Search

Background and Objectives: This study identified a clinically significant subset of patients with glioma with tumor outside of contrast enhancement present at autopsy and subsequently developed a method for detecting nonenhancing tumor using radio-pathomic mapping. We tested the hypothesis that autopsy-based radio-pathomic tumor probability maps would be able to noninvasively identify areas of infiltrative tumor beyond traditional imaging signatures., Methods: A total of 159 tissue samples from 65 subjects were aligned to MRI acquired nearest to death for this retrospective study. Demographic and survival characteristics for patients with and without tumor beyond the contrast-enhancing margin were computed. An ensemble algorithm was used to predict pixelwise tumor presence from pathological annotations using segmented cellularity (Cell), extracellular fluid, and cytoplasm density as input (6 train/3 test subjects). A second level of ensemble algorithms was used to predict voxelwise Cell, extracellular fluid, and cytoplasm on the full data set (43 train/22 test subjects) using 5-by-5 voxel tiles from T1, T1 + C, fluid-attenuated inversion recovery, and apparent diffusion coefficient as input. The models were then combined to generate noninvasive whole brain maps of tumor probability., Results: Tumor outside of contrast was identified in 41.5% of patients, who showed worse survival outcomes (hazard ratio = 3.90, P < .001). Tumor probability maps reliably tracked nonenhancing tumor on a range of local and external unseen data, identifying tumor outside of contrast in 69% of presurgical cases that also showed reduced survival outcomes (hazard ratio = 1.67, P = .027)., Conclusion: This study developed a multistage model for mapping gliomas using autopsy tissue samples as ground truth, which was able to identify regions of tumor beyond traditional imaging signatures., (Copyright © 2024 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the Congress of Neurological Surgeons.)

Background: Autopsy-based radio-pathomic maps of glioma pathology have shown substantial promise inidentifying areas of non-enhancing tumor presence, which may be able to differentiate subsets of patients that respond favorably to treatments such as bevacizumab that have shown mixed efficacy evidence. We tested the hypthesis that phenotypes of non-enhancing tumor fronts can distinguish between glioblastoma patients that will respond favorably to bevacizumab and will visually capture treatment response., Methods: T1, T1C, FLAIR, and ADC images were used to generate radio-pathomic maps of tumor characteristics for 79 pre-treatment patients with a primary GBM or high-grade IDH1-mutant astrocytoma for this study. Novel phenotyping (hypercellular, hypocellular, hybrid, or well-circumscribed front) of the non-enhancing tumor front was performed on each case. Kaplan Meier analyses were then used to assess differences in survival and bevacizumab efficacy between phenotypes. Phenotype compartment segmentations generated longitudinally for a subset of 26 patients over the course of bevacizumab treatment, where a mixed effect model was used to detect longitudinal changes., Results: Well-Circumscribed patients showed significant/trending increases in survival compared to Hypercellular Front (HR = 2.0, p = 0.05), Hypocellular Front (HR = 2.02, p = 0.03), and Hybrid Front tumors (HR = 1.75, p = 0.09). Only patients with hypocellular or hybrid fronts showed significant survival benefits from bevacizumab treatment (HR = 2.35, p = 0.02; and HR = 2.45, p = 0.03, respectively). Hypocellular volumes decreased by an average 50.52 mm 3 per day of bevacizumab treatment (p = 0.002)., Conclusion: Patients with a hypocellular tumor front identified by radio-pathomic maps showed improved treatment efficacy when treated with bevacizumab, and reducing hypocellular volumes over the course of treatment may indicate treatment response., (© 2024. The Author(s).)

