Artificial Intelligence-based Analytics for Diagnosis of Small Bowel Enteropathies and Black Box Feature Detection.

OBJECTIVES: Striking histopathological overlap between distinct but related conditions poses a disease diagnostic challenge. There is a major clinical need to develop computational methods enabling clinicians to translate heterogeneous biomedical images into accurate and quantitative diagnostics. This need is particularly salient with small bowel enteropathies; environmental enteropathy (EE) and celiac disease (CD). We built upon our preliminary analysis by developing an artificial intelligence (AI)-based image analysis platform utilizing deep learning convolutional neural networks (CNNs) for these enteropathies. METHODS: Data for the secondary analysis was obtained from three primary studies at different sites. The image analysis platform for EE and CD was developed using CNNs including one with multizoom architecture. Gradient-weighted class activation mappings (Grad-CAMs) were used to visualize the models' decision-making process for classifying each disease. A team of medical experts simultaneously reviewed the stain color normalized images done for bias reduction and Grad-CAMs to confirm structural preservation and biomedical relevance, respectively. RESULTS: Four hundred and sixty-one high-resolution biopsy images from 150 children were acquired. Median age (interquartile range) was 37.5 (19.0-121.5) months with a roughly equal sex distribution; 77 males (51.3%). ResNet50 and shallow CNN demonstrated 98% and 96% case-detection accuracy, respectively, which increased to 98.3% with an ensemble. Grad-CAMs demonstrated models' ability to learn different microscopic morphological features for EE, CD, and controls. CONCLUSIONS: Our AI-based image analysis platform demonstrated high classification accuracy for small bowel enteropathies which was capable of identifying biologically relevant microscopic features and emulating human pathologist decision-making process. Grad-CAMs illuminated the otherwise "black box" of deep learning in medicine, allowing for increased physician confidence in adopting these new technologies in clinical practice.

Artificial Intelligence and Its Role in Identifying Esophageal Neoplasia.

Randomized trials have demonstrated that ablation of dysplastic Barrett's esophagus can reduce the risk of progression to cancer. Endoscopic resection for early stage esophageal adenocarcinoma and squamous cell carcinoma can significantly reduce postoperative morbidity compared to esophagectomy. Unfortunately, current endoscopic surveillance technologies (e.g., high-definition white light, electronic, and dye-based chromoendoscopy) lack sensitivity at identifying subtle areas of dysplasia and cancer. Random biopsies sample only approximately 5% of the esophageal mucosa at risk, and there is poor agreement among pathologists in identifying low-grade dysplasia. Machine-based deep learning medical image and video assessment technologies have progressed significantly in recent years, enabled in large part by advances in computer processing capabilities. In deep learning, sequential layers allow models to transform input data (e.g., pixels for imaging data) into a composite representation that allows for classification and feature identification. Several publications have attempted to use this technology to help identify dysplasia and early esophageal cancer. The aims of this reviews are as follows: (a) discussing limitations in our current strategies to identify esophageal dysplasia and cancer, (b) explaining the concepts behind deep learning and convolutional neural networks using language appropriate for clinicians without an engineering background, (c) systematically reviewing the literature for studies that have used deep learning to identify esophageal neoplasia, and (d) based on the systemic review, outlining strategies on further work necessary before these technologies are ready for "prime-time," i.e., use in routine clinical care.

The intersection of video capsule endoscopy and artificial intelligence: addressing unique challenges using machine learning.

Introduction: Technical burdens and time-intensive review processes limit the practical utility of video capsule endoscopy (VCE). Artificial intelligence (AI) is poised to address these limitations, but the intersection of AI and VCE reveals challenges that must first be overcome. We identified five challenges to address. Challenge #1: VCE data are stochastic and contains significant artifact. Challenge #2: VCE interpretation is cost-intensive. Challenge #3: VCE data are inherently imbalanced. Challenge #4: Existing VCE AIMLT are computationally cumbersome. Challenge #5: Clinicians are hesitant to accept AIMLT that cannot explain their process. Methods: An anatomic landmark detection model was used to test the application of convolutional neural networks (CNNs) to the task of classifying VCE data. We also created a tool that assists in expert annotation of VCE data. We then created more elaborate models using different approaches including a multi-frame approach, a CNN based on graph representation, and a few-shot approach based on meta-learning. Results: When used on full-length VCE footage, CNNs accurately identified anatomic landmarks (99.1%), with gradient weighted-class activation mapping showing the parts of each frame that the CNN used to make its decision. The graph CNN with weakly supervised learning (accuracy 89.9%, sensitivity of 91.1%), the few-shot model (accuracy 90.8%, precision 91.4%, sensitivity 90.9%), and the multi-frame model (accuracy 97.5%, precision 91.5%, sensitivity 94.8%) performed well. Discussion: Each of these five challenges is addressed, in part, by one of our AI-based models. Our goal of producing high performance using lightweight models that aim to improve clinician confidence was achieved.

