Lung cancer lesion detection in histopathology images using graph-based sparse PCA network.

Early detection of lung cancer is critical for improvement of patient survival. To address the clinical need for efficacious treatments, genetically engineered mouse models (GEMM) have become integral in identifying and evaluating the molecular underpinnings of this complex disease that may be exploited as therapeutic targets. Assessment of GEMM tumor burden on histopathological sections performed by manual inspection is both time consuming and prone to subjective bias. Therefore, an interplay of needs and challenges exists for computer-aided diagnostic tools, for accurate and efficient analysis of these histopathology images. In this paper, we propose a simple machine learning approach called the graph-based sparse principal component analysis (GS-PCA) network, for automated detection of cancerous lesions on histological lung slides stained by hematoxylin and eosin (H&E). Our method comprises four steps: 1) cascaded graph-based sparse PCA, 2) PCA binary hashing, 3) block-wise histograms, and 4) support vector machine (SVM) classification. In our proposed architecture, graph-based sparse PCA is employed to learn the filter banks of the multiple stages of a convolutional network. This is followed by PCA hashing and block histograms for indexing and pooling. The meaningful features extracted from this GS-PCA are then fed to an SVM classifier. We evaluate the performance of the proposed algorithm on H&E slides obtained from an inducible K-rasG12D lung cancer mouse model using precision/recall rates, Fß-score, Tanimoto coefficient, and area under the curve (AUC) of the receiver operator characteristic (ROC) and show that our algorithm is efficient and provides improved detection accuracy compared to existing algorithms.

Automatic classification of dual-modalilty, smartphone-based oral dysplasia and malignancy images using deep learning.

With the goal to screen high-risk populations for oral cancer in low- and middle-income countries (LMICs), we have developed a low-cost, portable, easy to use smartphone-based intraoral dual-modality imaging platform. In this paper we present an image classification approach based on autofluorescence and white light images using deep learning methods. The information from the autofluorescence and white light image pair is extracted, calculated, and fused to feed the deep learning neural networks. We have investigated and compared the performance of different convolutional neural networks, transfer learning, and several regularization techniques for oral cancer classification. Our experimental results demonstrate the effectiveness of deep learning methods in classifying dual-modal images for oral cancer detection.

Automated High-Throughput Damage Scoring of Zebrafish Lateral Line Hair Cells After Ototoxin Exposure.

Zebrafish have emerged as a powerful biological system for drug development against hearing loss. Zebrafish hair cells, contained within neuromasts along the lateral line, can be damaged with exposure to ototoxins, and therefore, pre-exposure to potentially otoprotective compounds can be a means of identifying promising new drug candidates. Unfortunately, anatomical assays of hair cell damage are typically low-throughput and labor intensive, requiring trained experts to manually score hair cell damage in fluorescence or confocal images. To enhance throughput and consistency, our group has developed an automated damage-scoring algorithm based on machine-learning techniques that produce accurate damage scores, eliminate potential operator bias, provide more fidelity in determining damage scores that are between two levels, and deliver consistent results in a fraction of the time required for manual analysis. The system has been validated against trained experts using linear regression, hypothesis testing, and the Pearson's correlation coefficient. Furthermore, performance has been quantified by measuring mean absolute error for each image and the time taken to automatically compute damage scores. Coupling automated analysis of zebrafish hair cell damage to behavioral assays for ototoxicity produces a novel drug discovery platform for rapid translation of candidate drugs into preclinical mammalian models of hearing loss.

A Fully Automated High-Throughput Zebrafish Behavioral Ototoxicity Assay.

Zebrafish animal models lend themselves to behavioral assays that can facilitate rapid screening of ototoxic, otoprotective, and otoregenerative drugs. Structurally similar to human inner ear hair cells, the mechanosensory hair cells on their lateral line allow the zebrafish to sense water flow and orient head-to-current in a behavior called rheotaxis. This rheotaxis behavior deteriorates in a dose-dependent manner with increased exposure to the ototoxin cisplatin, thereby establishing itself as an excellent biomarker for anatomic damage to lateral line hair cells. Building on work by our group and others, we have built a new, fully automated high-throughput behavioral assay system that uses automated image analysis techniques to quantify rheotaxis behavior. This novel system consists of a custom-designed swimming apparatus and imaging system consisting of network-controlled Raspberry Pi microcomputers capturing infrared video. Automated analysis techniques detect individual zebrafish, compute their orientation, and quantify the rheotaxis behavior of a zebrafish test population, producing a powerful, high-throughput behavioral assay. Using our fully automated biological assay to test a standardized ototoxic dose of cisplatin against varying doses of compounds that protect or regenerate hair cells may facilitate rapid translation of candidate drugs into preclinical mammalian models of hearing loss.