Prostate cancer is the most commonly diagnosed cancer in men, accounting for 27% of the new male cancer diagnoses in 2022. If organ-confined, removal of the prostate through radical prostatectomy is considered curative; however, distant metastases may occur, resulting in a poor patient prognosis. This study sought to determine whether quantitative pathomic features of prostate cancer differ in patients who biochemically experience biological recurrence after surgery. Whole-mount prostate histology from 78 patients was analyzed for this study. In total, 614 slides were hematoxylin and eosin stained and digitized to produce whole slide images (WSI). Regions of differing Gleason patterns were digitally annotated by a genitourinary fellowship-trained pathologist, and high-resolution tiles were extracted from each annotated region of interest for further analysis. Individual glands within the prostate were identified using automated image processing algorithms, and histomorphometric features were calculated on a per-tile basis and across WSI and averaged by patients. Tiles were organized into cancer and benign tissues. Logistic regression models were fit to assess the predictive value of the calculated pathomic features across tile groups and WSI; additionally, models using clinical information were used for comparisons. Logistic regression classified each pathomic feature model at accuracies >80% with areas under the curve of 0.82, 0.76, 0.75, and 0.72 for all tiles, cancer only, noncancer only, and across WSI. This was comparable with standard clinical information, Gleason Grade Groups, and CAPRA score, which achieved similar accuracies but areas under the curve of 0.80, 0.77, and 0.70, respectively. This study demonstrates that the use of quantitative pathomic features calculated from digital histology of prostate cancer may provide clinicians with additional information beyond the traditional qualitative pathologist assessment. Further research is warranted to determine possible inclusion in treatment guidance., (Copyright © 2023 The Authors. Published by Elsevier Inc. All rights reserved.)

Prostate cancer (PCa) is the most diagnosed non-cutaneous cancer in men. Despite therapies such as radical prostatectomy, which is considered curative, distant metastases may form, resulting in biochemical recurrence (BCR). This study used radiomic features calculated from multi-parametric magnetic resonance imaging (MP-MRI) to evaluate their ability to predict BCR and PCa presence. Data from a total of 279 patients, of which 46 experienced BCR, undergoing MP-MRI prior to surgery were assessed for this study. After surgery, the prostate was sectioned using patient-specific 3D-printed slicing jigs modeled using the T2-weighted imaging (T2WI). Sectioned tissue was stained, digitized, and annotated by a GU-fellowship trained pathologist for cancer presence. Digitized slides and annotations were co-registered to the T2WI and radiomic features were calculated across the whole prostate and cancerous lesions. A tree regression model was fitted to assess the ability of radiomic features to predict BCR, and a tree classification model was fitted with the same radiomic features to classify regions of cancer. We found that 10 radiomic features predicted eventual BCR with an AUC of 0.97 and classified cancer at an accuracy of 89.9%. This study showcases the application of a radiomic feature-based tool to screen for the presence of prostate cancer and assess patient prognosis, as determined by biochemical recurrence.

Purpose: Digital pathology is becoming an increasingly popular area of advancement in both research and clinically. Pathologists are now able to manage and interpret slides digitally, as well as collaborate with external pathologists with digital copies of slides. Differences in slide scanners include variation in resolution, image contrast, and optical properties, which may influence downstream image processing. This study tested the hypothesis that varying slide scanners would result in differences in computed pathomic features on prostate cancer whole mount slides., Design: This study collected 192 unique tissue slides from 30 patients following prostatectomy. Tissue samples were paraffin-embedded, stained for hematoxylin and eosin (H&E), and digitized using 3 different scanning microscopes at the highest available magnification rate, for a total of 3 digitized slides per tissue slide. These scanners included a ( S1 ) Nikon microscope equipped with an automated sliding stage, an ( S2 ) Olympus VS120 slide scanner, and a ( S3 ) Huron TissueScope LE scanner. A color deconvolution algorithm was then used to optimize contrast by projecting the RGB image into color channels representing optical stain density. The resulting intensity standardized images were then computationally processed to segment tissue and calculate pathomic features including lumen, stroma, epithelium, and epithelial cell density, as well as second-order features including lumen area and roundness; epithelial area, roundness, and wall thickness; and cell fraction. For each tested feature, mean values of that feature per digitized slide were collected and compared across slide scanners using mixed effect models, fit to compare differences in the tested feature associated with all slide scanners for each slide, including a random effect of subject with a nested random effect of slide to account for repeated measures. Similar models were also computed for tissue densities to examine how differences in scanner impact downstream processing., Results: Each mean color channel intensity (i.e., Red, Green, Blue) differed between slide scanners (all P <.001). Of the color deconvolved images, only the hematoxylin channel was similar in all 3 scanners (all P >.05). Lumen and stroma densities between S3 and S1 slides, and epithelial cell density between S3 and S2 ( P >.05) were comparable but all other comparisons were significantly different ( P <.05). The second-order features were found to be comparable for all scanner comparisons, except for lumen area and epithelium area., Conclusion: This study demonstrates that both optical and computed properties of digitized histological samples are impacted by slide scanner differences. Future research is warranted to better understand which scanner properties influence the tissue segmentation process and to develop harmonization techniques for comparing data across multiple slide scanners., Competing Interests: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (© 2023 The Authors. Published by Elsevier Inc. on behalf of Association for Pathology Informatics.)

