Advanced OCTA imaging segmentation: Unsupervised, non-linear retinal vessel detection using modified self-organizing maps and joint MGRF modeling.

BACKGROUND AND OBJECTIVE: This paper proposes a fully automated and unsupervised stochastic segmentation approach using two-level joint Markov-Gibbs Random Field (MGRF) to detect the vascular system from retinal Optical Coherence Tomography Angiography (OCTA) images, which is a critical step in developing Computer-Aided Diagnosis (CAD) systems for detecting retinal diseases. METHODS: Using a new probabilistic model based on a Linear Combination of Discrete Gaussian (LCDG), the first level models the appearance of OCTA images and their spatially smoothed images. The parameters of the LCDG model are estimated using a modified Expectation Maximization (EM) algorithm. The second level models the maps of OCTA images, including the vascular system and other retina tissues, using MGRF with analytically estimated parameters from the input images. The proposed segmentation approach employs modified self-organizing maps as a MAP-based optimizer maximizing the joint likelihood and handles the Joint MGRF model in a new, unsupervised way. This approach deviates from traditional stochastic optimization approaches and leverages non-linear optimization to achieve more accurate segmentation results. RESULTS: The proposed segmentation framework is evaluated quantitatively on a dataset of 204 subjects. Achieving 0.92 ± 0.03 Dice similarity coefficient, 0.69 ± 0.25 95-percentile bidirectional Hausdorff distance, and 0.93 ± 0.03 accuracy, confirms the superior performance of the proposed approach. CONCLUSIONS: The conclusions drawn from the study highlight the superior performance of the proposed unsupervised and fully automated segmentation approach in detecting the vascular system from OCTA images. This approach not only deviates from traditional methods but also achieves more accurate segmentation results, demonstrating its potential in aiding the development of CAD systems for detecting retinal diseases.

A Deep Learning Pipeline for Grade Groups Classification Using Digitized Prostate Biopsy Specimens.

Prostate cancer is a significant cause of morbidity and mortality in the USA. In this paper, we develop a computer-aided diagnostic (CAD) system for automated grade groups (GG) classification using digitized prostate biopsy specimens (PBSs). Our CAD system aims to firstly classify the Gleason pattern (GP), and then identifies the Gleason score (GS) and GG. The GP classification pipeline is based on a pyramidal deep learning system that utilizes three convolution neural networks (CNN) to produce both patch- and pixel-wise classifications. The analysis starts with sequential preprocessing steps that include a histogram equalization step to adjust intensity values, followed by a PBSs' edge enhancement. The digitized PBSs are then divided into overlapping patches with the three sizes: 100 × 100 (CNNS), 150 × 150 (CNNM), and 200 × 200 (CNNL), pixels, and 75% overlap. Those three sizes of patches represent the three pyramidal levels. This pyramidal technique allows us to extract rich information, such as that the larger patches give more global information, while the small patches provide local details. After that, the patch-wise technique assigns each overlapped patch a label as GP categories (1 to 5). Then, the majority voting is the core approach for getting the pixel-wise classification that is used to get a single label for each overlapped pixel. The results after applying those techniques are three images of the same size as the original, and each pixel has a single label. We utilized the majority voting technique again on those three images to obtain only one. The proposed framework is trained, validated, and tested on 608 whole slide images (WSIs) of the digitized PBSs. The overall diagnostic accuracy is evaluated using several metrics: precision, recall, F1-score, accuracy, macro-averaged, and weighted-averaged. The (CNNL) has the best accuracy results for patch classification among the three CNNs, and its classification accuracy is 0.76. The macro-averaged and weighted-average metrics are found to be around 0.70-0.77. For GG, our CAD results are about 80% for precision, and between 60% to 80% for recall and F1-score, respectively. Also, it is around 94% for accuracy and NPV. To highlight our CAD systems' results, we used the standard ResNet50 and VGG-16 to compare our CNN's patch-wise classification results. As well, we compared the GG's results with that of the previous work.

Three-year functional outcome of transosseous-equivalent double-row vs. single-row repair of small and large rotator cuff tears: a double-blinded randomized controlled trial.

