Prostate Cancer Histopathology Research Articles

Artificial Intelligence Algorithms and Their Current Role in the Identification and Comparison of Gleason Patterns in Prostate Cancer Histopathology: A Comprehensive Review.

The development of the Gleason grading system has proven to be an irreplaceable tool in prostate cancer diagnostics within urology. Despite the advancements and developments in diagnostics, there remains a discrepancy in the grading process among even the most experienced pathologists. AI algorithms have demonstrated potential in detecting cancer and assigning Gleason grades, offering a solution to the issue of significant variability among pathologists' evaluations. Our paper explores the evolving role of AI in prostate cancer histopathology, with a key focus on outcomes and the reliability of various AI algorithms for Gleason pattern assessment. We conducted a non-systematic review of the published literature to examine the role of artificial intelligence in Gleason pattern diagnostics. The PubMed and Google Scholar databases were searched to gather pertinent information about recent advancements in artificial intelligence and their impact on Gleason patterns. We found that AI algorithms are increasingly being used to identify Gleason patterns in prostate cancer, with recent studies showing promising advancements that surpass traditional diagnostic methods. These findings highlight AI's potential to be integrated into clinical practice, enhancing pathologists' workflows and improving patient outcomes. The inter-observer variability in Gleason grading has seen an improvement in efficiency with the implementation of AI. Pathologists using AI have reported successful outcomes, demonstrating its effectiveness as a supplementary tool. While some refinements are still needed before AI can be fully implemented in clinical practice, its positive impact is anticipated soon.

Multi-scale relational graph convolutional network for multiple instance learning in histopathology images

Graph convolutional neural networks have shown significant potential in natural and histopathology images. However, their use has only been studied in a single magnification or multi-magnification with either homogeneous graphs or only different node types. In order to leverage the multi-magnification information and improve message passing with graph convolutional networks, we handle different embedding spaces at each magnification by introducing the Multi-Scale Relational Graph Convolutional Network (MS-RGCN) as a multiple instance learning method. We model histopathology image patches and their relation with neighboring patches and patches at other scales (i.e., magnifications) as a graph. We define separate message-passing neural networks based on node and edge types to pass the information between different magnification embedding spaces. We experiment on prostate cancer histopathology images to predict the grade groups based on the extracted features from patches. We also compare our MS-RGCN with multiple state-of-the-art methods with evaluations on several source and held-out datasets. Our method outperforms the state-of-the-art on all of the datasets and image types consisting of tissue microarrays, whole-mount slide regions, and whole-slide images. Through an ablation study, we test and show the value of the pertinent design features of the MS-RGCN.

Validation and three years of clinical experience in using an artificial intelligence algorithm as a second read system for prostate cancer diagnosis—real-world experience

BackgroundProstate cancer ranks as the most frequently diagnosed cancer in men in the USA, with significant mortality rates. Early detection is pivotal for optimal patient outcomes, providing increased treatment options and potentially less invasive interventions. There remain significant challenges in prostate cancer histopathology, including the potential for missed diagnoses due to pathologist variability and subjective interpretations. MethodsTo address these challenges, this study investigates the ability of artificial intelligence (AI) to enhance diagnostic accuracy. The Galen™ Prostate AI algorithm was validated on a cohort of Puerto Rican men to demonstrate its efficacy in cancer detection and Gleason grading. Subsequently, the AI algorithm was integrated into routine clinical practice during a 3-year period at a CLIA certified precision pathology laboratory. ResultsThe Galen™ Prostate AI algorithm showed a 96.7% (95% CI 95.6–97.8) specificity and a 96.6% (95% CI 93.3–98.8) sensitivity for prostate cancer detection and 82.1% specificity (95% CI 73.9–88.5) and 81.1% sensitivity (95% CI 73.7–87.2) for distinction of Gleason Grade Group 1 from Grade Group 2+. The subsequent AI integration into routine clinical use examined prostate cancer diagnoses on >122,000 slides and 9200 cases over 3 years and had an overall AI Impact ™ factor of 1.8%. ConclusionsThe potential of AI to be a powerful, reliable, and effective diagnostic tool for pathologists is highlighted, while the AI Impact™ in a real-world setting demonstrates the ability of AI to standardize prostate cancer diagnosis at a high level of performance across pathologists.

