The collateral map: prediction of lesion growth and penumbra after acute anterior circulation ischemic stroke.

OBJECTIVES: This study evaluated the collateral map's ability to predict lesion growth and penumbra after acute anterior circulation ischemic strokes. METHODS: This was a retrospective analysis of selected data from a prospectively collected database. The lesion growth ratio was the ratio of the follow-up lesion volume to the baseline lesion volume on diffusion-weighted imaging (DWI). The time-to-maximum (Tmax)/DWI ratio was the ratio of the baseline Tmax > 6 s volume to the baseline lesion volume. The collateral ratio was the ratio of the hypoperfused lesion volume of the phase_FU (phase with the hypoperfused lesions most approximate to the follow-up DWI lesion) to the hypoperfused lesion volume of the phase_baseline of the collateral map. Multiple logistic regression analyses were conducted to identify independent predictors of lesion growth. The concordance correlation coefficients of Tmax/DWI ratio and collateral ratio for lesion growth ratio were analyzed. RESULTS: Fifty-two patients, including twenty-six males (mean age, 74 years), were included. Intermediate (OR, 1234.5; p < 0.001) and poor collateral perfusion grades (OR, 664.7; p = 0.006) were independently associated with lesion growth. Phase_FUs were immediately preceded phases of the phase_baselines in intermediate or poor collateral perfusion grades. The concordance correlation coefficients of the Tmax/DWI ratio and collateral ratio for the lesion growth ratio were 0.28 (95% CI, 0.17-0.38) and 0.88 (95% CI, 0.82-0.92), respectively. CONCLUSION: Precise prediction of lesion growth and penumbra can be possible using collateral maps, allowing for personalized application of recanalization treatments. Further studies are needed to generalize the findings of this study. CLINICAL RELEVANCE STATEMENT: Precise prediction of lesion growth and penumbra can be possible using collateral maps, allowing for personalized application of recanalization treatments. KEY POINTS: • Cell viability in cerebral ischemia due to proximal arterial steno-occlusion mainly depends on the collateral circulation. • The collateral map shows salvageable brain extent, which can survive by recanalization treatments after acute anterior circulation ischemic stroke. • Precise estimation of salvageable brain makes it possible to make patient-specific treatment decision.

Biparametric MR signal characteristics can predict histopathological measures of prostate cancer.

OBJECTIVES: The aim of this study was to establish a new data-driven metric from MRI signal intensity that can quantify histopathological characteristics of prostate cancer. METHODS: This retrospective study was conducted on 488 patients who underwent biparametric MRI (bp-MRI), including T2-weighted imaging (T2W) and apparent diffusion coefficient (ADC) of diffusion-weighted imaging, and having biopsy-proven prostate cancer between August 2011 and July 2015. Forty-two of the patients who underwent radical prostatectomy and the rest of 446 patients constitute the labeled and unlabeled datasets, respectively. A deep learning model was built to predict the density of epithelium, epithelial nuclei, stroma, and lumen from bp-MRI, called MR-driven tissue density. On both the labeled validation set and the whole unlabeled dataset, the quality of MR-driven tissue density and its relation to bp-MRI signal intensity were examined with respect to different histopathologic and radiologic conditions using different statistical analyses. RESULTS: MR-driven tissue density and bp-MRI of 446 patients were evaluated. MR-driven tissue density was significantly related to bp-MRI (p < 0.05). The relationship was generally stronger in cancer regions than in benign regions. Regarding cancer grades, significant differences were found in the intensity of bp-MRI and MR-driven tissue density of epithelium, epithelial nuclei, and stroma (p < 0.05). Comparing MR true-negative to MR false-positive regions, MR-driven lumen density was significantly different, similar to the intensity of bp-MRI (p < 0.001). CONCLUSIONS: MR-driven tissue density could serve as a reliable histopathological measure of the prostate on bp-MRI, leading to an improved understanding of prostate cancer and cancer progression. KEY POINTS: • Semi-supervised deep learning enables non-invasive and quantitative histopathology in the prostate from biparametric MRI. • Tissue density derived from biparametric MRI demonstrates similar characteristics to the direct estimation of tissue density from histopathology images. • The analysis of MR-driven tissue density reveals significantly different tissue compositions among different cancer grades as well as between MR-positive and MR-negative benign.