A maximum-likelihood method to estimate a single ADC value of lesions using diffusion MRI.

PURPOSE: Design a statistically rigorous procedure to estimate a single apparent diffusion coefficient (ADC) of lesion from the mean lesion signal intensity in diffusion MRI. THEORY AND METHODS: A rigorous maximum-likelihood technique that incorporated the statistics of the mean lesion intensity and accounted for lesion heterogeneity was derived to estimate the ADC value. Performance evaluation included comparison with the conventionally used linear-regression and a statistically rigorous state-of-the-art ADC-map technique using realistic and clinically relevant simulation studies conducted with assistance of patient data for homogeneous and heterogeneous lesion models. RESULTS: The proposed technique outperformed the linear-regression and ADC-map approaches over a large spectrum of signal-to-noise ratio, ADC, lesion size, image-misalignment parameters, including at no image misalignment, and different amounts of lesion heterogeneity. The method was also superior at different sets of b values and in studies from specific patient-image-derived data. The technique took less than a second to execute. CONCLUSIONS: A rigorous, computationally fast, easy-to-implement, and convenient-to-use maximum-likelihood technique was proposed to estimate a single ADC value of the lesion. Results provide strong evidence in support of the method. Magn Reson Med 76:1919-1931, 2016. © 2016 International Society for Magnetic Resonance in Medicine.

Diffusion MRI with Semi-Automated Segmentation Can Serve as a Restricted Predictive Biomarker of the Therapeutic Response of Liver Metastasis.

PURPOSE: To assess the value of semi-automated segmentation applied to diffusion MRI for predicting the therapeutic response of liver metastasis. METHODS: Conventional diffusion weighted magnetic resonance imaging (MRI) was performed using b-values of 0, 150, 300 and 450s/mm(2) at baseline and days 4, 11 and 39 following initiation of a new chemotherapy regimen in a pilot study with 18 women with 37 liver metastases from primary breast cancer. A semi-automated segmentation approach was used to identify liver metastases. Linear regression analysis was used to assess the relationship between baseline values of the apparent diffusion coefficient (ADC) and change in tumor size by day 39. RESULTS: A semi-automated segmentation scheme was critical for obtaining the most reliable ADC measurements. A statistically significant relationship between baseline ADC values and change in tumor size at day 39 was observed for minimally treated patients with metastatic liver lesions measuring 2-5cm in size (p=0.002), but not for heavily treated patients with the same tumor size range (p=0.29), or for tumors of smaller or larger sizes. ROC analysis identified a baseline threshold ADC value of 1.33µm(2)/ms as 75% sensitive and 83% specific for identifying non-responding metastases in minimally treated patients with 2-5cm liver lesions. CONCLUSION: Quantitative imaging can substantially benefit from a semi-automated segmentation scheme. Quantitative diffusion MRI results can be predictive of therapeutic outcome in selected patients with liver metastases, but not for all liver metastases, and therefore should be considered to be a restricted biomarker.

Automated breast segmentation of fat and water MR images using dynamic programming.

RATIONALE AND OBJECTIVES: To develop and test an algorithm that outlines the breast boundaries using information from fat and water magnetic resonance images. MATERIALS AND METHODS: Three algorithms were implemented and tested using registered fat and water magnetic resonance images. Two of the segmentation algorithms are simple extensions of the techniques used for contrast-enhanced images: one algorithm uses clustering and local gradient (CLG) analysis and the other algorithm uses a Hessian-based sheetness filter (HSF). The third segmentation algorithm uses k-means++ and dynamic programming (KDP) for finding the breast pixels. All three algorithms separate the left and right breasts using either a fixed region or a morphological method. The performance is quantified using a mutual overlap (Dice) metric and a pectoral muscle boundary error. The algorithms are evaluated against three manual tracers using 266 breast images from 14 female subjects. RESULTS: The KDP algorithm has a mean overlap percentage improvement that is statistically significant relative to the HSF and CLG algorithms. When using a fixed region to remove the tissue between breasts with tracer 1 as a reference, the KDP algorithm has a mean overlap of 0.922 compared to 0.864 (P < .01) for HSF and 0.843 (P < .01) for CLG. The performance of KDP is very similar to tracers 2 (0.926 overlap) and 3 (0.929 overlap). The performance analysis in terms of pectoral muscle boundary error showed that the fraction of the muscle boundary within three pixels of reference tracer 1 is 0.87 using KDP compared to 0.578 for HSF and 0.617 for CLG. Our results show that the performance of the KDP algorithm is independent of breast density. CONCLUSIONS: We developed a new automated segmentation algorithm (KDP) to isolate breast tissue from magnetic resonance fat and water images. KDP outperforms the other techniques that focus on local analysis (CLG and HSF) and yields a performance similar to human tracers.