Applications of artificial intelligence in prostate cancer histopathology.

The diagnosis of prostate cancer (PCa) depends on the evaluation of core needle biopsies by trained pathologists. Artificial intelligence (AI) derived models have been created to address the challenges posed by pathologists' increasing workload, workforce shortages, and variability in histopathology assessment. These models with histopathological parameters integrated into sophisticated neural networks demonstrate remarkable ability to identify, grade, and predict outcomes for PCa. Though the fully autonomous diagnosis of PCa remains elusive, recently published data suggests that AI has begun to serve as an initial screening tool, an assistant in the form of a real-time interactive interface during histological analysis, and as a second read system to detect false negative diagnoses. Our article aims to describe recent advances and future opportunities for AI in PCa histopathology.

Histopathology of Prostate Cancer and its Precursors.

Starting in the mid-1970s, we formed a group of pathologists with a major interest in uropathology. Originally, it included 2 (R.M. and M.S.). In the years the followed, the group was enlarged to include 4 more people, 2 in the mid- and late-1980s (A.L.B. and L.C.) and another in the mid-1990s (R.Ma.); a sixth (A.C.) joined the group ∼5 years ago. Two have reached the retirement age (R.M. and M.S.), while others are in the process of joining the group to replace them. A fruitful collaboration spanned for ∼45 years. This contribution is based on a series of personal recollections of the successive changes in the interpretation of prostate cancer and its precursors, starting in the mid-1970s. Here we have retraced our involvement steps, sharing issues related to them with a junior uropathologist (A.C.).

Artificial intelligence for prostate cancer histopathology diagnostics.

Artificial intelligence for prostate cancer histopathology diagnostics.

Infiltrative growth pattern of prostate cancer is associated with lower uptake on PSMA PET and reduced diffusion restriction on mpMRI

PurposeRecently, a significant association was shown between novel growth patterns on histopathology of prostate cancer (PCa) and prostate-specific membrane antigen (PSMA) uptake on [68Ga]PSMA-PET. It is the aim of this study to evaluate the association between these growth patterns and ADC (mm2/1000 s) values in comparison to [68Ga]PSMA uptake on PET/MRI.MethodsWe retrospectively evaluated patients who underwent [68Ga]PSMA PET/MRI for staging or biopsy guidance, followed by radical prostatectomy at our institution between 07/2016 and 01/2020. The dominant lesion per patient was selected based on histopathology and correlated to PET/MRI in a multidisciplinary meeting, and quantified using SUVmax for PSMA uptake and ADCmean for diffusion restriction. PCa growth pattern was classified as expansive (EXP) or infiltrative (INF) according to its properties of forming a tumoral mass or infiltrating diffusely between benign glands by two independent pathologists. Furthermore, the corresponding WHO2016 ISUP tumor grade was evaluated. The t test was used to compare means, Pearson’s test for categorical correlation, Cohen’s kappa test for interrater agreement, and ROC curve to determine the best cutoff.ResultsSixty-two patients were included (mean PSA 11.7 ± 12.5). The interrater agreement between both pathologists was almost perfect with κ = 0.81. While 25 lesions had an EXP-growth with an ADCmean of 0.777 ± 0.109, 37 showed an INF-growth with a significantly higher ADCmean of 1.079 ± 0.262 (p < 0.001). We also observed a significant difference regarding PSMA SUVmax for the EXP-growth (19.2 ± 10.9) versus the INF-growth (9.4 ± 6.2, p < 0.001). Within the lesions encompassing the EXP- or the INF-growth, no significant correlation between the ISUP groups and ADCmean could be observed (p = 0.982 and p = 0.861, respectively).ConclusionPCa with INF-growth showed significantly lower SUVmax and higher ADCmean values compared to PCa with EXP-growth. Within the growth groups, ADCmean values were independent from ISUP grading.