Prostate Cancer: A Correlative Study of Multiparametric MR Imaging and Digital Histopathology.

Purpose To correlate multiparametric magnetic resonance (MR) imaging and quantitative digital histopathologic analysis (DHA) of the prostate. Materials and Methods This retrospective study was approved by the local institutional review board and was HIPAA compliant. Forty patients (median age, 60 years; age range, 44-71 years) who underwent prostate MR imaging consisting of T2-weighted and diffusion-weighted (DW) MR imaging along with subsequent robot-assisted radical prostatectomy gave informed consent to be included. Whole-mount tissue specimens were obtained with a patient-specific mold, and DHA was performed to assess the lumen, epithelium, stroma, and epithelial nucleus. These DHA images were registered with MR images and were correlated on a per-voxel basis. The relationship between MR imaging and DHA was assessed by using a linear mixed-effects model and the Pearson correlation coefficient. Results T2-weighted MR imaging, apparent diffusion coefficient (ADC) of DW imaging, and high-b-value DW imaging were significantly related to specific DHA parameters (P < .01). For instance, lumen density (ie, the percentage area of tissue components) was associated with T2-weighted MR imaging (slope = 0.36 ± 0.05 [standard error], γ = 0.35), ADC (slope = 0.47 ± 0.05, γ = 0.50), and high-b-value DW imaging (slope = -0.44 ± 0.05, γ = -0.44). Differences between regions harboring benign tissue and those harboring malignant tissue were observed at MR imaging and DHA (P < .01). Gleason score was significantly associated with MR imaging and DHA parameters (P < .05). For example, it was positively related to high-b-value DW imaging (slope = 0.21 ± 0.16, γ = 0.18) and negatively related to lumen density (slope = -0.19 ± 0.18, γ = -0.35). Conclusion Overall, significant associations were observed between MR imaging and DHA, regardless of prostate anatomy. © RSNA, 2017 Online supplemental material is available for this article.

Automated prostate tissue referencing for cancer detection and diagnosis.

BACKGROUND: The current practice of histopathology review is limited in speed and accuracy. The current diagnostic paradigm does not fully describe the complex and complicated patterns of cancer. To address these needs, we develop an automated and objective system that facilitates a comprehensive and easy information management and decision-making. We also develop a tissue similarity measure scheme to broaden our understanding of tissue characteristics. RESULTS: The system includes a database of previously evaluated prostate tissue images, clinical information and a tissue retrieval process. In the system, a tissue is characterized by its morphology. The retrieval process seeks to find the closest matching cases with the tissue of interest. Moreover, we define 9 morphologic criteria by which a pathologist arrives at a histomorphologic diagnosis. Based on the 9 criteria, true tissue similarity is determined and serves as the gold standard of tissue retrieval. Here, we found a minimum of 4 and 3 matching cases, out of 5, for ~80 % and ~60 % of the queries when a match was defined as the tissue similarity score ≥5 and ≥6, respectively. We were also able to examine the relationship between tissues beyond the Gleason grading system due to the tissue similarity scoring system. CONCLUSIONS: Providing the closest matching cases and their clinical information with pathologists will help to conduct consistent and reliable diagnoses. Thus, we expect the system to facilitate quality maintenance and quality improvement of cancer pathology.

Stromal-epithelial responses to fractionated radiotherapy in a breast cancer microenvironment.