One in eight men will be affected by prostate cancer (PCa) in their lives. While the current clinical standard prognostic marker for PCa is the Gleason score, it is subject to inter-reviewer variability. This study compares two machine learning methods for discriminating between cancerous regions on digitized histology from 47 PCa patients. Whole-slide images were annotated by a GU fellowship-trained pathologist for each Gleason pattern. High-resolution tiles were extracted from annotated and unlabeled tissue. Patients were separated into a training set of 31 patients (Cohort A, n = 9345 tiles) and a testing cohort of 16 patients (Cohort B, n = 4375 tiles). Tiles from Cohort A were used to train a ResNet model, and glands from these tiles were segmented to calculate pathomic features to train a bagged ensemble model to discriminate tumors as (1) cancer and noncancer, (2) high- and low-grade cancer from noncancer, and (3) all Gleason patterns. The outputs of these models were compared to ground-truth pathologist annotations. The ensemble and ResNet models had overall accuracies of 89% and 88%, respectively, at predicting cancer from noncancer. The ResNet model was additionally able to differentiate Gleason patterns on data from Cohort B while the ensemble model was not. Our results suggest that quantitative pathomic features calculated from PCa histology can distinguish regions of cancer; however, texture features captured by deep learning frameworks better differentiate unique Gleason patterns., Competing Interests: The authors have declared that no competing interests exist., (Copyright: © 2023 Duenweg et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.)

Background: Diffusion-weighted imaging (DWI) is commonly used to detect prostate cancer, and a major clinical challenge is differentiating aggressive from indolent disease., Purpose: To compare 14 site-specific parametric fitting implementations applied to the same dataset of whole-mount pathologically validated DWI to test the hypothesis that cancer differentiation varies with different fitting algorithms., Study Type: Prospective., Population: Thirty-three patients prospectively imaged prior to prostatectomy., Field Strength/sequence: 3 T, field-of-view optimized and constrained undistorted single-shot DWI sequence., Assessment: Datasets, including a noise-free digital reference object (DRO), were distributed to the 14 teams, where locally implemented DWI parameter maps were calculated, including mono-exponential apparent diffusion coefficient (MEADC), kurtosis (K), diffusion kurtosis (DK), bi-exponential diffusion (BID), pseudo-diffusion (BID*), and perfusion fraction (F). The resulting parametric maps were centrally analyzed, where differentiation of benign from cancerous tissue was compared between DWI parameters and the fitting algorithms with a receiver operating characteristic area under the curve (ROC AUC)., Statistical Test: Levene's test, P < 0.05 corrected for multiple comparisons was considered statistically significant., Results: The DRO results indicated minimal discordance between sites. Comparison across sites indicated that K, DK, and MEADC had significantly higher prostate cancer detection capability (AUC range = 0.72-0.76, 0.76-0.81, and 0.76-0.80 respectively) as compared to bi-exponential parameters (BID, BID*, F) which had lower AUC and greater between site variation (AUC range = 0.53-0.80, 0.51-0.81, and 0.52-0.80 respectively). Post-processing parameters also affected the resulting AUC, moving from, for example, 0.75 to 0.87 for MEADC varying cluster size., Data Conclusion: We found that conventional diffusion models had consistent performance at differentiating prostate cancer from benign tissue. Our results also indicated that post-processing decisions on DWI data can affect sensitivity and specificity when applied to radiological-pathological studies in prostate cancer., Level of Evidence: 1 TECHNICAL EFFICACY: Stage 3., (© 2021 The Authors. Journal of Magnetic Resonance Imaging published by Wiley Periodicals LLC on behalf of International Society for Magnetic Resonance in Medicine.)