Prostate cancer histopathology with label-free multispectral deep UV microscopy quantifies phenotypes of tumor grade and aggressiveness

Prostate cancer histopathology with label-free multispectral deep UV microscopy quantifies phenotypes of tumor grade and aggressiveness

Quantitative ultrasound shear wave elastography (USWE)-measured tissue stiffness correlates with PIRADS scoring of MRI and Gleason score on whole-mount histopathology of prostate cancer: implications for ultrasound image-guided targeting approach

ObjectiveTo correlate quantitative tissue stiffness measurements obtained by transrectal ultrasound shear wave elastography (USWE) with PI-RADS scoring of multiparametric magnetic imaging resonance (mpMRI) using Gleason scores of radical prostatectomy as a reference standard.Patients and methods196 men with localised prostate cancer were prospectively recruited into the study and had quantitative prostate tissue stiffness measurements in kilopascals (kPa) using transrectal USWE prior to radical prostatectomy. PI-RADS scores of mpMRI were also obtained in all the men. Imaging and histopathology of radical prostatectomy specimen were oriented to each other using patient specific customised 3D moulds to guide histopathology grossing of radical prostatectomy specimens. All included patients had confirmed PCa on TRUS-guided biopsies, had both USWE and mpMRI imaging data, and underwent radical prostatectomy. Chi-square test with 95% confidence interval was used to assess the difference between Gleason score (GS) of radical prostatectomy and PI-RADS classification, as well as GS of radical prostatectomy and stiffness (in Kpa) using USWE. The correlation coefficient (r) was calculated in order to investigate relation between PI-RADS classification and tissue stiffness in kPa.ResultsThere was a statistically significant correlation between USWE-measured tissue stiffness and GS (χ2 (2, N = 196) = 23.577, p < 0.001). Also, there was a statistically significant correlation between Gleason score and PI-RADS score (χ2 (2, N = 196) = 12.838, p = 0.002). High PIRADS on MRI and high stiffness on USWE (> 100 kPa) detected more than 80% and 90% high risk prostate cancer disease. However, a weak correlation coefficient of 0.231 was observed between PI-RADS score and level of tissue stiffness measured in kPa.ConclusionQuantitative USWE and mpMRI using PI-RADS classification provide a good degree of prediction for Gleason score of clinically significant prostate cancer (csPCa). Stiffer lesions on ultrasound showed a weak correlation with PI-RADS scoring system. USWE could be used to target suspected prostate cancer.

AUTHOR REPLY

Multiparametric magnetic resonance imaging (MP-MRI) provides essential information during the process of prostate cancer (PCa) management. However, a complete understanding of the association of the findings of MP-MRI with the histopathology of PCa should continue to be an area of active research. Indiscriminate use of MP-MRI, without appropriate Integration of other clinical variables, may cause harm to patients who may have been otherwise cured of PCa with surgery or radiation therapy. The goal should be to position MP-MRI within predictive models that include critical clinical variables to bank on the strengths of each while overcoming their limitations.

T1 and T2 MR fingerprinting measurements of prostate cancer and prostatitis correlate with deep learning-derived estimates of epithelium, lumen, and stromal composition on corresponding whole mount histopathology.