BACKGROUND: The stromal-epithelial-cell interactions that are responsible for directing normal breast-tissue development and maintenance play a central role in the progression of breast cancer. In the present study, we continued our development of three-dimensional (3-D) cell co-cultures used to study cancerous mammary cell responses to fractionated radiotherapy. In particular, we focused on the role of the reactive stroma in determining the therapeutic ratio for post-surgical treatment. METHODS: Cancerous human mammary epithelial cells (MRC-7) were cultured in a 3-D collagen matrix with human fibroblasts (MRC-5) stimulated by various concentrations of transforming growth factor beta 1 (TGF-ß1). These culture samples were designed to model the post-lumpectomy mammary stroma in the presence of residual cancer cells. We tracked over time the changes in medium stiffness, fibroblast-cell activation (MRC-5 converted to cancer activated fibroblasts (CAFs)), and proliferation of both cell types under a variety of fractionated radiotherapy protocols. Samples were exposed to 6 MV X-rays from a linear accelerator in daily fraction sizes of 90, 180 and 360 cGy over five days in a manner consistent with irradiation exposure during radiotherapy. RESULTS: We found in fractionation studies with MRC-5 fibroblasts and CAFs that higher doses per fraction may be more effective early on in deactivating cancer-harboring cellular environments. Higher-dose fraction schemes inhibit contractility in CAFs and prevent differentiation of fibroblasts, thereby metabolically uncoupling tumor cells from their surrounding stroma. However, higher dose fraction appears to increase ECM stiffening. CONCLUSIONS: The findings suggest that dose escalation to the region with residual disease can deactivate the reactive stroma, thus minimizing the cancer promoting features of the cellular environment. Large-fraction irradiation may be used to sterilize residual tumor cells and inhibit activation of intracellular transduction pathways that are promoted during the post-surgical wound-healing period. The higher dose fractions may slow wound healing and increase ECM stiffening that could stimulate proliferation of surviving cancer cells.

Efficient Data Mining for Local Binary Pattern in Texture Image Analysis.

Local binary pattern (LBP) is a simple gray scale descriptor to characterize the local distribution of the grey levels in an image. Multi-resolution LBP and/or combinations of the LBPs have shown to be effective in texture image analysis. However, it is unclear what resolutions or combinations to choose for texture analysis. Examining all the possible cases is impractical and intractable due to the exponential growth in a feature space. This limits the accuracy and time- and space-efficiency of LBP. Here, we propose a data mining approach for LBP, which efficiently explores a high-dimensional feature space and finds a relatively smaller number of discriminative features. The features can be any combinations of LBPs. These may not be achievable with conventional approaches. Hence, our approach not only fully utilizes the capability of LBP but also maintains the low computational complexity. We incorporated three different descriptors (LBP, local contrast measure, and local directional derivative measure) with three spatial resolutions and evaluated our approach using two comprehensive texture databases. The results demonstrated the effectiveness and robustness of our approach to different experimental designs and texture images.

DAX-Net: A dual-branch dual-task adaptive cross-weight feature fusion network for robust multi-class cancer classification in pathology images.

BACKGROUND AND OBJECTIVE: Multi-class cancer classification has been extensively studied in digital and computational pathology due to its importance in clinical decision-making. Numerous computational tools have been proposed for various types of cancer classification. Many of them are built based on convolutional neural networks. Recently, Transformer-style networks have shown to be effective for cancer classification. Herein, we present a hybrid design that leverages both convolutional neural networks and transformer architecture to obtain superior performance in cancer classification. METHODS: We propose a dual-branch dual-task adaptive cross-weight feature fusion network, called DAX-Net, which exploits heterogeneous feature representations from the convolutional neural network and Transformer network, adaptively combines them to boost their representation power, and conducts cancer classification as categorical classification and ordinal classification. For an efficient and effective optimization of the proposed model, we introduce two loss functions that are tailored to the two classification tasks. RESULTS: To evaluate the proposed method, we employed colorectal and prostate cancer datasets, of which each contains both in-domain and out-of-domain test sets. For colorectal cancer, the proposed method obtained an accuracy of 88.4%, a quadratic kappa score of 0.945, and an F1 score of 0.831 for the in-domain test set, and 84.4%, 0.910, and 0.768 for the out-of-domain test set. For prostate cancer, it achieved an accuracy of 71.6%, a kappa score of 0.635, and an F1 score of 0.655 for the in-domain test set, 79.2% accuracy, 0.721 kappa score, and 0.686 F1 score for the first out-of-domain test set, and 58.1% accuracy, 0.564 kappa score, and 0.493 F1 score for the second out-of-domain test set. It is worth noting that the performance of the proposed method outperformed other competitors by significant margins, in particular, with respect to the out-of-domain test sets. CONCLUSIONS: The experimental results demonstrate that the proposed method is not only accurate but also robust to varying conditions of the test sets in comparison to several, related methods. These results suggest that the proposed method can facilitate automated cancer classification in various clinical settings.

Impact of Collateral Circulation on Futile Endovascular Thrombectomy in Acute Anterior Circulation Ischemic Stroke.