Background and Purpose: Currently, contrast-enhancing margins on T1WI are used to guide treatment of gliomas, yet tumor invasion beyond the contrast-enhancing region is a known confounding factor. Therefore, this study used postmortem tissue samples aligned with clinically acquired MRIs to quantify the relationship between intensity values and cellularity as well as to develop a radio-pathomic model to predict cellularity using MR imaging data., Materials and Methods: This single-institution study used 93 samples collected at postmortem examination from 44 patients with brain cancer. Tissue samples were processed, stained with H&E, and digitized for nuclei segmentation and cell density calculation. Pre- and postgadolinium contrast T1WI, T2 FLAIR, and ADC images were collected from each patient's final acquisition before death. In-house software was used to align tissue samples to the FLAIR image via manually defined control points. Mixed-effects models were used to assess the relationship between single-image intensity and cellularity for each image. An ensemble learner was trained to predict cellularity using 5 × 5 voxel tiles from each image, with a two-thirds to one-third train-test split for validation., Results: Single-image analyses found subtle associations between image intensity and cellularity, with a less pronounced relationship in patients with glioblastoma. The radio-pathomic model accurately predicted cellularity in the test set (root mean squared error = 1015 cells/mm 2 ) and identified regions of hypercellularity beyond the contrast-enhancing region., Conclusions: A radio-pathomic model for cellularity trained with tissue samples acquired at postmortem examination is able to identify regions of hypercellular tumor beyond traditional imaging signatures., (© 2022 by American Journal of Neuroradiology.)

The presence and extent of cribriform patterned Gleason 4 (G4) glands are associated with poor prognosis following radical prostatectomy. This study used whole-mount prostate histology and multiparametric magnetic resonance imaging (MP-MRI) to evaluate diffusion differences in G4 gland morphology. Fourty-eight patients underwent MP-MRI prior to prostatectomy, of whom 22 patients had regions of both G4 cribriform glands and G4 fused glands (G4CG and G4FG, respectively). After surgery, the prostate was sliced using custom, patient-specific 3D-printed slicing jigs modeled according to the T2-weighted MR image, processed, and embedded in paraffin. Whole-mount hematoxylin and eosin-stained slides were annotated by our urologic pathologist and digitally contoured to differentiate the lumen, epithelium, and stroma. Digitized slides were co-registered to the T2-weighted MRI scan. Linear mixed models were fitted to the MP-MRI data to consider the different hierarchical structures at the patient and slide level. We found that Gleason 4 cribriform glands were more diffusion-restricted than fused glands.

Objectives: To determine the acute and early long-term associations of sport-related concussion (SRC) and subcortical and cortical structures in collegiate contact sport athletes., Methods: Athletes with a recent SRC (n=99) and matched contact (n=91) and non-contact sport controls (n=95) completed up to four neuroimaging sessions from 24 to 48 hours to 6 months postinjury. Subcortical volumes (amygdala, hippocampus, thalamus and dorsal striatum) and vertex-wise measurements of cortical thickness/volume were computed using FreeSurfer. Linear mixed-effects models examined the acute and longitudinal associations between concussion and structural metrics, controlling for intracranial volume (or mean thickness) and demographic variables (including prior concussions and sport exposure)., Results: There were significant group-dependent changes in amygdala volumes across visits (p=0.041); this effect was driven by a trend for increased amygdala volume at 6 months relative to subacute visits in contact controls, with no differences in athletes with SRC. No differences were observed in any cortical metric (ie, thickness or volume) for primary or secondary analyses., Conclusion: A single SRC had minimal associations with grey matter structure across a 6-month time frame., Competing Interests: Competing interests: None declared., (© Author(s) (or their employer(s)) 2021. No commercial re-use. See rights and permissions. Published by BMJ.)