Deep learning systems detect dysplasia with human-like accuracy using histopathology and probe-based confocal laser endomicroscopy.

Probe-based confocal laser endomicroscopy (pCLE) allows for real-time diagnosis of dysplasia and cancer in Barrett's esophagus (BE) but is limited by low sensitivity. Even the gold standard of histopathology is hindered by poor agreement between pathologists. We deployed deep-learning-based image and video analysis in order to improve diagnostic accuracy of pCLE videos and biopsy images. Blinded experts categorized biopsies and pCLE videos as squamous, non-dysplastic BE, or dysplasia/cancer, and deep learning models were trained to classify the data into these three categories. Biopsy classification was conducted using two distinct approaches-a patch-level model and a whole-slide-image-level model. Gradient-weighted class activation maps (Grad-CAMs) were extracted from pCLE and biopsy models in order to determine tissue structures deemed relevant by the models. 1970 pCLE videos, 897,931 biopsy patches, and 387 whole-slide images were used to train, test, and validate the models. In pCLE analysis, models achieved a high sensitivity for dysplasia (71%) and an overall accuracy of 90% for all classes. For biopsies at the patch level, the model achieved a sensitivity of 72% for dysplasia and an overall accuracy of 90%. The whole-slide-image-level model achieved a sensitivity of 90% for dysplasia and 94% overall accuracy. Grad-CAMs for all models showed activation in medically relevant tissue regions. Our deep learning models achieved high diagnostic accuracy for both pCLE-based and histopathologic diagnosis of esophageal dysplasia and its precursors, similar to human accuracy in prior studies. These machine learning approaches may improve accuracy and efficiency of current screening protocols.

SCREENING FOR BARRETT'S ESOPHAGUS WITH PROBE-BASED CONFOCAL LASER ENDOMICROSCOPY VIDEOS.

Histologic diagnosis of Barrett's esophagus and esophageal malignancy via probe-based confocal laser endomicroscopy (pCLE) allows for real-time examination of epithelial architecture and targeted biopsy sampling. Although pCLE demonstrates high specificity, sensitivity remains low. This study employs deep learning architectures in order to improve the accuracy of pCLE in diagnosing esophageal cancer and its precursors. pCLE videos are curated and annotated as belonging to one of the three classes: squamous, Barrett's (intestinal metaplasia without dysplasia), or dysplasia. We introduce two novel video architectures, AttentionPooling and Multi-Module AttentionPooling deep networks, that outperform other models and demonstrate a high degree of explainability.

Semi-Supervised Classification of Noisy, Gigapixel Histology Images.

One of the greatest obstacles in the adoption of deep neural networks for new medical applications is that training these models typically require a large amount of manually labeled training samples. In this body of work, we investigate the semi-supervised scenario where one has access to large amounts of unlabeled data and only a few labeled samples. We study the performance of MixMatch and FixMatch-two popular semi-supervised learning methods-on a histology dataset. More specifically, we study these models' impact under a highly noisy and imbalanced setting. The findings here motivate the development of semi-supervised methods to ameliorate problems commonly encountered in medical data applications.

Deep Learning for Whole-Slide Tissue Histopathology Classification: A Comparative Study in the Identification of Dysplastic and Non-Dysplastic Barrett's Esophagus.

The gold standard of histopathology for the diagnosis of Barrett's esophagus (BE) is hindered by inter-observer variability among gastrointestinal pathologists. Deep learning-based approaches have shown promising results in the analysis of whole-slide tissue histopathology images (WSIs). We performed a comparative study to elucidate the characteristics and behaviors of different deep learning-based feature representation approaches for the WSI-based diagnosis of diseased esophageal architectures, namely, dysplastic and non-dysplastic BE. The results showed that if appropriate settings are chosen, the unsupervised feature representation approach is capable of extracting more relevant image features from WSIs to classify and locate the precursors of esophageal cancer compared to weakly supervised and fully supervised approaches.