OBJECTIVE: Collateral circulation is associated with the differential treatment effect of endovascular thrombectomy (EVT) in acute ischemic stroke. We aimed to verify the ability of the collateral map to predict futile EVT in patients with acute anterior circulation ischemic stroke. METHODS: This secondary analysis of a prospective observational study included data from participants underwent EVT for acute ischemic stroke due to occlusion of the internal carotid artery and/or the middle cerebral artery within 8 hours of symptom onset. Multiple logistic regression analyses were conducted to identify independent predictors of futile recanalization (modified Rankin scale score at 90 days of 4-6 despite of successful reperfusion). RESULTS: In a total of 214 participants, older age (odds ratio [OR], 2.40; 95% confidence interval [CI], 1.56 to 3.67; p<0.001), higher baseline National Institutes of Health Stroke Scale (NIHSS) scores (OR, 1.12; 95% CI, 1.04 to 1.21; p=0.004), very poor collateral perfusion grade (OR, 35.09; 95% CI, 3.50 to 351.33; p=0.002), longer door-to-puncture time (OR, 1.08; 95% CI, 1.02 to 1.14; p=0.009), and failed reperfusion (OR, 3.73; 95% CI, 1.30 to 10.76; p=0.015) were associated with unfavorable functional outcomes. In 184 participants who achieved successful reperfusion, older age (OR, 2.30; 95% CI, 1.44 to 3.67; p<0.001), higher baseline NIHSS scores (OR, 1.12; 95% CI, 1.03 to 1.22; p=0.006), very poor collateral perfusion grade (OR, 4.96; 95% CI, 1.42 to 17.37; p=0.012), and longer door-to-reperfusion time (OR, 1.09; 95% CI, 1.03 to 1.15; p=0.003) were associated with unfavorable functional outcomes. CONCLUSION: The assessment of collateral perfusion status using the collateral map can predict futile EVT, which may help select ineligible patients for EVT, thereby potentially reducing the rate of futile EVT.

CaMeL-Net: Centroid-aware metric learning for efficient multi-class cancer classification in pathology images.

BACKGROUND AND OBJECTIVE: Cancer grading in pathology image analysis is a major task due to its importance in patient care, treatment, and management. The recent developments in artificial neural networks for computational pathology have demonstrated great potential to improve the accuracy and quality of cancer diagnosis. These improvements are generally ascribable to the advance in the architecture of the networks, often leading to increase in the computation and resources. In this work, we propose an efficient convolutional neural network that is designed to conduct multi-class cancer classification in an accurate and robust manner via metric learning. METHODS: We propose a centroid-aware metric learning network for an improved cancer grading in pathology images. The proposed network utilizes centroids of different classes within the feature embedding space to optimize the relative distances between pathology images, which manifest the innate similarities/dissimilarities between them. For improved optimization, we introduce a new loss function and a training strategy that are tailored to the proposed network and metric learning. RESULTS: We evaluated the proposed approach on multiple datasets of colorectal and gastric cancers. For the colorectal cancer, two different datasets were employed that were collected from different acquisition settings. the proposed method achieved an accuracy, F1-score, quadratic weighted kappa of 88.7%, 0.849, and 0.946 for the first dataset and 83.3%, 0.764, and 0.907 for the second dataset, respectively. For the gastric cancer, the proposed method obtained an accuracy of 85.9%, F1-score of 0.793, and quadratic weighted kappa of 0.939. We also found that the proposed method outperforms other competing models and is computationally efficient. CONCLUSIONS: The experimental results demonstrate that the prediction results by the proposed network are both accurate and reliable. The proposed network not only outperformed other related methods in cancer classification but also achieved superior computational efficiency during training and inference. The future study will entail further development of the proposed method and the application of the method to other problems and domains.

A novel dataset and efficient deep learning framework for automated grading of renal cell carcinoma from kidney histopathology images.