Purpose: Prostate cancer primarily arises from the glandular epithelium. Histomophometric techniques have been used to assess the glandular epithelium in automated detection and classification pipelines; however, they are often rigid in their implementation, and their performance suffers on large datasets where variation in staining, imaging, and preparation is difficult to control. The purpose of this study is to quantify performance of a pixelwise segmentation algorithm that was trained using different combinations of weak and strong stroma, epithelium, and lumen labels in a prostate histology dataset. Approach: We have combined weakly labeled datasets generated using simple morphometric techniques and high-quality labeled datasets from human observers in prostate biopsy cores to train a convolutional neural network for use in whole mount prostate labeling pipelines. With trained networks, we characterize pixelwise segmentation of stromal, epithelium, and lumen (SEL) regions on both biopsy core and whole-mount H&E-stained tissue. Results: We provide evidence that by simply training a deep learning algorithm on weakly labeled data generated from rigid morphometric methods, we can improve the robustness of classification over the morphometric methods used to train the classifier. Conclusions: We show that not only does our approach of combining weak and strong labels for training the CNN improve qualitative SEL labeling within tissue but also the deep learning generated labels are superior for cancer classification in a higher-order algorithm over the morphometrically derived labels it was trained on., (© The Authors. Published by SPIE under a Creative Commons Attribution 4.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.)

Prior studies have reported long-term differences in brain structure (brain morphometry) as being associated with cumulative concussion and contact sport participation. There is emerging evidence to suggest that similar effects of prior concussion and contact sport participation on brain morphometry may be present in younger cohorts of active athletes. We investigated the relationship between prior concussion and primary sport participation with subcortical and cortical structures in active collegiate contact sport and non-contact sport athletes. Contact sport athletes (CS; N = 190) and matched non-contact sport athletes (NCS; N = 95) completed baseline clinical testing and participated in up to four serial neuroimaging sessions across a 6-months period. Subcortical and cortical structural metrics were derived using FreeSurfer. Linear mixed-effects (LME) models examined the effects of years of primary sport participation and prior concussion (0, 1+) on brain structure and baseline clinical variables. Athletes with prior concussion across both groups reported significantly more baseline concussion and psychological symptoms (all p s < 0.05). The relationship between years of primary sport participation and thalamic volume differed between CS and NCS ( p = 0.015), driven by a significant inverse association between primary years of participation and thalamic volume in CS ( p = 0.007). Additional analyses limited to CS alone showed that the relationship between years of primary sport participation and dorsal striatal volume was moderated by concussion history ( p = 0.042). Finally, CS with prior concussion had larger hippocampal volumes than CS without prior concussion ( p = 0.015). Years of contact sport exposure and prior concussion(s) are associated with differences in subcortical volumes in young-adult, active collegiate athletes, consistent with prior literature in retired, primarily symptomatic contact sport athletes. Longitudinal follow-up studies in these athletes are needed to determine clinical significance of current findings., (Copyright © 2020 Brett, Bobholz, España, Huber, Mayer, Harezlak, Broglio, McAllister, McCrea, Meier and CARE Consortium Investigators.)

Magnetic resonance (MR)-derived radiomic features have shown substantial predictive utility in modeling different prognostic factors of glioblastoma and other brain cancers. However, the biological relationship underpinning these predictive models has been largely unstudied, and the generalizability of these models had been called into question. Here, we examine the localized relationship between MR-derived radiomic features and histology-derived "histomic" features using a data set of 16 patients with brain cancer. Tile-based radiomic features were collected on T1, post-contrast T1, FLAIR, and diffusion-weighted imaging (DWI)-derived apparent diffusion coefficient (ADC) images acquired before patient death, with analogous histomic features collected for autopsy samples coregistered to the magnetic resonance imaging. Features were collected for each original image, as well as a 3D wavelet decomposition of each image, resulting in 837 features per MR and histology image. Correlative analyses were used to assess the degree of association between radiomic-histomic pairs for each magnetic resonance imaging. The influence of several confounds was also assessed using linear mixed-effect models for the normalized radiomic-histomic distance, testing for main effects of different acquisition field strengths. Results as a whole were largely heterogeneous, but several features showed substantial associations with their histomic analogs, particularly those derived from the FLAIR and postcontrast T1W images. These features with the strongest association typically presented as stable across field strengths as well. These data suggest that a subset of radiomic features can consistently capture texture information on underlying tissue histology., Competing Interests: Conflict of Interest: None reported., (© 2020 The Authors. Published by Grapho Publications, LLC.)

