Abstract:Background and objective: Intravascular ultrasound (IVUS) is the gold standard modality for in vivo visualization of coronary artery and atherosclerotic plaque. Fine segmentation of lumen and external elastic membrane (EEM) is the prerequisite for clinically quantitative assessment of coronary diseases. Manual qualification is timeconsuming and labor-intensive, and limited by subjectivity of technicians. The purpose of this paper is to design an automatic pipeline for IVUS diagnosis and evaluate the potential of integrating the approach into clinical practice. Methods: The present pipeline contains modules of vessel wall segmentation, quantification and 3D visualization. A total of 35,930 cross-sectional images from 292 patients were acquired and categorized for training and evaluation. We compared the segmentation model with state-of-art methods and explored the factors impacting the segmentation performance on multicenter dataset. The correlation and consistency of clinical quantitative measurements between the pipeline and the expert observer were evaluated based on Bland-Altman analysis. The 3D remodeling of consecutive IVUS sequences with positions of stented segments was presented. Results: High agreement between output results and the ground truth was found in the segmentation of vessel, with an overall Jaccard index (JI) of 0.916 and 0.945 for lumen and EEM respectively. Compared with state-ofart works on the public dataset, our method achieved the highest JI of both lumen and EEM in 40 MHz IVUS images. The correlation of quantitative measurements between the pipeline and the expert observer was evaluated, which showed high agreement with r (95 % CI) > 0.9 for all clinical parameters. The performance of external validation on multicenter database demonstrated the applicability of proposed pipeline. Conclusion: The proposed method has equivalent performance compared with human experts on vessel segmentation and quantitative measurements, which could be further developed as the workflow of clinical practice.

An Effective Pipeline for Whole-Slide Image Glomerulus Segmentation

Semantic Segmentation Based Quality Control of Histopathology Whole Slide Images

An Accelerated Pipeline for Multi-label Renal Pathology Image Segmentation at the Whole Slide Image Level

Hierarchical Attention Guided Framework for Multi-resolution Collaborative Whole Slide Image Segmentation

Multi-scale Multi-site Renal Microvascular Structures Segmentation for Whole Slide Imaging in Renal Pathology

Glo-In-One: Holistic Glomerular Detection, Segmentation, and Lesion Characterization with Large-scale Web Image Mining

Fully Automatic Deep Learning Pipeline for Whole Slide Image Quality Assessment

CovHuSeg: An Enhanced Approach for Kidney Pathology Segmentation

An Automatic Pipeline for Segmentation and Quantification of Intravascular Ultrasound Images

User friendly, cloud based, whole slide image segmentation

Automatic No-Reference kidney tissue whole slide image quality assessment based on composite fusion models

MUSTANG: Multi-Stain Self-Attention Graph Multiple Instance Learning Pipeline for Histopathology Whole Slide Images

An image inpainting-based data augmentation method for improved sclerosed glomerular identification performance with the segmentation model EfficientNetB3-Unet

Glo-In-One-v2: Holistic Identification of Glomerular Cells, Tissues, and Lesions in Human and Mouse Histopathology

A Deep Learning-Based Approach for Glomeruli Instance Segmentation from Multistained Renal Biopsy Pathologic Images

Evaluation Kidney Layer Segmentation on Whole Slide Imaging using Convolutional Neural Networks and Transformers

Tissue Cross-Section and Pen Marking Segmentation in Whole Slide Images

MSMTSeg: Multi-Stained Multi-Tissue Segmentation of Kidney Histology Images via Generative Self-Supervised Meta-Learning Framework

Enhanced Pooling for Weakly Supervised Gigapixel WSI Training Improves Classification and Lesion Localization

A user-friendly tool for cloud-based whole slide image segmentation, with examples from renal histopathology

Unleash the Power of State Space Model for Whole Slide Image with Local Aware Scanning and Importance Resampling