To explore the associations between T1 and T2 magnetic resonance fingerprinting (MRF) measurements and corresponding tissue compartment ratios (TCRs) on whole mount histopathology of prostate cancer (PCa) and prostatitis. A retrospective, IRB-approved, HIPAA-compliant cohort consisting of 14 PCa patients who underwent 3T multiparametric MRI along with T1 and T2 MRF maps prior to radical prostatectomy was used. Correspondences between whole mount specimens and MRI and MRF were manually established. Prostatitis, PCa, and normal peripheral zone (PZ) regions of interest (ROIs) on pathology were segmented for TCRs of epithelium, lumen, and stroma using two U-net deep learning models. Corresponding ROIs were mapped to T2-weighted MRI (T2w), apparent diffusion coefficient (ADC), and T1 and T2 MRF maps. Their correlations with TCRs were computed using Pearson's correlation coefficient (R). Statistically significant differences in means were assessed using one-way ANOVA. Statistically significant differences (p < 0.01) in means of TCRs and T1 and T2 MRF were observed between PCa, prostatitis, and normal PZ. A negative correlation was observed between T1 and T2 MRF and epithelium (R = - 0.38, - 0.44, p < 0.05) of PCa. T1 MRF was correlated in opposite directions with stroma of PCa and prostatitis (R = 0.35, - 0.44, p < 0.05). T2 MRF was positively correlated with lumen of PCa and prostatitis (R = 0.57, 0.46, p < 0.01). Mean T2 MRF showed significant differences (p < 0.01) between PCa and prostatitis across both transition zone (TZ) and PZ, while mean T1 MRF was significant (p = 0.02) in TZ. Significant associations between MRF (T1 in the TZ and T2 in the PZ) and tissue compartments on corresponding histopathology were observed. • Mean T2 MRF measurements and ADC within cancerous regions of interest dropped with increasing ISUP prognostic groups (IPG). • Mean T1 and T2 MRF measurements were significantly different (p < 0.001) across IPGs, prostatitis, and normal peripheral zone (NPZ). • T2 MRF showed stronger correlations in the peripheral zone, while T1 MRF showed stronger correlations in the transition zone with histopathology for prostate cancer.

Bone Fracture Incidence After Androgen Deprivation Therapy-Investigational Agents: Results From Cancer Therapy Evaluation Program-Sponsored Early Phase Clinical Trials 2006–2013

Introduction: Androgen deprivation therapy (ADT) is a primary treatment option for patients diagnosed with locally advanced-stage or metastatic prostate cancer. Androgen deprivation can be achieved either by radical orchiectomy or by medical castration using a gonadotropin-releasing hormone agonist. ADT has been linked to an initial 12-month loss of bone mineral density, a risk factor for weight-bearing bone fracture, and therefore, a confounding hazard for adverse event when patients are enrolled on early phase trials. To better understand the frequency of ADT-investigational agent-related bone fracture, we conducted a retrospective study of National Cancer Institute Cancer Therapy Evaluation Program (CTEP)-sponsored early phase trials to determine the number of fractures observed among enrolled prostate cancer patients.Patients and Methods: 464 locally advanced-stage or metastatic prostate cancer patients were identified among seven ADT-investigational agent trials conducted between 2006 and 2013. Demographic, co-morbidity, treatment, and adverse event variables were abstracted from CTEP databases and descriptive statistics were used.Results: 464 men had a median age of 64 years, were mostly white (90%), and had a performance status of 0 or 1 (98%). The number of new bone fractures occurring on or after ADT-investigational agent treatment was very low (4.6 per 1000 person-years). The median pretrial prostate specific antigen level was 29 ng/mL and most men (71%) had prostate cancer histopathology Gleason 7 score or higher. In these trials, 43 percent of men had bone only and 35 percent had bone and visceral metastatic disease. The most frequent grade 1 or 2 adverse events were fatigue (36%), hot flashes (27%), and anemia (17%). Grade 3 or higher adverse events were rare, with hypertension (3%) and hyperglycemia (3%) observed.Conclusions: Identifying bone health factors may still be relevant in selected early phase ADT-investigational agent trial patients, emphasizing the need for improved methods for capturing baseline bone health and studying ADT-investigational agent and concurrent medication interactions on bone health.

