Girish Hiremath

Recently, circle representation has been introduced for medical imaging, designed specifically to enhance the detection of instance objects that are spherically shaped (e.g., cells, glomeruli, and nuclei). Given its outstanding effectiveness in instance detection, it is compelling to consider the application of circle representation for segmenting instance medical objects. In this study, we introduce CircleSnake, a simple end-to-end segmentation approach that utilizes circle contour deformation for segmenting ball-shaped medical objects at the instance level. The innovation of CircleSnake lies in these three areas: (1) It substitutes the complex bounding box-to-octagon contour transformation with a more consistent and rotation-invariant bounding circle-to-circle contour adaptation. This adaptation specifically targets ball-shaped medical objects. (2) The circle representation employed in CircleSnake significantly reduces the degrees of freedom to two, compared to eight in the octagon representation. This reduction enhances both the robustness of the segmentation performance and the rotational consistency of the method. (3) CircleSnake is the first end-to-end deep instance segmentation pipeline to incorporate circle representation, encompassing consistent circle detection, circle contour proposal, and circular convolution in a unified framework. This integration is achieved through the novel application of circular convolution within the context of circle detection and instance segmentation. In practical applications, such as the detection of glomeruli, nuclei, and eosinophils in pathological images, CircleSnake has demonstrated superior performance and greater rotation invariance when compared to benchmarks. The code has been made publicly available at: https://github.com/hrlblab/CircleSnake.

Telehealth remains understudied in patients with eosinophilic esophagitis and eosinophilic gastrointestinal diseases (EGIDs), yet may serve as an important tool for increasing access to providers with EGID-specific expertise. The online patient-centered research network, EGID Partners, provided insight into EGID-related telehealth utilization. Respondents reported that telehealth visits offered the ability to spend adequate time with their healthcare provider and communicate just as effectively as an in-person visit, while also incurring lower travel-related costs and less missed work or school. Here, we provide lessons learned that telehealth can be an effective, acceptable, and feasible method of delivering care to EGID patients.

Summary: Given the chronic and progressive course of eosinophilic esophagitis (EoE), patients with pediatric-onset EoE will require uninterrupted gastroenterology (GI) care as they reach adulthood. Yet, the effectiveness of transferring and integrating EoE patients from pediatric GI (pGI) to adult GI (aGI) care has not been studied. To address this gap, we assessed loss to follow-up, duration from the last pGI to the first aGI encounters (clinic visit and EGD), and its impact on clinical course and medication non-adherence in EoE patients. We identified 58 EoE patients who initially received pGI care and were transferred to aGI between 2017 and 2023 within our institution’s shared electronic medical record environment. Demographic, clinical, endoscopic, and histologic data were analyzed using descriptive statistics, survival analysis, Cox regression models, and paired comparisons. Loss to follow-up was 16%. The median duration from the last pGI clinic visit to the first aGI clinic visit was 299 days, and that for the last pGI EGD to the first aGI EGD was 730 days. A significantly higher odds of heartburn (McNemar P-value = 0.01) and higher medication non-adherence rates (7% vs. 26%) were noted in 49 patients who established care with the aGI. The endoscopic and histologic severity remained unchanged. In EoE patients, transferring from pGI to aGI care is associated with loss of follow-up, deterioration of symptoms, and medication non-adherence. There is a critical need to develop optimized protocols to ensure a seamless transfer of care for EoE patients.

Summary: Little is known about the extent or importance of overlapping gastrointestinal (GI) tract involvement in eosinophilic gastrointestinal diseases (EGIDs), how presentations differ by areas of involvement, and whether overlap impacts treatments. We aimed to evaluate overlapping GI tract involvement in EGIDs and whether clinical differences existed. To do this, we assessed the EGID Partners cohort, an online patient-centered research network. Adults (≥18 years) and caregivers of children <18 years old with EoE or non-EoE EGIDs could join. Surveys were completed at enrollment, comparing patients with EoE alone, EGID without esophageal involvement (‘EGID-NE’), and EGID with esophageal involvement (‘EGID-WE’). Of 527 cases enrolled, 402 had EoE alone and 125 had non-EoE EGID, 57 (46%) with EGID-NE, and 68 (53%) with EGID-WE. There were 10, 18, and 9 with eosinophilic gastritis, gastroenteritis, and colitis alone, respectively; 88 had overlap. EGID-NE had a higher proportion of females (79%; P < 0.001), and family history of EoE/EGID was more common in EGID-WE (19% vs. 11% in EoE and 7% in EGID-NE; P = 0.007). Patient-Reported Outcomes Measurement Information System measures for anxiety were above general population averages and highest for EGID-WE. Treatments such as elemental formula (47% vs. 32% vs. 20%; P = 0.001), systemic steroids (33% vs. 56% vs. 14%; P < 0.001), and biologics were also more common in EGID-WE and EGID-NE. In conclusion, overlap in regions with eosinophilic infiltration is common for non-EoE EGIDs, with more than half of non-EoE EGIDs having esophageal involvement and a high proportion of multisegmental involvement. EGID-WE patients tended to have more disease burden.