Background: Corpora amylacea (CAM), in benign prostatic acini, contain acute-phase proteins. Do CAM coincide with carcinoma?, Methods: Within 270 biopsies, 83 prostatectomies, and 33 transurethral resections (TURs), CAM absence was designated CAM 0; corpora in less than 5% of benign acini: CAM 1; in 5% to 25%: CAM 2; in more than 25%: CAM 3. CAM were compared against carcinoma presence, clinicopathologic findings, and grade groups (GG) 1 to 2 vs 3 to 5. The frequency of CAM according to anatomic zone was counted. A pilot study was conducted using paired initial benign and repeat biopsies (33 benign, 24 carcinoma)., Results: A total of 68.9% of biopsies, 96.4% of prostatectomies, and 66.7% of TURs disclosed CAM. CAM ≥1 was common at an older age (P = .019). In biopsies, 204 cases (75%) had carcinoma; and CAM of 2 to 3 (compared to 0-1) were recorded in 25.0% of carcinomas but only 7.4% of benign biopsies (P = .005; odds ratio [OR] = 5.1). CAM correlated with high percent Gleason pattern 3, low GG (P = .035), and chronic inflammation (CI). CI correlated inversely with carcinoma (P = .003). CAM disclosed no association with race, body mass index, serum prostate specific antigen (PSA), acute inflammation (in biopsies), atrophy, or carcinoma volume. With CAM 1, the odds of GG 3 to 5 carcinoma, by comparison to CAM 0, decreased more than 2× (OR = 0.48; P = .032), with CAM 2, more than 3× (OR = 0.33; P = .005), and with CAM 3, almost 3× (OR = 0.39, P = .086). For men aged less than 65, carcinoma predictive model was: Score = (2 × age) + (5 × PSA) - (20 × degree of CAM); using our data, area under the ROC curve was 78.17%. When the transition zone was involved by cancer, it showed more CAM than in cases where it was uninvolved (P = .012); otherwise zonal distributions were similar. In the pilot study, CAM ≥1 predicted carcinoma on repeat biopsy (P < .05; OR = 8), as did CAM 2 to 3 (P < .0001; OR = 30). CI was not significant, and CAM retained significance after adjusting for CI., Conclusion: CAM correlate with carcinoma. Whether abundant CAM in benign biopsies adds value amidst high clinical suspicion, warrants further study., (© 2020 Wiley Periodicals, Inc.)

This study investigated the association between processing speed and cortical morphometry in children with idiopathic epilepsies (n = 81) versus healthy controls (n = 57), age 8-18. Participants underwent 1.5 T MRI scanning and cognitive testing including assessment of psychomotor speed (Digit Symbol) at or near the time of epilepsy diagnosis. Vertex analyses of cortical volume, thickness, surface area, and local gyrification index (LGI), as well as volume-based analyses of subcortical structures and cerebellum, were used to determine the morphometric correlates of Digit Symbol performance. Group comparisons revealed that the epilepsy and control groups exhibited different patterns of morphometric association with Digit Symbol performance - controls exhibited several areas of correlation between LGI and psychomotor speed, whereas participants with focal epilepsies exhibited different areas of correlation in different directions, and participants with generalized epilepsy exhibited no correlations. The other cortical morphometric measures showed no regions of significant correlation with Digit Symbol performance. In addition, cerebellum and brain stem volumes correlated with Digit Symbol performance in the control group, but not in epilepsy patients. These results suggest that LGI analysis is able to capture nuanced relationships between features of cortical and subcortical morphology with psychomotor speed, these relationships disrupted in different ways in children with epilepsy.