Lightweight preprocessing and template matching facilitate streamlined ischemic myocardial scar classification.

Purpose: Ischemic myocardial scarring (IMS) is a common outcome of coronary artery disease that potentially leads to lethal arrythmias and heart failure. Late-gadolinium-enhanced cardiac magnetic resonance (CMR) imaging scans have served as the diagnostic bedrock for IMS, with recent advancements in machine learning enabling enhanced scar classification. However, the trade-off for these improvements is intensive computational and time demands. As a solution, we propose a combination of lightweight preprocessing (LWP) and template matching (TM) to streamline IMS classification. Approach: CMR images from 279 patients (151 IMS, 128 control) were classified for IMS presence using two convolutional neural networks (CNNs) and TM, both with and without LWP. Evaluation metrics included accuracy, sensitivity, specificity, F1-score, area under the receiver operating characteristic curve (AUROC), and processing time. External testing dataset analysis encompassed patient-level classifications (PLCs) and a CNN versus TM classification comparison (CVTCC). Results: LWP enhanced the speed of both CNNs (4.9x) and TM (21.9x). Furthermore, in the absence of LWP, TM outpaced CNNs by over 10x, while with LWP, TM was more than 100x faster. Additionally, TM performed similarly to the CNNs in accuracy, sensitivity, specificity, F1-score, and AUROC, with PLCs demonstrating improvements across all five metrics. Moreover, the CVTCC revealed a substantial 90.9% agreement. Conclusions: Our results highlight the effectiveness of LWP and TM in streamlining IMS classification. Anticipated enhancements to LWP's region of interest (ROI) isolation and TM's ROI targeting are expected to boost accuracy, positioning them as a potential alternative to CNNs for IMS classification, supporting the need for further research.

Frozen Section Timeout: Pilot Study to Reconcile Margins Using 3D Resected Specimen and Defect Scans.

OBJECTIVE: Opportunities exist to improve intraoperative communication and documentation of resection margin details. We instituted a "frozen section timeout" that centers around visualization of the paired resection specimen and surgical defect-facilitating effective, bidirectional exchange of information. METHODS: We designed an interactive form for use during the "frozen section timeout" including annotated 3D virtual models of the resected specimen and surgical defect, plus a "line-item" table for primary and supplemental margin results. The "timeout" was conducted over a Zoom call between the operating room and frozen section laboratory. The form was simultaneously projected and discussed while all members of the surgical care team stopped activities. Nurses, co-surgeons, and all other members of the surgical team were encouraged to take part in this process. RESULTS: Twenty-six frozen section timeouts were conducted during head and neck surgeries in the Department of Otolaryngology at Mount Sinai West Hospital. These timeouts were facilitated by the lead surgeon, and all other activities were halted to ensure that critical information was shared, documented, and agreed upon. During the timeout, the annotated specimen and defect scans were displayed, clearly demonstrating the at-risk margins and the corresponding location and breadth of supplemental margins harvested. CONCLUSION: Incorporating a frozen section timeout can improve intraoperative communication, increase transparency, and potentially eliminate uncertainty regarding margin status and tumor clearance. Visualization of at-risk margins and the corresponding location and breadth of supplemental margins promises an unprecedented level of documentation and understanding. This novel technique can establish a new and improved standard of care. LEVEL OF EVIDENCE: NA Laryngoscope, 134:725-731, 2024.

iPad Annotation of 3D Surgical Models Using Procreate®: Novel Documentation of Supplemental Margins.

We present a novel, efficient approach to demonstrating supplemental margins during oncologic resection. Surgeons and pathologists annotated 10 virtual models of surgical defects and resection specimens in 3D using an iPad-based application, Procreate®. Incorporating this method into the surgical workflow can improve interdepartmental communication and provide visual documentation of surgical steps taken to address at-risk margins. Laryngoscope, 2023.

