Cloud-Based Benchmarking of Medical Image Analysis.

• Author: Hanbury, Allan.
• TeilnehmendeR: Müller, Henning.; Langs, Georg.
• Place / Publishing House: Cham : : Springer International Publishing AG,, 2017.; ©2017.
• Year of Publication: 2017
• Edition: 1st ed.
• Language: English
• Physical Description: 1 online resource (256 pages)

Table of Contents

Intro -- Preface -- Acknowledgements -- Contents -- Contributors -- Acronyms -- Part I Evaluation-as-a-Service -- 1 VISCERAL: Evaluation-as-a-Service for Medical Imaging -- 1.1 Introduction -- 1.2 VISCERAL Benchmarks -- 1.2.1 Anatomy Benchmarks -- 1.2.2 Detection Benchmark -- 1.2.3 Retrieval Benchmark -- 1.3 Evaluation-as-a-Service in VISCERAL -- 1.4 Main Outcomes of VISCERAL -- 1.4.1 Gold Corpus -- 1.4.2 Silver Corpus -- 1.4.3 Evaluation Metric Calculation Software -- 1.5 Experience with EaaS in VISCERAL -- 1.6 Conclusion -- References -- 2 Using the Cloud as a Platform for Evaluation and Data Preparation -- 2.1 Introduction -- 2.2 VISCERAL Registration System -- 2.2.1 Registration -- 2.2.2 Participant Dashboard -- 2.2.3 Management of Participants -- 2.2.4 Open Source Software Release -- 2.3 Continuous Evaluation in the Cloud -- 2.3.1 Submission -- 2.3.2 Isolation of the VM -- 2.3.3 Initial Test -- 2.3.4 Executing Algorithms and Saving the Results -- 2.3.5 Evaluation of Results -- 2.4 Cloud-Based Evaluation Infrastructure -- 2.4.1 Setting up a Cloud Environment -- 2.4.2 Setting up a Benchmark in the Cloud -- 2.4.3 Cloud Set-Up for the VISCERAL Benchmarks -- 2.4.4 Cloud Infrastructure Setup and Management Experience Report -- 2.5 Conclusion -- References -- Part II VISCERAL Datasets -- 3 Ethical and Privacy Aspects of Using Medical Image Data -- 3.1 Introduction -- 3.2 Ethical and Privacy Aspects for Data Access -- 3.2.1 Review by the Medical Ethics Committee -- 3.2.2 Handling of Informed Consent Procedures -- 3.2.3 Anonymization -- 3.2.4 Data Distribution During and After the Benchmarks -- 3.3 Relevant Legislation -- 3.4 Procedures Implemented by Data Providers -- 3.4.1 Agencia D'Informació, Avaluació i Qualitat en Salut, Spain -- 3.4.2 Medizinische Universität Wien (Austria) -- 3.4.3 Universitätsklinikum Heidelberg (Germany).

3.5 Aspects, Recommendations and Conditions for Obtaining Approval from Ethical Committees -- 3.6 Conclusion -- References -- 4 Annotating Medical Image Data -- 4.1 Introduction -- 4.2 3D Annotation Software -- 4.2.1 Evaluation Criteria -- 4.2.2 Reviewed Annotation Tools -- 4.2.3 Tool Comparison -- 4.2.4 Selected Software and Technical Aspects -- 4.3 VISCERAL Ticketing Tool/Framework -- 4.3.1 Ticketing System Database -- 4.3.2 Annotation Ticket Life Cycle -- 4.3.3 Manual Annotation Instructions -- 4.4 Inter-annotator Agreement -- 4.5 Conclusion -- References -- 5 Datasets Created in VISCERAL -- 5.1 Introduction -- 5.2 Anatomy Gold Corpus -- 5.3 Anatomy Silver Corpus -- 5.4 Detection Gold Corpus -- 5.5 Retrieval Gold Corpus -- 5.6 Retrieval Silver Corpus -- 5.7 Summary -- References -- Part III VISCERAL Benchmarks -- 6 Evaluation Metrics for Medical Organ Segmentation and Lesion Detection -- 6.1 Introduction -- 6.2 Metrics for VISCERAL Benchmarks -- 6.2.1 Metrics for Segmentation -- 6.2.2 Metrics for Lesion Detection -- 6.3 Analysis of Fuzzy Segmentation Metrics -- 6.3.1 Metric Sensitivity Against Fuzzification -- 6.3.2 Ranking Systems Using Binary/Fuzzy Ground Truth -- 6.4 Analysis of Metrics Using Manual Rankings -- 6.4.1 Dataset -- 6.4.2 Manual Versus Metric Rankings at Segmentation Level -- 6.4.3 Manual Versus Metric Rankings at System Level -- 6.4.4 Discussion of the Manual Ranking Analysis -- 6.5 Conclusion -- References -- 7 VISCERAL Anatomy Benchmarks for Organ Segmentation and Landmark Localization: Tasks and Results -- 7.1 Introduction -- 7.2 Data and Data Format -- 7.2.1 Data -- 7.2.2 Gold Corpus: Training Set -- 7.2.3 Gold Corpus: Test Set -- 7.2.4 Data Format -- 7.3 Tasks -- 7.4 Results -- 7.4.1 Anatomy1 -- 7.4.2 Anatomy2: Intermediate Results at the ISBI Challenge -- 7.4.3 Anatomy2: Main Benchmark -- 7.4.4 Anatomy3 -- 7.4.5 Discussion.