Analysis of second-harmonic-generation microscopy in a mouse model of ovarian carcinoma.

Second-harmonic-generation (SHG) imaging of mouse ovaries ex vivo was used to detect collagen structure changes accompanying ovarian cancer development. Dosing with 4-vinylcyclohexene diepoxide and 7,12-dimethylbenz[a]anthracene resulted in histologically confirmed cases of normal, benign abnormality, dysplasia, and carcinoma. Parameters for each SHG image were calculated using the Fourier transform matrix and gray-level co-occurrence matrix (GLCM). Cancer versus normal and cancer versus all other diagnoses showed the greatest separation using the parameters derived from power in the highest-frequency region and GLCM energy. Mixed effects models showed that these parameters were significantly different between cancer and normal (P<0.008). Images were classified with a support vector machine, using 25% of the data for training and 75% for testing. Utilizing all images with signal greater than the noise level, cancer versus not-cancer specimens were classified with 81.2% sensitivity and 80.0% specificity, and cancer versus normal specimens were classified with 77.8% sensitivity and 79.3% specificity. Utilizing only images with greater than of 75% of the field of view containing signal improved sensitivity and specificity for cancer versus normal to 81.5% and 81.1%. These results suggest that using SHG to visualize collagen structure in ovaries could help with early cancer detection.

Task-based evaluation of segmentation algorithms for diffusion-weighted MRI without using a gold standard.

In many studies, the estimation of the apparent diffusion coefficient (ADC) of lesions in visceral organs in diffusion-weighted (DW) magnetic resonance images requires an accurate lesion-segmentation algorithm. To evaluate these lesion-segmentation algorithms, region-overlap measures are used currently. However, the end task from the DW images is accurate ADC estimation, and the region-overlap measures do not evaluate the segmentation algorithms on this task. Moreover, these measures rely on the existence of gold-standard segmentation of the lesion, which is typically unavailable. In this paper, we study the problem of task-based evaluation of segmentation algorithms in DW imaging in the absence of a gold standard. We first show that using manual segmentations instead of gold-standard segmentations for this task-based evaluation is unreliable. We then propose a method to compare the segmentation algorithms that does not require gold-standard or manual segmentation results. The no-gold-standard method estimates the bias and the variance of the error between the true ADC values and the ADC values estimated using the automated segmentation algorithm. The method can be used to rank the segmentation algorithms on the basis of both the ensemble mean square error and precision. We also propose consistency checks for this evaluation technique.

A Maximum-Likelihood Approach for ADC Estimation of Lesions in Visceral Organs.

Accurate estimation of the apparent diffusion coefficient (ADC) of lesions in diffusion-weighted magnetic resonance imaging (DWMRI) is important to predict and monitor anti-cancer therapy response. The task of ADC estimation of lesions is complicated due to noise in the image, different variances in signal strengths at different b values and other random phenomena. In organs that have visceral motion, due to motion across scans, estimating the ADC becomes even more complex. To get rid of inaccuracies due to motion, only a single ADC value of the lesion is estimated, conventionally using a linear-regression (LR) approach. The LR approach is based on an inaccurate noise model and also suffers from other deficiencies. In this paper, we propose an easy-to-implement and computationally-fast maximum-likelihood (ML) method to estimate the ADC value of heterogeneous lesions in visceral organs. The proposed method takes into account the Rician distribution of noise in DWMRI. In the process, we also derive the statistical model for the measured mean signal intensity in DWMRI. We show using Monte-Carlo simulations that that the proposed method is more accurate than the LR method.