Functional loss of tumor suppressor genes detected by loss of heterozygosity, but not driver mutations, predicts aggressive lymph node status in papillary thyroid carcinoma.

BACKGROUND: Recognizing aggressive tumor biology is essential to optimizing patient management for papillary thyroid carcinomas (PTC). Aggressive lymph node (ALN) status is one feature that influences decision-making. We evaluated genomic deletions in regions of tumor suppressor genes, detected by loss of heterozygosity (LOH) analysis, to understand causal alterations linked to thyroid cancer aggressiveness and to serve as a molecular diagnostic biomarker for ALN status. METHODS: We analyzed 105 primary PTC enriched for patients with ALN (64% with, 36% without). We also analyzed 39 positive lymph nodes (79% with, 21% without ALN). LOH was determined using a panel of 25 polymorphic microsatellite alleles targeting 10 genomic loci harboring common tumor suppressor genes. Additionally, ThyGeNEXT® and ThyraMIR® assays were performed. RESULTS: LOH was detected in 43/67 primary PTC from patients with ALN status, compared with only 5/38 primary PTC without ALN (minimal metastatic burden) (P=0.0000003). This is further supported by post hoc analyses of paired primary and metastatic samples. Paired samples from patients with ALN are more likely to harbor LOH, compared to the ALN negative group (P=0.0125). Additionally, 12/31 paired samples from patients with ALN demonstrated additional or different LOH loci in metastatic samples compared to the primary tumor samples. No association was seen between ALN and mutational, translocation, or microRNA data. CONCLUSIONS: LOH detected in primary PTC significantly predicts ALN status. Analysis of paired primary and metastatic samples from patients with / without ALN status further supports this relationship. The acquisition of LOH at additional loci is common in lymph nodes from patients with ALN status. SIMPLE SUMMARY: A subset of patients with papillary thyroid carcinoma (PTC) will develop recurrent disease. One known predictor of recurrence is the American Thyroid Association category "Aggressive Lymph Node" (ALN) disease, considering metastatic burden. Loss of heterozygosity (LOH) - chromosomal loss in regions of tumor suppressor genes - has yet to be investigated as a possible mechanism driving ALN status in PTC. The ability to predict ALN status prior to surgery can guide the extent of surgery and postoperative treatment options. We found that paired samples from patients with ALN are more likely to harbor LOH, compared to patients without ALN disease. 38% of patients with ALN demonstrated additional or different LOH loci in metastatic samples compared to the primary tumor samples. LOH complements current molecular analysis of thyroid cancer when searching for evidence of aggressive biology.

Intraoperative three-dimensional scanning of head and neck surgical defects: Enhanced communication and documentation of harvested supplemental margins.

BACKGROUND: We have demonstrated the effectiveness of 3D resection specimen scanning for communicating margin results. We now address the corresponding surgical defect by debuting 3D defect models, which allow for accurate annotations of harvested supplemental margins. METHODS: Surgical defects were rendered into 3D models, which were annotated to document the precise location of harvested supplemental margins. 3D defect scans were also compared with routine 2D photography and were analyzed for quality, clarity, and the time required to complete the scan. RESULTS: Forty defects were scanned from procedures including segmental mandibulectomy, maxillectomy, and laryngopharyngectomy. Average duration of defect scan was 6 min, 45 s. In six of ten 2D photographs, the surgeon was unable to precisely annotate the extent of at least one supplemental margin. CONCLUSION: 3D defect scanning offers advantages in that this technique enables documentation of the precise location and breadth of supplemental margins harvested to address margins at-risk.

Pelvic Floor Pressures Differ Based on Location in the Pelvis and Body Position: A Cadaver Mode.

