spatial characteristics of cell loss, implying that multiple mechanisms underlie clustered cell loss. Our method of cellular mapping is effective for system-level phenotyping of the organ of Corti under both physiological and pathological conditions.

The neuroanatomical pattern of sex and gender in cis/transgender individuals revealed by machine learning-based pattern recognition with NeuroMiner

The medial prefrontal cortex (mPFC) has reciprocal projections with many cerebral structures that are crucial in the control of food ingestion behavior and reward processing; Thus the mPFC has an important function in taste memory recognition. Previous results indicate that long-term consumption of sugar produces changes in appetitive re-learning and suggest that this could trigger an escalating consumption due to the inability to learn new negative consequences related to the same taste. Further evidence suggests that general identity reward value could be encoded in the mPFC. Therefore, the purpose of this study was to evaluate in rats whether after 21 days of sugar consumption the increase in sweet taste preference and latent inhibition of conditioned taste aversion (CTA) were affected differentially by pharmacological activation or blockage of dopaminergic and β-adrenergic receptors, in the mPFC, during CTA acquisition. Results showed that after long-term sugar exposure, mPFC activation of β-adrenergic receptors with clenbuterol delayed aversive memory extinction, but the blockade with propranolol or activation of dopaminergic receptors with apomorphine increased CTA latent inhibition and accelerated aversive memory extinction only after acute sugar exposure. Only dopaminergic blockade with haloperidol prevented sweet taste preference expression after long-term sugar consumption, increased CTA latent inhibition and accelerated extinction after acute sugar exposure. Taken together, the present data provide evidence that catecholaminergic receptors in the mPFC after prolonged sugar consumption underwent functional changes related to re-learning and new aversive taste learning.

A Novel Technique for Imaging and Analysis of Hair Cells in the Organ of Corti Using Modified Sca/eS and Machine Learning.

Here, we describe a sorbitol-based optical clearing method, called modified Sca/eS that can be used to image all hair cells (HCs) in the mouse cochlea. This modification of Sca/eS is defined by three steps: decalcification, de-lipidation, and refractive index matching, which can all be completed within 72 h. Furthermore, we established automated analysis programs that perform machine learning-based pattern recognition. These programs generate 1) a linearized image of HCs, 2) the coordinates of HCs, 3) a holocochleogram, and 4) clusters of HC loss. In summary, a novel approach that integrates modified Sca/eS and programs based on machine learning facilitates quantitative and comprehensive analysis of the physiological and pathological properties of all HCs.

A machine learning approach for automated wide-range frequency tagging analysis in embedded neuromonitoring systems.

EEG is a standard non-invasive technique used in neural disease diagnostics and neurosciences. Frequency-tagging is an increasingly popular experimental paradigm that efficiently tests brain function by measuring EEG responses to periodic stimulation. Recently, frequency-tagging paradigms have proven successful with low stimulation frequencies (0.5-6Hz), but the EEG signal is intrinsically noisy in this frequency range, requiring heavy signal processing and significant human intervention for response estimation. This limits the possibility to process the EEG on resource-constrained systems and to design smart EEG based devices for automated diagnostic. We propose an algorithm for artifact removal and automated detection of frequency tagging responses in a wide range of stimulation frequencies, which we test on a visual stimulation protocol. The algorithm is rooted on machine learning based pattern recognition techniques and it is tailored for a new generation parallel ultra low power processing platform (PULP), reaching performance of more that 90% accuracy in the frequency detection even for very low stimulation frequencies (<1Hz) with a power budget of 56mW.

Abstract 5123: Multiplex in situ immunoprofiling and survival outcomes in non-small cell lung cancer