7.5 Conclusion -- References -- 8 Retrieval of Medical Cases for Diagnostic Decisions: VISCERAL Retrieval Benchmark -- 8.1 Introduction -- 8.2 Dataset -- 8.3 Medical Case-Based Retrieval -- 8.4 Evaluation -- 8.4.1 Relevance Judgements -- 8.4.2 Metrics -- 8.5 Participants -- 8.6 Results -- 8.7 Conclusion -- References -- Part IV VISCERAL Anatomy Participant Reports -- 9 Automatic Atlas-Free Multiorgan Segmentation of Contrast-Enhanced CT Scans -- 9.1 Introduction -- 9.2 Method -- 9.2.1 Process 1: Scan-Specific Characterization -- 9.2.2 Process 2: Generic Four-Step Segmentation -- 9.2.3 Process 2: Implementation details -- 9.2.4 Post-processing at the End of Process 2 -- 9.3 The VISCERAL Benchmark -- 9.4 Results and Discussion -- 9.5 VISCERAL Benchmark Perspective -- 9.6 Conclusion -- References -- 10 Multiorgan Segmentation Using Coherent Propagating Level Set Method Guided by Hierarchical Shape Priors and Local Phase Information -- 10.1 Introduction -- 10.2 Statistical Shape-Prior-Guided Level Set Segmentation -- 10.3 Multiorgan Segmentation Using Hierarchical Shape Priors -- 10.3.1 Building Hierarchical Shape Priors -- 10.3.2 Multiorgan Segmentation Using Hierarchical Shape Priors -- 10.3.3 Region-Based External Speed Function -- 10.4 Improving Segmentation Accuracy Using Model-Guided Local Phase Analysis -- 10.4.1 Quadrature Filters and Model-Guided Local Phase Analysis -- 10.4.2 Integrating Region-Based and Edge-Based Energy in the Level Set Method -- 10.5 Speeding up Level Set Segmentation Using Coherent Propagation -- 10.6 Experiments and Results -- 10.7 Discussion and Conclusion -- References -- 11 Automatic Multiorgan Segmentation Using Hierarchically Registered Probabilistic Atlases -- 11.1 Introduction and Related Work -- 11.2 Methods -- 11.2.1 SURF Keypoint-Based Image Registration -- 11.2.2 Organ Atlas Construction.

11.2.3 Image Clustering -- 11.2.4 Multiorgan Image Segmentation -- 11.3 Evaluation Results and Discussion -- 11.4 Concluding Remarks and Future Work -- References -- 12 Multiatlas Segmentation Using Robust Feature-Based Registration -- 12.1 Introduction -- 12.1.1 Related Work -- 12.1.2 Our Approach -- 12.2 Methods -- 12.2.1 Pairwise Registration -- 12.2.2 Label Fusion with a Random Forest Classifier -- 12.2.3 Graph Cut Segmentation with a Potts Model -- 12.3 Experimental Evaluation -- 12.3.1 Challenge Results -- 12.3.2 Detailed Evaluation -- 12.4 Conclusions -- References -- Part V VISCERAL Retrieval Participant Reports -- 13 Combining Radiology Images and Clinical Metadata for Multimodal Medical Case-Based Retrieval -- 13.1 Introduction -- 13.2 Materials and Methods -- 13.2.1 Dataset -- 13.2.2 VISCERAL Retrieval Benchmark Evaluation Setup -- 13.2.3 Multimodal Medical Case Retrieval -- 13.3 Results -- 13.3.1 Lessons Learned -- 13.4 Conclusions -- References -- 14 Text- and Content-Based Medical Image Retrieval in the VISCERAL Retrieval Benchmark -- 14.1 Introduction -- 14.2 Methods -- 14.2.1 Term Weighting Retrieval -- 14.2.2 Semantics Retrieval -- 14.2.3 BoVW Retrieval -- 14.2.4 Retrieval Result Refinement -- 14.2.5 Fusion Retrieval -- 14.3 Results and Discussion -- 14.4 Conclusion -- References -- Index.