BACKGROUND: The pelvic floor is a bowl-shaped complex of multiple muscles and fascia, which functions to support the pelvic organs, and it aids in controlling continence. In pelvic floor disease, this complex becomes weakened or damaged leading to urinary, fecal incontinence, and pelvic organ prolapse. It is unclear whether the position of the body impacts the forces on the pelvic floor. PURPOSE: The primary objective of this work is to measure force applied to the pelvic floor of a cadaver in sitting, standing, supine, and control positions. The secondary objective is to map the forces across the pelvic floor. METHODS: An un-embalmed female cadaver without pelvic floor dysfunction was prepared for pelvic floor pressure measurement using a pressure sensory array placed on top of the pelvic floor, and urodynamic catheters were placed in the hollow of the sacrum, the retropubic space, and at the vaginal apex. Pressure measurements were recorded with the cadaver in the supine position, sitting cushioned without external pelvic floor support, and standing. Pressure array data were analyzed along with imaging of the cadaver. Together, these data were mapped into a three-dimensional reconstruction of the pressure points in pelvic floor and corresponding pelvic organs. RESULTS: pressures were higher at the symphysis than in the hollow of the sacrum in the standing position. Pressure array measurements were lowest in the standing position and highest in the sitting position. Three-dimensional reconstruction confirmed the location and accuracy of our measurements. CONCLUSIONS: The findings of increased pressures behind the symphysis are in line with the higher incidence of anterior compartment prolapse. Our findings support our hypothesis that the natural shape and orientation of the pelvis in the standing position shields the pelvic floor from downward forces of the viscera.

Advancing Osteoporosis Evaluation Procedures: Detailed Computational Analysis of Regional Structural Vulnerabilities in Osteoporotic Bone.

Osteoporotic fractures of the femur are associated with poor healing, disability, reduced quality of life, and high mortality rates within 1 year. Moreover, osteoporotic fractures of the femur are still considered to be an unsolved problem in orthopedic surgery. In order to more effectively identify osteoporosis-related fracture risk and develop advanced treatment approaches for femur fractures, it is necessary to acquire a greater understanding of how osteoporosis alters the diaphyseal structure and biomechanical characteristics. The current investigation uses computational analyses to comprehensively examine how femur structure and its associated properties differ between healthy and osteoporotic bones. The results indicate statistically significant differences in multiple geometric properties between healthy femurs and osteoporotic femurs. Additionally, localized disparities in the geometric properties are evident. Overall, this approach will be beneficial in the development of new diagnostic procedures for highly detailed patient-specific detection of fracture risk, for establishing novel injury prevention treatments, and for informing advanced surgical solutions.

The role of anatomical context in soft-tissue multi-organ segmentation of cadaveric non-contrast-enhanced whole body CT.

BACKGROUND: Cadaveric computed tomography (CT) image segmentation is a difficult task to solve, especially when applied to whole-body image volumes. Traditional algorithms require preprocessing using registration, or highly conserved organ morphologies. These requirements cannot be fulfilled by cadaveric specimens, so deep learning must be used to overcome this limitation. Further, the widespread use of 2D algorithms for volumetric data ignores the role of anatomical context. The use of 3D spatial context for volumetric segmentation of CT scans as well as the anatomical context required to optimize the segmentation has not been adequately explored. PURPOSE: To determine whether 2D slice-by-slice UNet algorithms or 3D volumetric UNet (VNet) algorithms provide a more effective method for segmenting 3D volumes, and to what extent anatomical context plays in the segmentation of soft-tissue organs in cadaveric, noncontrast-enhanced (NCE) CT. METHODS: We tested five CT segmentation algorithms: 2D UNets with and without 3D data augmentation (3D rotations) as well as VNets with three levels of anatomical context (implemented via image downsampling at 1X, 2X, and 3X) for their performance via 3D Dice coefficients, and Hausdorff distance calculations. The classifiers were trained to segment the kidneys and liver and the performance was evaluated using Dice coefficient and Hausdorff distance on the segmentation versus the ground truth annotation. RESULTS: Our results demonstrate that VNet algorithms perform significantly better ( p < 0.05 $p<0.05$ ) than 2D models. Among the VNet classifiers, those that use some level of image downsampling outperform (as calculated through Dice coefficients) the VNet without downsampling. Additionally, the optimal amount of downsampling depends on the target organ. CONCLUSIONS: Anatomical context is an important component of soft-tissue, multi-organ segmentation in cadaveric, NCE CT imaging of the whole body. Different amounts of anatomical contexts are optimal depending on the size, position, and surrounding tissue of the organ.