Trends of kidney cancer cases worldwide are expected to increase persistently and this inspires the modification of the traditional diagnosis system to respond to future challenges. Renal Cell Carcinoma (RCC) is the most common kidney cancer and responsible for 80-85% of all renal tumors. This study proposed a robust and computationally efficient fully automated Renal Cell Carcinoma Grading Network (RCCGNet) from kidney histopathology images. The proposed RCCGNet contains a shared channel residual (SCR) block which allows the network to learn feature maps associated with different versions of the input with two parallel paths. The SCR block shares the information between two different layers and operates the shared data separately by providing beneficial supplements to each other. As a part of this study, we also introduced a new dataset for the grading of RCC with five different grades. We obtained 722 Hematoxylin & Eosin (H &E) stained slides of different patients and associated grades from the Department of Pathology, Kasturba Medical College (KMC), Mangalore, India. We performed comparable experiments which include deep learning models trained from scratch as well as transfer learning techniques using pre-trained weights of the ImageNet. To show the proposed model is generalized and independent of the dataset, we experimented with one additional well-established data called BreakHis dataset for eight class-classification. The experimental result shows that proposed RCCGNet is superior in comparison with the eight most recent classification methods on the proposed dataset as well as BreakHis dataset in terms of prediction accuracy and computational complexity.

Multi-cell type and multi-level graph aggregation network for cancer grading in pathology images.

In pathology, cancer grading is crucial for patient management and treatment. Recent deep learning methods, based upon convolutional neural networks (CNNs), have shown great potential for automated and accurate cancer diagnosis. However, these do not explicitly utilize tissue/cellular composition, and thus difficult to incorporate the existing knowledge of cancer pathology. In this study, we propose a multi-cell type and multi-level graph aggregation network (MMGA-Net) for cancer grading. Given a pathology image, MMGA-Net constructs multiple cell graphs at multiple levels to represent intra- and inter-cell type relationships and to incorporate global and local cell-to-cell interactions. In addition, it extracts tissue contextual information using a CNN. Then, the tissue and cellular information are fused to predict a cancer grade. The experimental results on two types of cancer datasets demonstrate the effectiveness of MMGA-Net, outperforming other competing models. The results also suggest that the information fusion of multiple cell types and multiple levels via graphs is critical for improved pathology image analysis.

PAIP 2020: Microsatellite instability prediction in colorectal cancer.

Microsatellite instability (MSI) refers to alterations in the length of simple repetitive genomic sequences. MSI status serves as a prognostic and predictive factor in colorectal cancer. The MSI-high status is a good prognostic factor in stage II/III cancer, and predicts a lack of benefit to adjuvant fluorouracil chemotherapy in stage II cancer but a good response to immunotherapy in stage IV cancer. Therefore, determining MSI status in patients with colorectal cancer is important for identifying the appropriate treatment protocol. In the Pathology Artificial Intelligence Platform (PAIP) 2020 challenge, artificial intelligence researchers were invited to predict MSI status based on colorectal cancer slide images. Participants were required to perform two tasks. The primary task was to classify a given slide image as belonging to either the MSI-high or the microsatellite-stable group. The second task was tumor area segmentation to avoid ties with the main task. A total of 210 of the 495 participants enrolled in the challenge downloaded the images, and 23 teams submitted their final results. Seven teams from the top 10 participants agreed to disclose their algorithms, most of which were convolutional neural network-based deep learning models, such as EfficientNet and UNet. The top-ranked system achieved the highest F1 score (0.9231). This paper summarizes the various methods used in the PAIP 2020 challenge. This paper supports the effectiveness of digital pathology for identifying the relationship between colorectal cancer and the MSI characteristics.

Analysis of variance in spectroscopic imaging data from human tissues.

The analysis of cell types and disease using Fourier transform infrared (FT-IR) spectroscopic imaging is promising. The approach lacks an appreciation of the limits of performance for the technology, however, which limits both researcher efforts in improving the approach and acceptance by practitioners. One factor limiting performance is the variance in data arising from biological diversity, measurement noise or from other sources. Here we identify the sources of variation by first employing a high throughout sampling platform of tissue microarrays (TMAs) to record a sufficiently large and diverse set data. Next, a comprehensive set of analysis of variance (ANOVA) models is employed to analyze the data. Estimating the portions of explained variation, we quantify the primary sources of variation, find the most discriminating spectral metrics, and recognize the aspects of the technology to improve. The study provides a framework for the development of protocols for clinical translation and provides guidelines to design statistically valid studies in the spectroscopic analysis of tissue.