Synoptic Versus Narrative Reporting of Prostate Biopsies at a Tertiary Healthcare Institution: Challenges, Successes, and Expectations

Abstract Objectives Cancer pathology reports are expected to contain all information required for patient management and disease surveillance. Moreover, reports for patients with prostate cancer have become increasingly complex with the addition of more pathological details. This study aimed to compare narrative and synoptic prostate cancer reports for core needle biopsies received at a tertiary hospital in Nigeria in order to determine which form was most complete according to international standards. Methods This study was conducted from January 2010 to December 2015 at the Lagos University Teaching Hospital, Lagos, Nigeria. All malignant prostate cancer histopathology reports received during this period were analyzed for the presence of important clinicopathological parameters, including the numbers of cores taken and those involved by the tumor, percentage of tumor involvement, Gleason score, and the presence of high-grade prostatic intraepithelial neoplasms (HGPINs) and perineural and lymphovascular invasion. Results A total of 83 reports were reviewed, of which 27 were in narrative and 56 in synoptic format. The documentation of clinicopathological characteristics in narrative reports was significantly incomplete compared to synoptic reports in recording the number of cores (33.3% versus 96.4%), number of cores involved by the tumor (11.1% versus 94.6%), percentage of cores involved by the tumor (3.7% versus 100.0%), and the presence of HGPINs (7.4% versus 100.0%) and perineural (59.3% versus 98.2%) and lymphovascular (48.1% versus 100.0%) invasion (P &lt; .001 each). Conclusion Synoptic reports of malignant prostate cancer biopsies received at the Lagos University Teaching Hospital were found to contain more complete information than narrative reports.

Persistent Homology for the Quantitative Evaluation of Architectural Features in Prostate Cancer Histology

The current system for evaluating prostate cancer architecture is the Gleason grading system which divides the morphology of cancer into five distinct architectural patterns, labeled 1 to 5 in increasing levels of cancer aggressiveness, and generates a score by summing the labels of the two most dominant patterns. The Gleason score is currently the most powerful prognostic predictor of patient outcomes; however, it suffers from problems in reproducibility and consistency due to the high intra-observer and inter-observer variability amongst pathologists. In addition, the Gleason system lacks the granularity to address potentially prognostic architectural features beyond Gleason patterns. We evaluate prostate cancer for architectural subtypes using techniques from topological data analysis applied to prostate cancer glandular architecture. In this work we demonstrate the use of persistent homology to capture architectural features independently of Gleason patterns. Specifically, using persistent homology, we compute topological representations of purely graded prostate cancer histopathology images of Gleason patterns 3,4 and 5, and show that persistent homology is capable of clustering prostate cancer histology into architectural groups through a ranked persistence vector. Our results indicate the ability of persistent homology to cluster prostate cancer histopathology images into unique groups with dominant architectural patterns consistent with the continuum of Gleason patterns. In addition, of particular interest, is the sensitivity of persistent homology to identify specific sub-architectural groups within single Gleason patterns, suggesting that persistent homology could represent a robust quantification method for prostate cancer architecture with higher granularity than the existing semi-quantitative measures. The capability of these topological representations to segregate prostate cancer by architecture makes them an ideal candidate for use as inputs to future machine learning approaches with the intent of augmenting traditional approaches with topological features for improved diagnosis and prognosis.

Histologic Changes in Prostate Cancer Detected Subsequent to the 2012 United States Preventive Services Task Force (USPSTF) Prostate Cancer Screening Recommendation.

We report changes in the histopathology of prostate cancer diagnosed in a large urology group practice after the final United States Preventive Services Task Force (USPSTF) Grade D recommendation against prostate-specific antigen screening. All prostate biopsies performed from 2011 through 2015 in a large urology group practice were retrospectively reviewed; 2012 was excluded as a transition year. The changes in biopsy data in years following the USPSTF decision (2013-2015) were then compared with baseline (2011). A total of 10,944 biopsies were evaluated during the study period. Positive biopsy rates rose from 39.1% at baseline to 45.2% in 2015 (P < 0.01) with a marked shift toward more aggressive cancer throughout the study period. The absolute number of patients presenting with Gleason Grade Group 4 or 5 increased from 155/year at baseline to 231, 297, and 285 in 2013, 2014, and 2015, respectively (P < 0.05), unrelated to age or racial changes over time. Black men represented 16% of the cohort. Since the USPSTF recommendation against prostate cancer screening, trends toward a substantial upward grade migration and increased volume of cancers were noted in a cohort of nearly 11,000 patients in a real-world clinical practice. Additionally, continuing reductions in cancer detection in the United States may exacerbate these trends.