Developing Patient-Specific Statistical Reconstructions of Healthy Anatomical Structures to Improve Patient Outcomes.

There are still numerous problems with modern joint replacement prostheses, which negatively influence patient health and recovery. For example, it is especially important to avoid failures and complications following hip arthroplasty because the loss of hip joint function is commonly associated with increased demand on the healthcare system, reoperation, loss of independence, physical disability, and death. The current study uses hip arthroplasty as a model system to present a new strategy of computationally generating patient-specific statistical reconstructions of complete healthy anatomical structures from computed tomography (CT) scans of damaged anatomical structures. The 3D model morphological data were evaluated from damaged femurs repaired with prosthetic devices and the respective damaged femurs that had been restored using statistical reconstruction. The results from all morphological measurements (i.e., maximum femoral length, Hausdorff distance, femoral neck anteversion, length of rotational center divergence, and angle of inclination) indicated that the values of femurs repaired with traditional prostheses did not fall within the +/-3 standard deviations of the respective patient-specific healthy anatomical structures. These results demonstrate that there are quantitative differences in the morphology of femurs repaired with traditional prostheses and the morphology of patient-specific statistical reconstructions. This approach of generating patient-specific statistical reconstructions of healthy anatomical structures might help to inform prosthetic designs so that new prostheses more closely resemble natural healthy morphology and preserve biomechanical function. Additionally, the patient-specific statistical reconstructions of healthy anatomical structures might be valuable for surgeons in that prosthetic devices could be selected and positioned to more accurately restore natural biomechanical function. All in all, this contribution establishes the novel approach of generating patient-specific statistical reconstructions of healthy anatomical structures from the CT scans of individuals' damaged anatomical structures to improve treatments and patient outcomes.

Histology segmentation using active learning on regions of interest in oral cavity squamous cell carcinoma.

In digital pathology, deep learning has been shown to have a wide range of applications, from cancer grading to segmenting structures like glomeruli. One of the main hurdles for digital pathology to be truly effective is the size of the dataset needed for generalization to address the spectrum of possible morphologies. Small datasets limit classifiers' ability to generalize. Yet, when we move to larger datasets of whole slide images (WSIs) of tissue, these datasets may cause network bottlenecks as each WSI at its original magnification can be upwards of 100 000 by 100 000 pixels, and over a gigabyte in file size. Compounding this problem, high quality pathologist annotations are difficult to obtain, as the volume of necessary annotations to create a classifier that can generalize would be extremely costly in terms of pathologist-hours. In this work, we use Active Learning (AL), a process for iterative interactive training, to create a modified U-net classifier on the region of interest (ROI) scale. We then compare this to Random Learning (RL), where images for addition to the dataset for retraining are randomly selected. Our hypothesis is that AL shows benefits for generating segmentation results versus randomly selecting images to annotate. We show that after 3 iterations, that AL, with an average Dice coefficient of 0.461, outperforms RL, with an average Dice Coefficient of 0.375, by 0.086.

Tall cell variant papillary thyroid carcinoma impacts disease-free survival at the 10 % cut-point on multivariate analysis.