Automated registration of sequential breath-hold dynamic contrast-enhanced MR images: a comparison of three techniques.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is increasingly in use as an investigational biomarker of response in cancer clinical studies. Proper registration of images acquired at different time points is essential for deriving diagnostic information from quantitative pharmacokinetic analysis of these data. Motion artifacts in the presence of time-varying intensity due to contrast enhancement make this registration problem challenging. DCE-MRI of chest and abdominal lesions is typically performed during sequential breath-holds, which introduces misregistration due to inconsistent diaphragm positions and also places constraints on temporal resolution vis-à-vis free-breathing. In this work, we have employed a computer-generated DCE-MRI phantom to compare the performance of two published methods, Progressive Principal Component Registration and Pharmacokinetic Model-Driven Registration, with Sequential Elastic Registration (SER) to register adjacent time-sample images using a published general-purpose elastic registration algorithm. In all three methods, a 3D rigid-body registration scheme with a mutual information similarity measure was used as a preprocessing step. The DCE-MRI phantom images were mathematically deformed to simulate misregistration, which was corrected using the three schemes. All three schemes were comparably successful in registering large regions of interest (ROIs) such as muscle, liver, and spleen. SER was superior in retaining tumor volume and shape, and in registering smaller but important ROIs such as tumor core and tumor rim. The performance of SER on clinical DCE-MRI data sets is also presented.

Evaluating segmentation algorithms for diffusion-weighted MR images: a task-based approach.

Apparent Diffusion Coefficient (ADC) of lesions obtained from Diffusion Weighted Magnetic Resonance Imaging is an emerging biomarker for evaluating anti-cancer therapy response. To compute the lesion's ADC, accurate lesion segmentation must be performed. To quantitatively compare these lesion segmentation algorithms, standard methods are used currently. However, the end task from these images is accurate ADC estimation, and these standard methods don't evaluate the segmentation algorithms on this task-based measure. Moreover, standard methods rely on the highly unlikely scenario of there being perfectly manually segmented lesions. In this paper, we present two methods for quantitatively comparing segmentation algorithms on the above task-based measure; the first method compares them given good manual segmentations from a radiologist, the second compares them even in absence of good manual segmentations.

Computer-aided identification of ovarian cancer in confocal microendoscope images.

The confocal microendoscope is an instrument for imaging the surface of the human ovary. Images taken with this instrument from normal and diseased tissue show significant differences in cellular distribution. A real-time computer-aided system to facilitate the identification of ovarian cancer is introduced. The cellular-level structure present in ex vivo confocal microendoscope images is modeled as texture. Features are extracted based on first-order statistics, spatial gray-level-dependence matrices, and spatial-frequency content. Selection of the features is performed using stepwise discriminant analysis, forward sequential search, a nonparametric method, principal component analysis, and a heuristic technique that combines the results of these other methods. The selected features are used for classification, and the performance of various machine classifiers is compared by analyzing areas under their receiver operating characteristic curves. The machine classifiers studied included linear discriminant analysis, quadratic discriminant analysis, and the k-nearest-neighbor algorithm. The results suggest it is possible to automatically identify pathology based on texture features extracted from confocal microendoscope images and that the machine performance is superior to that of a human observer.

Performance analysis of a dual-buffer architecture for digital flow cytometry.

BACKGROUND: Most current commercial flow cytometers employ analog circuitry to provide feature values describing the pulse waveforms produced from suspended cells and particles. This restricts the type of features that can be extracted (typically pulse height, width, and integral) and consequently places a limit on classification performance. In previous work, we described a first-generation digital data acquisition and processing system that was used to demonstrate the classification advantages provided by the extraction of additional waveform features. An improved version of the system is discussed in this paper, focusing on dual-buffering to ensure increased pulse capture. A mathematical model of the system is also presented for performance analysis. METHODS: The second-generation system incorporates fast digitization of analog pulse waveforms, instantaneous pulse detection hardware, and a novel dual-buffering scheme. A mathematical model of the system was developed to theoretically compute the capture-rate performance. RESULTS: The capture rate of the system was theoretically analyzed and empirically measured. Under typical conditions, a capture rate of 8,000 pulses/s was experimentally achieved. CONCLUSIONS: Based on these results, the dual-buffer architecture shows great potential for use in flow cytometry.