BACKGROUND: The trial aimed to prospectively compare the functional outcomes of patients undergoing arthroscopic rotator cuff repair using transosseous-equivalent double-row (TEDR) or single-row (SR) suture anchor techniques at 3 years postoperatively for both large (>3 cm) and small (<3 cm) tears. METHODS: Eighty patients with a symptomatic and magnetic resonance imaging (MRI)-proven full-thickness rotator cuff tear, who had failed conservative management of at least 6 months' duration and who had a complete passive range of motion of the affected shoulder, were enrolled in the trial. Patients were randomized to TEDR repair (n = 40) or SR repair (n = 40). Subgroup analysis was conducted for tears <3 cm (TEDR n = 17, SR n = 19) and tears >3 cm (TEDR n = 23, SR n = 21). Primary outcomes included the Oxford Shoulder Score (OSS), the University of California, Los Angeles (UCLA) score, and the Constant-Murley score (CMS). The secondary outcomes included a 0-100-mm visual analog scale (VAS) score for pain, range of motion (ROM), and EQ-5D scores. All patients completed a follow-up of 3 years. RESULTS: There was a significant difference in the mean OSS postoperative score for tears >3 cm (P = .01) and mean improvement from baseline in the TEDR group (P = .001). For tears >3 cm, mean postoperative scores were also significantly higher in the TEDR group for UCLA (P = .015) and CMS (P = .001). Post hoc testing showed that the differences between these groups was statistically significant (P < .05). For tears <3 cm, a significant postoperative difference in favor of SR repair was seen in the mean CMSs (P = .011), and post hoc testing showed that the difference was statistically significant (P = .015). No significant difference was seen with mean postoperative OSS or UCLA, and post hoc testing did not show a statistically significant difference between groups. CONCLUSIONS: TEDR repair showed improved functional outcomes for tears >3 cm compared with SR repair. For tears <3 cm, no clear benefit was seen with either technique.

Using resting state functional MRI to build a personalized autism diagnosis system.

Autism spectrum disorder (ASD) is a neuro-developmental disorder associated with social impairments, communication difficulties, and restricted and repetitive behaviors. Yet, there is no confirmed cause identified for ASD. Studying the functional connectivity of the brain is an emerging technique used in diagnosing and understanding ASD. In this study, we obtained the resting state functional MRI data of 283 subjects from the National Database of Autism Research (NDAR). An automated autism diagnosis system was built using the data from NDAR. The proposed system is machine learning based. Power spectral densities (PSDs) of time courses corresponding to the spatial activation areas are used as input features, feeds them to a stacked autoencoder then builds a classifier using probabilistic support vector machines. Over the used dataset, around 90% of sensitivity, specificity and accuracy was achieved by our machine learning system. Moreover, the system generalization ability was checked over two different prevalence values, one for the general population and the other for the of high risk population, and the system proved to be very generalizable, especially among the population of high risk. The proposed system generates a full personalized report for each subject, along with identifying the global differences between ASD and typically developed (TD) subjects and its ability to diagnose autism. It shows the impacted areas and the severity of implications. From the clinical aspect, this report is considered very valuable as it helps in both predicting and understanding behavior of autistic subjects. Moreover, it helps in designing a plan for personalized treatment per each individual subject. The proposed work is taking a step towards achieving personalized medicine in autism which is the ultimate goal of our group's research efforts in this area.

yuDetecting the percent of peripheral blood mononuclear cells displaying p-STAT-3 in malignant glioma patients.