INTRODUCTION: The diagnosis of tall cell variant papillary thyroid carcinoma (TCV-PTC) corresponds to the feature of "aggressive histology" within the framework of the American Thyroid Association (ATA) Risk of Recurrence (ROR) guidelines. Using the current World Health Organization (WHO) definition for TCV-PTC (tall cells with height at least twice the width, distribution ≥ 30 %), we examined the impact of this diagnosis on disease-free survival (DFS). METHODS: The study cohort consisted of 347 patients treated for primary papillary thyroid carcinoma (PTC). Current ATA guidelines were followed for the extent of surgery and the administration of adjuvant radioiodine therapy. Clinical surveillance included ultrasound examination and biochemical parameters according to ATA standards. The outcome was measured as time from surgery to first disease recurrence (DR) versus time from surgery until the last documented disease-free encounter (no evidence of disease, NED). Disease-free patients with fewer than 6 months of follow-up were excluded from this cohort. Structural recurrences are documented by histology or cytology whereas biochemical recurrences are documented by rising serum thyroglobulin in the absence of structural disease. All slides on all patients were examined by two pathologists with the substantial interobserver agreement (Kappa = 73 %). The primary tumors are categorically classified either as (1) TCV-PTC (definition above), (2) Papillary thyroid carcinoma with tall cell features (PTC-TCF) (≥ 10 % < 30 % tall cells), or (3) Control (< 10 % tall cells). Tumor size is categorized as either (1) ≤ 10 mm, (2) 11-29 mm, or (3) ≥ 30 mm. Degree of ETE is categorized as either intrathyroidal, microscopic ETE, histologic spread to strap muscles, or pT4 disease. RESULTS: 185 patients are classified as TCV-PTC (≥ 30 % tall cells), 62 as PTC-TCF (≥ 10 % < 30 % tall cells), and 100 as control group (< 10 % tall cells). TCV-PTC is associated with ≥ 30 mm size (p = .0246) and invasion of strap muscles and/or pT4 (p = .0325). There was no relationship between TCV-PTC and aggressive lymph node (ALN) status defined by ATA. Overall follow-up ranged from two months (one patient death) to 203 months (mean 40.8, median 33.0). DR occurred in 61 patients (mean 31.4 months, range 0 -184, 59 structural recurrences, 2 biochemical recurrences). Three models for TCV-PTC were examined: Model 1 - Tall cells ≥ 10% versus control, Model 2 - TCV-PTC versus TCF-PTC versus control, and Model 3 - TCV-PTC versus control. Kaplan Meier curves demonstrated decreased DFS with ALN status (p = .0001), ETE (p = .0295), and TCV-PTC (Model 1, p = .041). On multivariate analysis, TCV-PTC (Model 1) remained significantly predictive when adjusted for ALN (p = .0059). ETE dropped out of the model. CONCLUSION: TCV-PTC is significantly associated with larger tumors and a greater degree of ETE. The diagnosis of TCV-PTC significantly impacts DFS at the 10 % cut-point on multivariate analysis.

Enhancing biomedical data validity with standardized segmentation finite element analysis.

Finite element analysis is a powerful computational technique for augmenting biomedical research, prosthetics design, and preoperative surgical assessment. However, the validity of biomechanical data obtained from finite element analysis is dependent on the quality of the preceding data processing. Until now, little information was available about the effect of the segmentation process on finite element models and biomechanical data. The current investigation applied 4 segmentation approaches to 129 femur specimens, yielding a total of 516 finite element models. Biomechanical data including average displacement, pressure, stress, and strain were collected from experimental groups based on the different segmentation approaches. The results indicate that only a 5.0% variation in the segmentation process leads to statistically significant differences in all 4 biomechanical measurements. These results suggest that it is crucial for consistent segmentation procedures to be applied to all specimens within a study. This methodological advancement will help to ensure that finite element data will be more accurate and that research conclusions will have greater validity.

Sleep and occupational well-being in active duty special operations soldiers: A replication and expansion.