A 3D Multi-task Regression and Ordinal Regression Deep Neural Network for Collateral Imaging from Dynamic Susceptibility Contrast-Enhanced MR perfusion in Acute Ischemic Stroke.

BACKGROUND AND OBJECTIVE: Cerebral collaterals have been identified as one of the primary determinants for treatment options in acute ischemic stroke. Several works have been proposed, but these have not been adopted for a routine clinical usage due to their manual and heuristic nature as well as inconsistency and instability of the assessment. Herein, we present an advanced deep learning-based method that can automatically generate a multiphase collateral imaging (collateral map) derived from dynamic susceptibility contrast-enhanced MR perfusion (DSC-MRP) in an accurate and robust manner. METHODS: We develop a 3D multi-task regression and ordinal regression deep neural network for generating collateral maps from DSC-MRP, which formulates the prediction of collateral maps as both a regression task and an ordinal regression task. For an ordinal regression task, we introduce a spacing-decreasing discretization (SDD) strategy to represent the intensity of the collateral status on a discrete, ordinal scale. We also devise loss functions to achieve effective and efficient multi-task learning. RESULTS: We systematically evaluated the performance of the proposed network using DSC-MRP from 802 patients. On average, the proposed network achieved ≥0.900 squared correlation coefficient (R-Squared), ≥0.916 Tanimoto measure (TM), ≥0.0913 structural similarity index measure (SSIM), and ≤0.564 × 10-1 mean absolute error (MAE), outperforming eight competing models that have been recently developed in medical imaging and computer vision. We also found that the proposed network could provide an improved contrast between the low and high intensity regions in the collateral maps, which is a key to an accurate evaluation of the collateral status. CONCLUSIONS: The experimental results demonstrate that the proposed network is able to generate collateral maps with high accuracy, facilitating a timely and prompt assessment of the collateral status in clinlcs. The future study will entail the optimization of the proposed network and its clinical evalution in a prospective manner.

Multimodal microscopy for automated histologic analysis of prostate cancer.

BACKGROUND: Prostate cancer is the single most prevalent cancer in US men whose gold standard of diagnosis is histologic assessment of biopsies. Manual assessment of stained tissue of all biopsies limits speed and accuracy in clinical practice and research of prostate cancer diagnosis. We sought to develop a fully-automated multimodal microscopy method to distinguish cancerous from non-cancerous tissue samples. METHODS: We recorded chemical data from an unstained tissue microarray (TMA) using Fourier transform infrared (FT-IR) spectroscopic imaging. Using pattern recognition, we identified epithelial cells without user input. We fused the cell type information with the corresponding stained images commonly used in clinical practice. Extracted morphological features, optimized by two-stage feature selection method using a minimum-redundancy-maximal-relevance (mRMR) criterion and sequential floating forward selection (SFFS), were applied to classify tissue samples as cancer or non-cancer. RESULTS: We achieved high accuracy (area under ROC curve (AUC) >0.97) in cross-validations on each of two data sets that were stained under different conditions. When the classifier was trained on one data set and tested on the other data set, an AUC value of ~0.95 was observed. In the absence of IR data, the performance of the same classification system dropped for both data sets and between data sets. CONCLUSIONS: We were able to achieve very effective fusion of the information from two different images that provide very different types of data with different characteristics. The method is entirely transparent to a user and does not involve any adjustment or decision-making based on spectral data. By combining the IR and optical data, we achieved high accurate classification.

Unsupervised Tumor Characterization via Conditional Generative Adversarial Networks.

Grading for cancer, based upon the degree of cancer differentiation, plays a major role in describing the characteristics and behavior of the cancer and determining treatment plan for patients. The grade is determined by a subjective and qualitative assessment of tissues under microscope, which suffers from high inter- and intra-observer variability among pathologists. Digital pathology offers an alternative means to automate the procedure as well as to improve the accuracy and robustness of cancer grading. However, most of such methods tend to mimic or reproduce cancer grade determined by human experts. Herein, we propose an alternative, quantitative means of assessing and characterizing cancers in an unsupervised manner. The proposed method utilizes conditional generative adversarial networks to characterize tissues. The proposed method is evaluated using whole slide images (WSIs) and tissue microarrays (TMAs) of colorectal cancer specimens. The results suggest that the proposed method holds a potential for quantifying cancer characteristics and improving cancer pathology.