Pathobiome driven gut inflammation in Pakistani children with Environmental Enteric Dysfunction.

Environmental Enteric Dysfunction (EED) is an acquired small intestinal inflammatory condition underlying high rates of stunting in children <5 years of age in low- and middle-income countries. Children with EED are known to have repeated exposures to enteropathogens and environmental toxins that leads to malabsorptive syndrome. We aimed to characterize association of linear growth faltering with enteropathogen burden and subsequent changes in EED biomarkers. In a longitudinal birth cohort (n = 272), monthly anthropometric measurements (Length for Age Z score- LAZ) of asymptomatic children were obtained up to 18 months. Biological samples were collected at 6 and 9 months for the assessment of biomarkers. A customized TaqMan array card was used to target 40 enteropathogens in fecal samples. Linear regression was applied to study the effect of specific enteropathogen infection on change in linear growth (ΔLAZ). Presence of any pathogen in fecal sample correlated with serum flagellin IgA (6 mo, r = 0.19, p = 0.002), fecal Reg 1b (6 mo, r = 0.16, p = 0.01; 9mo, r = 0.16, p = 0.008) and serum Reg 1b (6 mo, r = 0.26, p<0.0001; 9 mo, r = 0.15, p = 0.008). At 6 months, presence of Campylobacter [ß (SE) 7751.2 (2608.5), p = 0.003] and ETEC LT [ß (SE) 7089.2 (3015.04), p = 0.019] was associated with increase in MPO. Giardia was associated with increase in Reg1b [ß (SE) 72.189 (26.394), p = 0.006] and anti-flic IgA[ß (SE) 0.054 (0.021), p = 0.0091]. Multiple enteropathogen infections in early life negatively correlated with ΔLAZ, and simultaneous changes in gut inflammatory and permeability markers. A combination vaccine targeting enteropathogens in early life could help in the prevention of future stunting.

High SMAD7 and p-SMAD2,3 expression is associated with environmental enteropathy in children.

Enteropathies such as Crohn's disease are associated with enteric inflammation characterized by impaired TGF-ß signaling, decreased expression of phosphorylated (p)-SMAD2,3 and increased expression of SMAD7 (an inhibitor of SMAD3 phosphorylation). Environmental enteropathy (EE) is an acquired inflammatory disease of the small intestine (SI), which is associated with linear growth disruption, cognitive deficits, and reduced oral vaccine responsiveness in children <5 y in resource-poor countries. We aimed to characterize EE inflammatory pathways by determining SMAD7 and p-SMAD2,3 levels (using Western blotting) in EE duodenal biopsies (N = 19 children, 7 from Pakistan, 12 from Zambia) and comparing these with healthy controls (Ctl) and celiac disease (CD) patients from Italy. Densitometric analysis of immunoblots showed that EE SI biopsies expressed higher levels of both SMAD7 (mean±SD in arbitrary units [a.u.], Ctl = 0.47±0.20 a.u., EE = 1.13±0.25 a.u., p-value = 0.03) and p-SMAD2,3 (mean±SD, Ctl = 0.38±0.14 a.u., EE = 0.60±0.10 a.u., p-value = 0.03). Immunohistochemistry showed that, in EE, SMAD7 is expressed in both the epithelium and in mononuclear cells of the lamina propria (LP). In contrast, p-SMAD3 in EE is expressed much more prominently in epithelial cells than in the LP. The high SMAD7 immunoreactivity and lack of p-SMAD3 expression in the LP suggests defective TGF-ß signaling in the LP in EE similar to a previously reported SMAD7-mediated inflammatory pathway in refractory CD and Crohn's disease. However, Western blot densitometry showed elevated p-SMAD2,3 levels in EE, possibly suggesting a different inflammatory pathway than Crohn's disease but more likely reflecting cumulative protein expression from across all compartments of the mucosa as opposed to the LP alone. Further studies are needed to substantiate these preliminary results and to illustrate the relationship between SMAD proteins, TGF-ß signaling, and inflammatory cytokine production, all of which may be potential therapeutic targets.

Prostate-Specific Antigen 5 Years following Stereotactic Body Radiation Therapy for Low- and Intermediate-Risk Prostate Cancer: An Ablative Procedure?