OBJECTIVE: To assess the relationship between sleep quality and occupational well-being in active duty military Service Members. DESIGN: Longitudinal prospective analysis. SETTING: An annual military training event. PARTICIPANTS: US Army special operations Soldiers (n = 60; 100% male; age 25.41 ± 3.74). INTERVENTION: None. MEASUREMENTS: The Pittsburgh Sleep Quality Index (PSQI) was administered prior to the training event, and the Emotional Exhaustion Scale, the Role Overload Scale, the Walter Reed Army Institute of Research Soldier-Specific Functional Impairment Scale, and the Perceived Stress Scale were administered after the event. Linear regression models were used to assess the relationship between sleep and occupational wellness measures, and the outcome measures of "good" and "poor" sleepers (per the PSQI scoring criteria) were compared with Student's t tests. RESULTS: Higher (poorer) PSQI Global Scores predicted poorer occupational wellness of all measures (emotional exhaustion: B = 1.60, P < .001, R2 = 0.25; functional impairment: B = 0.29, P = .03, R2 = 0.14; role overload: B = 0.28, P = .008, R2 = 0.12; and perceived stress: B = 0.34, P = .004, R2 = 0.20). There were additional relationships between specific PSQI component scores and occupational wellness measures, which is a replication of This team's previous work. Furthermore, emotional exhaustion (t(58) = -4.18, P < .001), functional impairment (t(59)= -3.68, P = .001), role overload (t(58) = -3.20, P = .002), and perceived stress (t(58) = -2.43, P = .02) were all higher in poor sleepers. CONCLUSIONS: The findings of this study suggest that US Army special operations Soldiers who have poorer sleep quality may be at increased risk for having poorer occupational well-being.

Prevalence of Eosinophilic Granulomatosis With Polyangiitis and Associated Health Care Utilization Among Patients With Concomitant Asthma in US Commercial Claims Database.

OBJECTIVE: To estimate the prevalence and associated disease burden of eosinophilic granulomatosis with polyangiitis (EGPA) in patients with asthma from a US claims database. METHODS: Two cohorts were defined using enrollees (aged ≥18 years) from the Optum deidentified Clinformatics Datamart claims database 2010-2014, based on validated EGPA case definitions with varying specificity: EGPA 1 (main cohort; more specific; patients with 2 codes [in any combination] within 12 months of each other for eosinophilia, vasculitis, or mononeuritis multiplex) and EGPA 2 (sensitivity analysis cohort; less specific; patients with 2 codes of above conditions and/or neurologic symptoms within 12 months of each other). Patients had 3 or more asthma medications in the 12-month baseline before index date (date of the second code). Eosinophilic granulomatosis with polyangiitis prevalence, asthma severity during the baseline period, oral corticosteroid (OCS) use, and health care utilization during the 12-month follow-up period were determined. RESULTS: Overall, 88 and 604 patients were included in main cohort EGPA 1 and sensitivity analysis cohort EGPA 2, respectively; corresponding annual EGPA prevalence rates were 3.2 to 5.9 and 23.4 to 30.7 cases/million patients. Approximately 75% of patients were prescribed OCS and ~30% experienced 1 or more hospitalization; 75% in EGPA 1 and 52% in EGPA 2 with 1 or more non-OCS prescription in the 90 days before index date had severe asthma. CONCLUSIONS: Eosinophilic granulomatosis with polyangiitis prevalence estimates varied based on specificity of the case definition but were generally consistent with previous country-specific estimates. Despite differences in prevalence, both cohorts displayed a generally similar, high burden of OCS use and health care utilization, highlighting the substantial disease burden among patients with EGPA and the need for specific treatments.

Disrupting the spatio-temporal symmetry of the electron dynamics in atmospheric pressure plasmas by voltage waveform tailoring.

Single frequency, geometrically symmetric Radio-Frequency (rf) driven atmospheric pressure plasmas exhibit temporally and spatially symmetric patterns of electron heating, and consequently, charged particle densities and fluxes. Using a combination of phase-resolved optical emission spectroscopy and kinetic plasma simulations, we demonstrate that tailored voltage waveforms consisting of multiple rf harmonics induce targeted disruption of these symmetries. This confines the electron heating to small regions of time and space and enables the electron energy distribution function to be tailored.