Eosinophilic esophagitis (EoE) is an immunoinflammatory disease of the esophagus attributable to a complex interaction between genetic and environmental factors. While several genetic risk variants have been linked with EoE, we report a novel association between B‐cell lymphoma/leukemia 11B genetic mutation in a child with dysmorphic facies, developmental delays, atopic comorbidities, and difficult‐to‐treat EoE. After a prolonged course of EoE and multiple esophagogastroduodenoscopies with biopsies, this patient achieved clinical and histologic remission on a combination of swallowed topical steroids and high‐dose proton pump inhibitor (PPI) therapy. However, her EoE relapsed when we attempted to wean her off PPI, and it was finally controlled after adding PPI back to her regimen. This report underscores the importance of genetic testing in patients with unusual clinical features and difficult‐to‐treat EoE. Relevant to real‐world clinical practice, this case also raises the question of the treatment goals in children with EoE and underlying genetic mutation(s).

Background: A combination of proton‐pump inhibitors (PPI) and topical steroids (TS) is used to treat children with eosinophilic esophagitis (EoE). However, a subset of children do not respond to this combination therapy. We aimed to identify the esophageal transcriptional, cell composition, and microbial differences between the non‐responders (EoE‐PPI‐TSnr; n = 7) and responders (EoE‐PPI‐TSr; n = 7) to the combination therapy for EoE and controls ( n = 9) using metatranscriptomics. Methods: Differential gene expression analysis was used to identify transcriptional differences, validated using the EoE diagnostic panel (EDP). Deconvolution analysis was performed to identify differences in their cell type composition. Microbiome analysis was conducted from esophageal biopsies RNAseq data, and microbial abundance was correlated with esophageal gene expression. Results: In all, 3164 upregulated and 3154 downregulated genes distinguished EoE‐PPI‐TSnr from EoE‐PPI‐TSr. Eosinophilic inflammatory response, cytokine signaling, and collagen formation pathways were significantly upregulated in EoE‐PPI‐TSnr. There was a 56% overlap in dysregulated genes between EoE‐PPI‐TSnr and EDP, with a perfect agreement in the directionality of modulation. Eosinophils, dendritic cells (DCs), immature DCs, megakaryocytic‐erythroid progenitors, and T helper type 1 cells were significantly higher in EoE‐PPI‐TSnr. There was no significant difference in microbiome diversity. The relative abundance of Fusobacterium sp. and Acinetobacter sp. notably differed in EoE‐PPI‐TSnr and correlated with the key pathways. Conclusion: Our results provide critical insights into the molecular, cellular, and microbial factors associated with the lack of response to PPI and TS combination therapy in children with EoE. This study advances our understanding of the pathobiology of EoE while guiding personalized treatment strategies.

Eosinophilic esophagitis (EoE) is a chronic inflammatory condition characterized by an intense infiltration of eosinophils into the esophageal epithelium. When not adequately controlled, eosinophilic inflammation can lead to changes in components of the extracellular matrix (ECM) of the lamina propria. Particularly, alterations to the collagen fiber matrix can lead to lamina propria fibrosis (LPF), which plays an important role in the fibrostenotic complications of EoE. Current approaches to assess LPF in EoE are prone to inter-observer inconsistencies and provide limited insight into the structural remodeling of the ECM. An objective approach to quantify LPF can eliminate inter-observer inconsistencies and provide novel insights into the fibrotic transformation of the lamina propria in EoE. Second harmonic generation (SHG) microscopy is a powerful modality for objectively quantifying disease associated alterations in ECM collagen structure that is finding increasing use for clinical research. We used SHG with morphometric analysis (SHG-MA) to characterize lamina propria collagen fibers and ECM porosity in esophageal biopsies collected from children with active EoE (n = 11), inactive EoE (n = 11), and non-EoE (n = 11). The collagen fiber width quantified by SHG-MA correlated positively with peak eosinophil count (r = 0.65, p < 0.005) and histopathologist scoring of LPF (r = 0.52, p < 0.005) in the esophageal biopsies. Patients with active EoE had a significant enlargement of ECM pores compared to inactive EoE and non-EoE (p < 0.005), with the mean pore area correlating positively with EoE activity (r = 0.76, p < 0.005) and LPF severity (r = 0.65, p < 0.005). These results indicate that SHG-MA can be utilized to objectively characterize and provide novel insights into lamina propria ECM structural remodeling in children with EoE, which could aid in monitoring disease progression.