Synoptic Versus Narrative Reporting of Prostate Biopsies at a Tertiary Healthcare Institution: Challenges, successes and expectations

Cancer pathology reports are expected to contain all information required for patient management and disease surveillance. Moreover, reports for patients with prostate cancer have become increasingly complex with the addition of more pathological details. This study aimed to compare narrative and synoptic prostate cancer reports for core needle biopsies received at a tertiary hospital in Nigeria in order to determine which form was most complete according to international standards. This study was conducted from January 2010 to December 2015 at the Lagos University Teaching Hospital, Lagos, Nigeria. All malignant prostate cancer histopathology reports received during this period were analysed for the presence of important clinicopathological parameters, including the numbers of cores taken and those involved by the tumour, percentage of tumour involvement, Gleason score and the presence of high-grade prostatic intraepithelial neoplasms (HGPINs) and perineural and lymphovascular invasion. A total of 83 reports were reviewed, of which 27 were in narrative and 56 in synoptic format. The documentation of clinicopathological characteristics in narrative reports was significantly incomplete compared to synoptic reports in recording the number of cores (33.3% versus 96.4%), number of cores involved by the tumour (11.1% versus 94.6%), percentage of cores involved by the tumour (3.7% versus 100.0%) and the presence of HGPINs (7.4% versus 100.0%) and perineural (59.3% versus 98.2%) and lymphovascular (48.1% versus 100.0%) invasion (P <0.001 each). Synoptic reports of malignant prostate cancer biopsies received at the Lagos University Teaching Hospital were found to contain more complete information than narrative reports.

The present and future of prostate cancer histopathology.

The present review describes recent changes in the histologic grading of prostatic adenocarcinoma and emerging data suggesting areas for further grade optimization. The grading of prostatic adenocarcinoma has evolved over time, and optimization for active surveillance management has been one major driving force. Recent changes include adoption of the Grade Group system, which stratifies patients into one of five prognostic groups. Although it provides clearer labels for patient understanding and defines a more homogenous low-risk group (i.e. Grade Group 1 of 5), emerging data suggest that further prognostic stratification may be achieved by incorporating other histologic findings. The impact of 'cribriform histology' and intraductal carcinoma as aggressive features of prostate cancer has gained significant recognition. Furthermore, quantifying the fraction of each individual Gleason pattern component may also add prognostic stratification. In addition to Gleason score and Grade Group, some institutions now identify and report high-risk histologic patterns (e.g. 'cribriform histology') and the percentage of Gleason pattern 4 on biopsies. While early in adoption, these additional features are being used to further stratify patients beyond the Grade Group system, particularly for patients with Grade Group 2 carcinomas. These histologic findings will likely be incorporated into future modifications of the grading system to further optimize patient stratification and ensure reporting consistency. The histologic evaluation of prostatic adenocarcinoma provides strong prognostic information for clinical management. The recent Grade Group system offers many improvements, but further optimization based on specific histologic patterns may evolve.

Histopathology of Prostate Cancer.

This review focuses on histopathological aspects of carcinoma of the prostate. A tissue diagnosis of adenocarcinoma is often essential for establishing a diagnosis of prostate cancer, and the foundation for a tissue diagnosis is currently light microscopic examination of hematoxylin and eosin (H&E)-stained tissue sections. Markers detected by immunohistochemistry on tissue sections can support a diagnosis of adenocarcinoma that is primary in the prostate gland or metastatic. Histological variants of carcinoma of the prostate are important for diagnostic recognition of cancer or as clinicopathologic entities that have prognostic and/or therapeutic significance. Histological grading of adenocarcinoma of the prostate, including use of the 2014 International Society of Urological Pathology (ISUP) modified Gleason grades and the new grade groups, is one of the most powerful prognostic indicators for clinically localized prostate cancer, and is one of the most critical factors in determination of management of these patients.