Anti-IL-5 treatments in patients with severe asthma by blood eosinophil thresholds: Indirect treatment comparison.

BACKGROUND: Three anti-IL-5 pathway-directed therapies are approved for use in patients with severe eosinophilic asthma (SEA); however, no head-to-head comparison data are available. OBJECTIVE: We sought to compare the efficacy of licensed doses of mepolizumab, benralizumab, and reslizumab in patients with SEA, according to baseline blood eosinophil counts. METHODS: This indirect treatment comparison (ITC) used data from a Cochrane review and independent searches. Eligible studies were randomized controlled trials in patients aged 12 years or greater with SEA. End points included annualized rate of clinically significant exacerbations and change from baseline in Asthma Control Questionnaire score and FEV1. An ITC was performed in patients with Asthma Control Questionnaire scores of 1.5 or greater and stratified by baseline blood eosinophil count. RESULTS: Eleven studies were included. All treatments significantly reduced the rate of clinically significant exacerbations and improved asthma control versus placebo in all blood eosinophil count subgroups. Mepolizumab reduced clinically significant exacerbations by 34% to 45% versus benralizumab across subgroups (rate ratio ≥400 cells/µL: 0.55 [95% CI, 0.35-0.87]; ≥300 cells/µL: 0.61 [95% CI, 0.37-0.99]; and ≥150 cells/µL: 0.66 [95% CI, 0.49-0.89]; all P < .05) and by 45% versus reslizumab in the 400 cells/µL or greater subgroup (rate ratio, 0.55 [95% CI, 0.36-0.85]; P = .007). Asthma control was significantly improved with mepolizumab versus benralizumab (all subgroups: P < .05) and versus reslizumab in the 400 cells/µL or greater subgroup (P = .004). Benralizumab significantly improved lung function versus reslizumab in the 400 cells/µL or greater subgroup (P = .025). CONCLUSIONS: This ITC of the licensed doses suggests that mepolizumab was associated with significantly greater improvements in clinically significant exacerbations and asthma control compared with reslizumab or benralizumab in patients with similar blood eosinophil counts.

Quantitative nuclear histomorphometry predicts oncotype DX risk categories for early stage ER+ breast cancer.

BACKGROUND: Gene-expression companion diagnostic tests, such as the Oncotype DX test, assess the risk of early stage Estrogen receptor (ER) positive (+) breast cancers, and guide clinicians in the decision of whether or not to use chemotherapy. However, these tests are typically expensive, time consuming, and tissue-destructive. METHODS: In this paper, we evaluate the ability of computer-extracted nuclear morphology features from routine hematoxylin and eosin (H&E) stained images of 178 early stage ER+ breast cancer patients to predict corresponding risk categories derived using the Oncotype DX test. A total of 216 features corresponding to the nuclear shape and architecture categories from each of the pathologic images were extracted and four feature selection schemes: Ranksum, Principal Component Analysis with Variable Importance on Projection (PCA-VIP), Maximum-Relevance, Minimum Redundancy Mutual Information Difference (MRMR MID), and Maximum-Relevance, Minimum Redundancy - Mutual Information Quotient (MRMR MIQ), were employed to identify the most discriminating features. These features were employed to train 4 machine learning classifiers: Random Forest, Neural Network, Support Vector Machine, and Linear Discriminant Analysis, via 3-fold cross validation. RESULTS: The four sets of risk categories, and the top Area Under the receiver operating characteristic Curve (AUC) machine classifier performances were: 1) Low ODx and Low mBR grade vs. High ODx and High mBR grade (Low-Low vs. High-High) (AUC = 0.83), 2) Low ODx vs. High ODx (AUC = 0.72), 3) Low ODx vs. Intermediate and High ODx (AUC = 0.58), and 4) Low and Intermediate ODx vs. High ODx (AUC = 0.65). Trained models were tested independent validation set of 53 cases which comprised of Low and High ODx risk, and demonstrated per-patient accuracies ranging from 75 to 86%. CONCLUSION: Our results suggest that computerized image analysis of digitized H&E pathology images of early stage ER+ breast cancer might be able predict the corresponding Oncotype DX risk categories.