BACKGROUND: Our previous work on early PSA kinetics following prostate stereotactic body radiation therapy (SBRT) demonstrated that an initial rapid and then slow PSA decline may result in very low PSA nadirs. This retrospective study sought to evaluate the PSA nadir 5 years following SBRT for low- and intermediate-risk prostate cancer (PCa). METHODS: 65 low- and 80 intermediate-risk PCa patients were treated definitively with SBRT to 35-37.5 Gy in 5 fractions at Georgetown University Hospital between January 2008 and October 2011. Patients who received androgen deprivation therapy were excluded from this study. Biochemical relapse was defined as a PSA rise >2 ng/ml above the nadir and analyzed using the Kaplan-Meier method. The PSA nadir was defined as the lowest PSA value prior to biochemical relapse or as the lowest value recorded during follow-up. Prostate ablation was defined as a PSA nadir <0.2 ng/ml. Univariate logistic regression analysis was used to evaluate relevant variables on the likelihood of achieving a PSA nadir <0.2 ng/ml. RESULTS: The median age at the start of SBRT was 72 years. These patients had a median prostate volume of 36 cc with a median 25% of total cores involved. At a median follow-up of 5.6 years, 86 and 37% of patients achieved a PSA nadir ≤0.5 and <0.2 ng/ml, respectively. The median time to PSA nadir was 36 months. Two low and seven intermediate risk patients experienced a biochemical relapse. Regardless of the PSA outcome, the median PSA nadir for all patients was 0.2 ng/ml. The 5-year biochemical relapse free survival (bRFS) rate for low- and intermediate-risk patients was 98.5 and 95%, respectively. Initial PSA (p = 0.024) and a lower testosterone at the time of the PSA nadir (p = 0.049) were found to be significant predictors of achieving a PSA nadir <0.2 ng/ml. CONCLUSION: SBRT for low- and intermediate-risk PCa is a convenient treatment option with low PSA nadirs and a high rate of early bRFS. Fewer than 40% of patients, however, achieved an ablative PSA nadir. Thus, the role of further dose escalation is an area of active investigation.

Acute Urinary Morbidity Following Stereotactic Body Radiation Therapy for Prostate Cancer with Prophylactic Alpha-Adrenergic Antagonist and Urethral Dose Reduction.

BACKGROUND: Stereotactic body radiation therapy (SBRT) delivers high doses of radiation to the prostate while minimizing radiation to the adjacent critical organs. Large fraction sizes may increase urinary morbidity due to unavoidable treatment of the prostatic urethra. This study reports rates of acute urinary morbidity following SBRT for localized prostate cancer with prophylactic alpha-adrenergic antagonist utilization and urethral dose reduction (UDR). METHODS: From April 2013 to September 2014, 102 patients with clinically localized prostate cancer were treated with robotic SBRT to a total dose of 35-36.25 Gy in five fractions. UDR was employed to limit the maximum point dose of the prostatic urethra to 40 Gy. Prophylactic alpha-adrenergic antagonists were initiated 5 days prior to SBRT and continued until resolution of urinary symptoms. Quality of life (QoL) was assessed before and after treatment using the American Urological Association Symptom Score (AUA) and the Expanded Prostate Cancer Index Composite-26 (EPIC-26). Clinical significance was assessed using a minimally important difference (MID) of one half SD change from baseline. RESULTS: One hundred two patients underwent definitive prostate SBRT with UDR and were followed for 3 months. No patient experienced acute urinary retention requiring catheterization. A mean baseline AUA symptom score of 9.06 significantly increased to 11.83 1-week post-SBRT (p = 0.0024) and 11.84 1-month post-SBRT (p = 0.0023) but returned to baseline by 3 months. A mean baseline EPIC-26 irritative/obstructive score of 87.7 decreased to 74.1 1-week post-SBRT (p < 0.0001) and 77.8 1-month post-SBRT (p < 0.0001) but returned to baseline at 3 months. EPIC-26 irritative/obstructive score changes were clinically significant, exceeding the MID of 6.0. At baseline, 8.9% of men described their urinary function as a moderate to big problem, and that proportion increased to 37.6% 1 week following completion of SBRT before returning to baseline by 3 months. CONCLUSION: Stereotactic body radiation therapy for localized prostate cancer with utilization of prophylactic alpha-adrenergic antagonist and UDR was well tolerated as determined by acute urinary function and bother, and symptoms were comparable to those observed following conventionally fractionated external beam radiation therapy (EBRT). Longer follow-up is required to assess long-term toxicity and efficacy following SBRT with UDR.