Content-based image retrieval of digitized histopathology in boosted spectrally embedded spaces

Context: Content-based image retrieval (CBIR) systems allow for retrieval of images from within a database that are similar in visual content to a query image. This is useful for digital pathology, where text-based descriptors alone might be inadequate to accurately describe image content. By representing images via a set of quantitative image descriptors, the similarity between a query image with respect to archived, annotated images in a database can be computed and the most similar images retrieved. Recently, non-linear dimensionality reduction methods have become popular for embedding high-dimensional data into a reduced-dimensional space while preserving local object adjacencies, thereby allowing for object similarity to be determined more accurately in the reduced-dimensional space. However, most dimensionality reduction methods implicitly assume, in computing the reduced-dimensional representation, that all features are equally important. Aims: In this paper we present boosted spectral embedding (BoSE), which utilizes a boosted distance metric to selectively weight individual features (based on training data) to subsequently map the data into a reduced-dimensional space. Settings and Design: BoSE is evaluated against spectral embedding (SE) (which employs equal feature weighting) in the context of CBIR of digitized prostate and breast cancer histopathology images. Materials and Methods: The following datasets, which were comprised of a total of 154 hematoxylin and eosin stained histopathology images, were used: (1) Prostate cancer histopathology (benign vs. malignant), (2) estrogen receptor (ER) + breast cancer histopathology (low vs. high grade), and (3) HER2+ breast cancer histopathology (low vs. high levels of lymphocytic infiltration). Statistical Analysis Used: We plotted and calculated the area under precision-recall curves (AUPRC) and calculated classification accuracy using the Random Forest classifier. Results: BoSE outperformed SE both in terms of CBIR-based (area under the precision-recall curve) and classifier-based (classification accuracy) on average across all of the dimensions tested for all three datasets: (1) Prostate cancer histopathology (AUPRC: BoSE = 0.79, SE = 0.63; Accuracy: BoSE = 0.93, SE = 0.80), (2) ER + breast cancer histopathology (AUPRC: BoSE = 0.79, SE = 0.68; Accuracy: BoSE = 0.96, SE = 0.96), and (3) HER2+ breast cancer histopathology (AUPRC: BoSE = 0.54, SE = 0.44; Accuracy: BoSE = 0.93, SE = 0.91). Conclusion: Our results suggest that BoSE could serve as an important tool for CBIR and classification of high-dimensional biomedical data.

Explicit shape descriptors: novel morphologic features for histopathology classification.

Object morphology, defined as shape and size characteristics, observed on medical imagery is often an important marker for disease presence and/or aggressiveness. In the context of prostate cancer histopathology, gland morphology is an integral component of the Gleason grading system which enables discrimination between low and high grade disease. However, clinicians are often unable to distinguish between subtle differences in object morphology, as evidenced by high inter-observer variability in Gleason grading. Boundary-based morphologic descriptors, such as the variance in the distance from points on the boundary of an object to its center, may not have the requisite discriminability to separate objects with subtle shape differences. In this paper, we present a set of novel explicit shape descriptors (ESDs) which are capable of distinguishing subtle shape differences between prostate glands of intermediate Gleason grades (grades 3 and 4) on prostate cancer histopathology. Calculation of ESDs involves: (1) representing object morphology using an explicit shape model (e.g. medial axis); (2) aligning the shape models via a non-rigid registration scheme with a diffeomorphic constraint and quantifying shape model dissimilarity; and (3) applying a non-linear dimensionality reduction scheme (e.g. Graph Embedding) to learn a low dimensional projection encoding the shape differences between objects. ESDs are hence the principal eigenvectors in the reduced embedding space. In this work we demonstrate that ESDs in conjunction with a Support Vector Machine classifier are able to correctly distinguish between 888 prostate glands corresponding to different Gleason grades (benign, grade 3, or grade 4) of prostate cancer from 58 needle biopsy specimens with a maximum accuracy of 0.89 and corresponding area under the receiver operating characteristic curve of 0.78.