Semi-supervised learning for an improved diagnosis of COVID-19 in CT images.

Coronavirus disease 2019 (COVID-19) has been spread out all over the world. Although a real-time reverse-transcription polymerase chain reaction (RT-PCR) test has been used as a primary diagnostic tool for COVID-19, the utility of CT based diagnostic tools have been suggested to improve the diagnostic accuracy and reliability. Herein we propose a semi-supervised deep neural network for an improved detection of COVID-19. The proposed method utilizes CT images in a supervised and unsupervised manner to improve the accuracy and robustness of COVID-19 diagnosis. Both labeled and unlabeled CT images are employed. Labeled CT images are used for supervised leaning. Unlabeled CT images are utilized for unsupervised learning in a way that the feature representations are invariant to perturbations in CT images. To systematically evaluate the proposed method, two COVID-19 CT datasets and three public CT datasets with no COVID-19 CT images are employed. In distinguishing COVID-19 from non-COVID-19 CT images, the proposed method achieves an overall accuracy of 99.83%, sensitivity of 0.9286, specificity of 0.9832, and positive predictive value (PPV) of 0.9192. The results are consistent between the COVID-19 challenge dataset and the public CT datasets. For discriminating between COVID-19 and common pneumonia CT images, the proposed method obtains 97.32% accuracy, 0.9971 sensitivity, 0.9598 specificity, and 0.9326 PPV. Moreover, the comparative experiments with respect to supervised learning and training strategies demonstrate that the proposed method is able to improve the diagnostic accuracy and robustness without exhaustive labeling. The proposed semi-supervised method, exploiting both supervised and unsupervised learning, facilitates an accurate and reliable diagnosis for COVID-19, leading to an improved patient care and management.

Joint categorical and ordinal learning for cancer grading in pathology images.

Cancer grading in pathology image analysis is one of the most critical tasks since it is related to patient outcomes and treatment planning. Traditionally, it has been considered a categorical problem, ignoring the natural ordering among the cancer grades, i.e., the higher the grade is, the more aggressive it is, and the worse the outcome is. Herein, we propose a joint categorical and ordinal learning framework for cancer grading in pathology images. The approach simultaneously performs both categorical classification and ordinal classification and aims to leverage the distinctive features from the two tasks. Moreover, we propose a new loss function for the ordinal classification task that offers an improved contrast between the correctly classified examples and misclassified examples. The proposed method is evaluated on multiple collections of colorectal and prostate pathology images that underwent different acquisition and processing procedures. Both quantitative and qualitative assessments of the experimental results confirm the effectiveness and robustness of the proposed method in comparison to other competing methods. The results suggest that the proposed approach could permit improved histopathologic analysis of cancer grades in pathology images.

Region-aggregated attention CNN for disease detection in fruit images.

BACKGROUND: Diseases and pests have a profound effect on a yearly harvest and productivity in agriculture. A precise and accurate detection of the diseases and pests could facilitate timely treatment and management of the diseases and pests and lessen the resultant loss in economy and health. Herein, we propose an improved design of the disease detection system for plant images. METHODS: Built upon the two-stage framework of object detection neural networks such as Mask R-CNN, the proposed network involves three types of extensions, including the addition of additional level of feature pyramids to improve the exploration and proposal of candidate regions, the aggregation of feature maps from all levels of feature pyramids per candidate region to fully exploit the information from feature pyramids, and the introduction of a squeeze-and-excitation block to the construction of feature pyramids and the aggregated feature maps to improve the representation of feature maps. RESULTS: The proposed network was evaluated using 74 images of infected apple fruits. In 3-fold cross-validation, the proposed network achieved averaged precision (AP) of 72.26, AP at 0.5 threshold of 88.51 and AP at 0.75 threshold of 82.30. In the comparative experiments, the proposed network outperformed the other competing networks. The utility of the three extensions was also demonstrated in comparison to Mask R-CNN. CONCLUSIONS: The experimental results suggest that the proposed network could identify and localize the symptom of the disease with high accuracy, leading to an early diagnosis and treatment of the disease, and thus holding the potential for improving crop yield and quality.