A resolution adaptive deep hierarchical (RADHicaL) learning scheme applied to nuclear segmentation of digital pathology images.

Deep learning (DL) has recently been successfully applied to a number of image analysis problems. However, DL approaches tend to be inefficient for segmentation on large image data, such as high-resolution digital pathology slide images. For example, typical breast biopsy images scanned at 40× magnification contain billions of pixels, of which usually only a small percentage belong to the class of interest. For a typical naïve deep learning scheme, parsing through and interrogating all the image pixels would represent hundreds if not thousands of hours of compute time using high performance computing environments. In this paper, we present a resolution adaptive deep hierarchical (RADHicaL) learning scheme wherein DL networks at lower resolutions are leveraged to determine if higher levels of magnification, and thus computation, are necessary to provide precise results. We evaluate our approach on a nuclear segmentation task with a cohort of 141 ER+ breast cancer images and show we can reduce computation time on average by about 85%. Expert annotations of 12,000 nuclei across these 141 images were employed for quantitative evaluation of RADHicaL. A head-to-head comparison with a naïve DL approach, operating solely at the highest magnification, yielded the following performance metrics: .9407 vs .9854 Detection Rate, .8218 vs .8489 F-score, .8061 vs .8364 true positive rate and .8822 vs 0.8932 positive predictive value. Our performance indices compare favourably with state of the art nuclear segmentation approaches for digital pathology images.

Developing the Quantitative Histopathology Image Ontology (QHIO): A case study using the hot spot detection problem.

Interoperability across data sets is a key challenge for quantitative histopathological imaging. There is a need for an ontology that can support effective merging of pathological image data with associated clinical and demographic data. To foster organized, cross-disciplinary, information-driven collaborations in the pathological imaging field, we propose to develop an ontology to represent imaging data and methods used in pathological imaging and analysis, and call it Quantitative Histopathological Imaging Ontology - QHIO. We apply QHIO to breast cancer hot-spot detection with the goal of enhancing reliability of detection by promoting the sharing of data between image analysts.

Mentor's brain functional connectivity network during robotic assisted surgery mentorship.

In many complicated cognitive-motor tasks mentoring is inevitable during the learning process. Although mentors are expert in doing the task, trainee's operation might be new for a mentor. This makes mentoring a very difficult task which demands not only the knowledge and experience of a mentor, but also his/her ability to follow trainee's movements and patiently advise him/her during the operation. We hypothesize that information binding throughout the mentor's brain areas, contributed to the task, changes as the expertise level of the trainee improves from novice to intermediate and expert. This can result in the change of mentor's level of satisfaction. The brain functional connectivity network is extracted by using brain activity of a mentor during mentoring novice and intermediate surgeons, watching expert surgeon operation, and doing Urethrovesical Anasthomosis (UVA) procedure by himself. By using the extracted network, we investigate the role of modularity and neural activity efficiency in mentoring. Brain activity is measured by using a 24-channel ABM Neuro-headset with the frequency of 256 Hz. One mentor operates 26 UVA procedures and three trainees with the expertise level of novice, intermediate, and expert perform 26 UVA procedures under the supervision of mentor. Our results indicate that the modularity of functional connectivity network is higher when mentor performs the task or watches the expert operation comparing mentoring the novice and intermediate surgeons. At the end of each operation, mentor subjectively assesses the quality of operation by giving scores to NASA-TLX indexes. Performance score is used to discuss our results. The extracted significant positive correlation between performance level and modularity (r = 0.38, p - value <; 0.005) shows the increase of automaticity and decrease in neural activity cost by improving the performance.