BACKGROUND: The signal transducer and activator of transcription 3 (STAT-3) is frequently overexpressed in cancer cells, propagates tumorigenesis, and is a key regulator of immune suppression in cancer patients. The presence of phosphorylated STAT-3 (p-STAT-3) in the tumor can induce p-STAT-3 in tumor-associated immune cells that can return to the circulatory system. We hypothesized that the number of peripheral blood mononuclear cells (PBMCs) displaying p-STAT-3 would be increased in glioma patients, which would correlate with the extent of tumor-expressed p-STAT-3, and that higher p-STAT-3 levels in peripheral blood would correlate with a higher fraction of immune-suppressive regulatory T cells (Tregs). METHODS: We measured the percentage of PBMCs displaying p-STAT-3 in 19 healthy donors and 45 patients with primary brain tumors. The level of p-STAT-3 in tumor tissue was determined by immunohistochemistry. The degree of immune suppression was determined based on the fraction of Tregs in the CD4 compartment. RESULTS: Healthy donors had 4.8 +/- 3.6% of PBMCs that expressed p-STAT-3, while the mean proportion of PBMCs displaying p-STAT-3 in patients with GBM was 11.8 +/- 13.5% (P = 0.03). We did not observe a correlation by Spearman correlation between the degree of p-STAT-3 levels in the tumor and the percent of PBMCs displaying p-STAT-3. Furthermore, the percent of PBMCs displaying p-STAT-3 in glioma patients was not directly correlated with the fraction of Tregs in the CD4 compartment. CONCLUSION: We conclude that the percent of PBMCs displaying p-STAT-3 may be increased in malignant glioma patients.

A novel phosphorylated STAT3 inhibitor enhances T cell cytotoxicity against melanoma through inhibition of regulatory T cells.

The activation of signal transducer and activator of transcription 3 (STAT3) has been identified as a key mediator that drives the fundamental components of melanoma malignancy, including immune suppression in melanoma patients. Increasing evidence also suggests that regulatory T cells (Tregs) are important in suppressing anti-tumor immunity and play a dominant role in negating efficacious immunotherapy approaches. We hypothesized that WP1066, a novel inhibitor of STAT3 signaling, reverses immune suppression through the inhibition of Tregs and that this contributes to the antitumor activity of this agent against melanoma brain metastases. We found that the mean percentage of peripheral blood mononuclear cells expressing phosphorylated STAT3 (p-STAT3) was significantly elevated in samples from patients with melanoma brain metastases compared to healthy donors, 16.13 +/- 2.48% versus 4.17 +/- 1.79%. The p-STAT3 inhibitor WP1066 enhanced CD3+ (which contained Tregs) but not CD8+ T cell cytotoxicity against human A375 melanoma cells, indicating that this p-STAT3 blockade agent did not directly activate CD8+ T cells. Furthermore, the p-STAT3 inhibitor did not enhance the cytotoxicity of CD3+CD25- T cells (from which Tregs were excluded), indicating that the enhanced cytotoxicity of WP1066 is secondary to its inhibition of Tregs. This was confirmed by demonstrating that WP1066 inhibited FoxP3+ Treg induction in a dose-dependent manner. Moreover, CD3+ T cells exhibited markedly enhanced levels of phosphorylated ZAP-70, a critical proximal signal in T cell activation, after exposure to WP1066. Similar effects were not observed in Treg-depleted CD3+CD25- T cell populations, confirming that the T cell activation by WP compounds is secondary to their inhibition of the Tregs. These results suggest that WP1066 enhances T cell cytotoxicity against melanoma through inhibition of Tregs.

The incidence, correlation with tumor-infiltrating inflammation, and prognosis of phosphorylated STAT3 expression in human gliomas.

PURPOSE: The signal transducer and activator of transcription 3 (STAT3) is frequently overexpressed in most cancers, propagates tumorigenesis, and is a key regulator of immune suppression in cancer patients. We sought to determine the incidence of phosphorylated STAT3 (p-STAT3) expression in malignant gliomas of different pathologic types, whether p-STAT3 expression is a negative prognostic factor, and whether p-STAT3 expression influences the inflammatory response within gliomas. METHODS: Using immunohistochemical analysis, we measured the incidence of p-STAT3 expression in 129 patients with gliomas of various pathologic types in a glioma tissue microarray. We categorized our results according to the total number of p-STAT3-expressing cells within the gliomas and correlated this number with the number of infiltrating T cells and T regulatory cells. We then evaluated the association between p-STAT3 expression and median survival time using univariate and multivariate analyses. RESULTS: We did not detect p-STAT3 expression in normal brain tissues or low-grade astrocytomas. We observed significant differences in the incidence of p-STAT3 expression between the different grades of astrocytomas and different pathologic glioma types. p-STAT3 expression was associated with the population of tumor-infiltrating immune cells but not with that of T regulatory cells. On univariate analysis, we found that p-STAT3 expression within anaplastic astrocytomas was a negative prognostic factor. CONCLUSIONS: p-STAT3 expression is common within gliomas of both the astrocytic and oligodendroglial lineages and portends poor survival in patients with anaplastic astrocytomas. p-STAT3 expression differs significantly between gliomas of different pathologic types and grades and correlated with the degree of immune infiltration.