Abstract The clinical activity of immune checkpoint blockade (“immunotherapies”) in non-small cell lung cancer (NSCLC) has highlighted the importance of the immune system in controlling cancer growth. Immune cells play a central role in determining response to immunotherapies, yet obtaining phenotypic information about the various immune cells in intact tissue sections and measuring cell-to-cell interactions is challenging. For this study, we applied multiplex immunofluorescence imaging to simultaneously resolve both phenotypic and morphologic information in order to determine the immune subset in the tumor -microenvironment that dictated the prognosis of NSCLC. One-hundred-twenty nine non-metastatic NSCLC tumors were studied using Tissue Microarray (TMA), with each tumor represented in triplicate. Majority of the patients were females (56%), had stage 2 (11%) or stage 3 (84%) disease, and all patients had primary resection and postoperative radiotherapy. TMA sections were stained for CD4, CD8, CD68, FOXP3, CD274 (PD-L1) and cytokeratin, with a DAPI counterstain, using a sequential multiplex IHC approach involving antibody stripping and tyramide signal amplification. Slides were imaged using an automated multispectral microscope. We determined cellular phenotype counts and per-cell marker expression in tumor and stromal areas in each core of the TMA, using machine learning-based pattern recognition and cell classification approaches. On average, 4,770 cells were analyzed in each TMA core. The densities of phenotyped cells (counts/mm2) for tumor and stroma were calculated. We analyzed the tumor to stroma ratios in each core for each of the immune subsets (CD8, CD4, FoxP3+ Treg, and CD68+ macrophages). In multivariate analysis, after controlling for patient and tumor characteristics, high CD4 tumor:stroma ratio predicted significantly for poorer overall survival (OS) (5 year OS of CD4 high: 27% vs. CD4 medium/low: 50%, p = 0.001). High CD4 content also correlated significantly to high FoxP3+ Treg levels (p&amp;lt;0.0001, Fisher's Exact test) and lower PD-L1 expression in the tumor (p = 0.038). This initial data suggests the immunosuppressive tumor microenvironment, as reflected by high tumor CD4 content, portends to poor prognosis in NSCLC patients. This is an example of assessments enabled by data sets such as this, that include per-cell expression of checkpoint proteins that retain the exact cellular locations of the various cell phenotypes within the microenvironment context. We will present analyses that explore more in depth on the immuno-biology cell-to-cell interactions, including associations between PD-L1 expression and outcomes. This study highlights the power of multiplex immunofluorescence characterization to unravel the complex immunologic network within the tumor microenvironment that may not only be prognostic but also predictive of immunotherapy responsiveness. Citation Format: Steven H. Lin, James R. Mansfield, Xiayu Rao, Wen Jiang, Penny Fang, Kristin Roman, Chichung Wang, Ritsuko Komaki, Stephen Hahn, Stephen G. Swisher, Junya Fujimoto, Ignacio Wistuba, Jing Wang, Clifford C. Hoyt. Multiplex in situ immunoprofiling and survival outcomes in non-small cell lung cancer. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 5123.

Joint modulation format/bit‐rate classification and signal‐to‐noise ratio estimation in multipath fading channels using deep machine learning

A novel algorithm for simultaneous modulation format/bit-rate classification and non-data-aided (NDA) signal-to-noise ratio (SNR) estimation in multipath fading channels by applying deep machine learning-based pattern recognition on signals’ asynchronous delay-tap plots (ADTPs) is proposed. The results for three widely-used modulation formats at two different bit-rates demonstrate classification accuracy of 99.8%. In addition, NDA SNR estimation over a wide range of 0−30 dB is shown with mean error of 1 dB. The proposed method requires low-speed, asynchronous sampling of signal and is thus ideal for low-cost multiparameter estimation under real-world channel conditions.

Machine learning based pattern recognition applied to microarray data.

Microarrays have allowed the expression level of thousands of genes or proteins to be measured simultaneously. Data sets generated by these arrays consist of a small number of observations (e.g., 20-100 samples) on a very large number of variables (e.g., 10,000 genes or proteins). The observations in these data sets often have other attributes associated with them such as a class label denoting the pathology of the subject. Finding the genes or proteins that are correlated to these attributes is often a difficult task since most of the variables do not contain information about the pathology and as such can mask the identity of the relevant features. We describe a genetic algorithm (GA) that employs both supervised and unsupervised learning to mine gene expression and proteomic data. The pattern recognition GA selects features that increase clustering, while simultaneously searching for features that optimize the separation of the classes in a plot of the two or three largest principal components of the data. Because the largest principal components capture the bulk of the variance in the data, the features chosen by the GA contain information primarily about differences between classes in the data set. The principal component analysis routine embedded in the fitness function of the GA acts as an information filter, significantly reducing the size of the search space since it restricts the search to feature sets whose principal component plots show clustering on the basis of class. The algorithm integrates aspects of artificial intelligence and evolutionary computations to yield a smart one pass procedure for feature selection, clustering, classification, and prediction.