Cancer Prevention Through Early Detection : First International Workshop, CaPTion 2022, Held in Conjunction with MICCAI 2022, Singapore, September 22, 2022, Proceedings / edited by Sharib Ali, Fons van der Sommen, Bartłomiej Władysław Papież, Maureen van Eijnatten, Yueming Jin, Iris Kolenbrander.

Format:

Book

Contributor:

Alī, Śāriba, editor.

Series:

Lecture Notes in Computer Science, 1611-3349 ; 13581

Language:

English

Subjects (All):

Image processing--Digital techniques.

Image processing.

Computer vision.

Machine learning.

Computers.

Application software.

Computer Imaging, Vision, Pattern Recognition and Graphics.

Machine Learning.

Computing Milieux.

Computer and Information Systems Applications.

Local Subjects:

Computer Imaging, Vision, Pattern Recognition and Graphics.

Machine Learning.

Computing Milieux.

Computer and Information Systems Applications.

Physical Description:

1 online resource (175 pages)

Edition:

1st ed. 2022.

Place of Publication:

Cham : Springer Nature Switzerland : Imprint: Springer, 2022.

Summary:

This book constitutes the refereed proceedings of the first International Workshop on Cancer Prevention through Early Detection, CaPTion, held in conjunction with the 25th International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2022, in Singapore, Singapore, in September 2022. The 16 papers presented at CaPTion 2022 were carefully reviewed and selected from 21 submissions. The workshop invites researchers to submit their work in the field of medical imaging around the central theme of early cancer detection, and it strives to address the challenges that are required to be overcomed to translate computational methods to clinical practice through well designed, generalizable (robust), interpretable and clinically transferable methods.

Contents:

Intro

Preface

Organization

Contents

Classification

3D-Morphomics, Morphological Features on CT Scans for Lung Nodule Malignancy Diagnosis

1 Introduction

2 Methods

2.1 Data Sets

2.2 Data Analysis Models

3 Results

3.1 3D-Morphomics

3.2 Lung Nodule Diagnosis Performances of 3D-Morphomics

4 Conclusions

References

.26em plus .1em minus .1emSelf-supervised Approach for a Fully Assistive Esophageal Surveillance: Quality, Anatomy and Neoplasia Guidance

2 Related Work

3 Method

3.1 Self-supervision Solving Jigsaw Puzzle

3.2 Fine-Tuning with Angular Margin Loss

4 Experiments and Results

4.1 Implementation Details

4.2 Data Collection and Evaluation Metrics

4.3 Comparison with SOTA Methods

4.4 Qualitative Analysis

5 Conclusion

Multi-scale Deformable Transformer for the Classification of Gastric Glands: The IMGL Dataset

2 Related Works

3 Materials and Methods

3.1 IMGL Dataset Description

3.2 The Proposed IMGL-VTNet Architecture

3.3 Multi-scale Deformable Transformer Encoder

4 Experimental Results

4.1 A Comparison of State-of-the-Art Methods: IMGL Dataset

4.2 Feature Map Scales Analysis

4.3 Application of the Proposed Model to Pedestrian Detection

Parallel Classification of Cells in Thinprep Cytology Test Image for Cervical Cancer Screening

2 Method

2.1 Overview

2.2 Dual Classifiers in Parallel

2.3 Intra-class Compactness

2.4 Implementation Details

3 Experimental Results

3.1 Datasets

3.2 Classification Performance

3.3 Evolving of the Latent Space

4 Discussion and Conclusion

Detection and Diagnosis

Lightweight Transformer Backbone for Medical Object Detection

2 Methodology.

2.1 Overview of Proposed Method

2.2 Feature Map Rearrangement and Reconstruction

2.3 Lightweight Transformer on Feature Patches

3 Experiments and Results

3.1 Dataset and Evaluation Metrics

3.2 Implementation Details

3.3 Experimental Results

4 Conclusion

Contrastive and Attention-Based Multiple Instance Learning for the Prediction of Sentinel Lymph Node Status from Histopathologies of Primary Melanoma Tumours

2 Materials and Methods

2.1 Dataset

2.2 Multiple Instance Learning

2.3 Proposed Model

2.4 Self-supervised Contrastive Learning:

3 Experimental Set-Up and Results

3.1 Feature Extraction

3.2 Experiments

4 Discussion

5 Conclusions

Knowledge Distillation with a Class-Aware Loss for Endoscopic Disease Detection

3 Materials and Method

3.2 Proposed Knowledge-Distillation Framework

4.1 Experimental Setup and Evaluation Metrics

4.2 Results

IF3: An Interpretable Feature Fusion Framework for Lesion Risk Assessment Based on Auto-constructed Fuzzy Cognitive Maps

2 Methodology

2.1 Fuzzy Cognitive Maps

2.2 Proposed Framework

3.1 Dataset Description and Parameter Settings

3.2 Interpretable Example of Risk Assessment Using IF3

3.3 Performance Evaluation of IF3

4 Discussion and Conclusions

Lesion Characterization

A CAD System for Real-Time Characterization of Neoplasia in Barrett's Esophagus NBI Videos

2.1 Data

2.2 Network Architecture, Training and Evaluation

2.3 Video Analysis Methods

References.

Efficient Out-of-Distribution Detection of Melanoma with Wavelet-Based Normalizing Flows

2 Background

2.1 Normalizing Flows

2.2 Wavelet Flow

3 Methods

4 Results and Discussion

Robust Colorectal Polyp Characterization Using a Hybrid Bayesian Neural Network

2.2 Bayesian Neural Networks

2.3 Model Architecture

2.4 Evaluation Metrics

3.1 Experimental Setting

3.2 Calibration-performance Assessment

3.3 Model Performance Comparison

3.4 Generalization and Robustness to Over-Fitting Assessment

Active Data Enrichment by Learning What to Annotate in Digital Pathology

2.1 Annotation Protocol

2.2 Dataset Enrichment

3.1 Unsupervised Data Enrichment

3.2 Supervised Active Data Enrichment

Segmentation, Registration, and Image-Guided Intervention

Comparing Training Strategies Using Multi-Assessor Segmentation Labels for Barrett's Neoplasia Detection

2.1 Data Set

2.2 Segmentation Ground-truth Assembly

2.3 Network Architecture

2.4 Training Details

3.1 Metrics

3.2 Results

Improved Pancreatic Tumor Detection by Utilizing Clinically-Relevant Secondary Features

2 Related Work on PDAC Detection

3.1 Data Collection

3.2 Segmentation Model for Classification and Localization

3.3 Experiments

3.4 Data Preparation and Training Details

Strategising Template-Guided Needle Placement for MR-targeted Prostate Biopsy

2 Method.

2.1 Patient-specific Prostate MR-derived Biopsy Environment

2.2 The MDP Components

2.3 Policy Learning

3 Experiments

4 Results

5 Discussion and Conclusion

Semantic-Aware Registration with Weakly-Supervised Learning

2.1 Structural Constraints

2.2 Adaptive Registration

3.1 Registration Results

Author Index.

Notes:

Includes bibliographical references and index.

Other Format:

Print version: Ali, Sharib Cancer Prevention Through Early Detection

ISBN:

9783031179792

303117979X