A novel inhibitor of signal transducers and activators of transcription 3 activation is efficacious against established central nervous system melanoma and inhibits regulatory T cells.

PURPOSE: Activation of signal transducers and activators of transcription 3 (STAT3) has been identified as a central mediator of melanoma growth and metastasis. We hypothesized that WP1066, a novel STAT3 blockade agent, has marked antitumor activity, even against the melanoma metastasis to brain, a site typically refractory to therapies. EXPERIMENTAL DESIGN: The antitumor activities and related mechanisms of WP1066 were investigated both in vitro on melanoma cell lines and in vivo on mice with subcutaneously syngeneic melanoma or with intracerebral melanoma tumors. RESULTS: WP1066 achieved an IC(50) of 1.6, 2.3, and 1.5 mumol/L against melanoma cell line A375, B16, and B16EGFRvIII, respectively. WP1066 suppressed the phosphorylation of Janus-activated kinase 2 and STAT3 (Tyr705) in these cells. Tumor growth in mice with subcutaneously established syngeneic melanoma was markedly inhibited by WP1066 compared with that in controls. Long-term survival (>78 days) was observed in 80% of mice with established intracerebral syngeneic melanoma treated with 40 mg/kg of WP1066 in contrast to control mice who survived for a median of 15 days. Although WP1066 did not induce immunologic memory or enhance humoral responses to EGFRvIII, this compound reduced the production of immunosuppressive cytokines and chemokines (transforming growth factor-beta, RANTES, MCP-1, vascular endothelial growth factor), markedly inhibited natural and inducible Treg proliferation, and significantly increased cytotoxic immune responses of T cells. CONCLUSIONS: The antitumor cytotoxic effects of WP1066 and its ability to induce antitumor immune responses suggest that this compound has potential for the effective treatment of melanoma metastatic to brain.

Incidence and prognostic impact of FoxP3+ regulatory T cells in human gliomas.

PURPOSE: The incidence of regulatory T cells (Treg) in intrinsic central nervous system malignancies is unknown. Immunotherapeutic approaches that inhibit the Treg population may be limited to a subset of patients with gliomas. Our hypothesis is that only the most malignant gliomas have a prominent glioma-infiltrating Treg population that contributes to the immunosuppressive biology and that the presence of Tregs is a negative prognostic variable. EXPERIMENTAL DESIGN: We measured the incidence of Tregs in 135 glial tumors (including all pathologic types) in a glioma microarray using immunohistochemical analysis. Results were categorized according to the total number of Tregs within the tumors. Correlation of the presence of Tregs with prognosis was evaluated using univariate and multivariate analyses. RESULTS: Tregs were not present in normal brain tissue and were very rarely found in low-grade gliomas and oligodendrogliomas. We observed significant differences in the prevalence of Tregs between astrocytic and oligodendroglial tumors, between tumors of different grades, and between different pathologic types of tumors. We identified Tregs most frequently in glioblastoma multiforme (GBM) but very rarely in low-grade astrocytomas. The presence of Tregs within GBMs did not alter the median survival in patients from whom the tumors were obtained. CONCLUSIONS: Treg infiltration differed significantly in the tumors according to lineage, pathology, and grade. Tregs seemed to have the highest predilection for tumors of the astrocytic lineage and specifically in the high-grade gliomas, such as GBM. In both univariate and multivariate analysis, the presence of Tregs in GBMs seemed to be prognostically neutral.