Giger, FDA 2009 Accepting CAD for Clinical Practice Maryellen L. Giger, Ph.D., FAAPM Professor & Vice-Chair for Basic Science Research Department of Radiology.

Slides:



Advertisements
Similar presentations
FDA QS reg/CLIA Comparison: Overview
Advertisements

1 Testing in the Open Market Testing in the Open Market AAAS Colloquium on Personalized Medicine: Planning for the Future June 2, 2009 Courtney C. Harper,
3D Imaging Research Massachusetts General Hospital Harvard Medical School 3D Computer-Aided Diagnosis (CAD) in Radiology Hiro Yoshida, PhD.
Chapter 4 Pattern Recognition Concepts: Introduction & ROC Analysis.
ACR and SBI Statement Margarita Zuley, MD Associate Professor, Radiology Medical Director, Breast Imaging University of Pittsburgh.
FDA-QA-DAS/2010 FDA’s Public Meeting: Device Improvements to Reduce the Number of Under-doses, Over-doses, and Misaligned Exposures from Therapeutic Radiation.
COURSE: JUST 3900 INTRODUCTORY STATISTICS FOR CRIMINAL JUSTICE Instructor: Dr. John J. Kerbs, Associate Professor Joint Ph.D. in Social Work and Sociology.
Computer Aided Diagnosis: CAD overview
Options for Regulation and the Impact of Regulation on the Marketplace 29 November 2005 Alan Kent
A 3D Approach for Computer-Aided Liver Lesion Detection Reed Tompkins DePaul Medix Program 2008 Mentor: Kenji Suzuki, Ph.D. Special Thanks to Edmund Ng.
Theresa Tsosie-Robledo MS RN-BC February 15, 2012
Data and Data Collection Quantitative – Numbers, tests, counting, measuring Fundamentally--2 types of data Qualitative – Words, images, observations, conversations,
Radiological Devices Advisory Committee Meeting November 18, 2009 John A. DeLucia iCAD, Inc.
1 History and Lessons from FDA Regulation of Digital Radiology Kyle J. Myers, Ph.D. Division of Imaging and Applied Mathematics OSEL/CDRH/FDA October 22,
AMERICAN COLLEGE OF RADIOLOGY March 23, OUR MISSION To foster the ongoing development of widely acceptable consistent imaging.
The Learning Agreement, Intellectual Property Rights and Project Approval Professor Dianne Ford Director of PhD Studies, Faculty of Medical Sciences.
Is there life after HEP - The Myths and the Reality of Working in Industry Kathy O’Shaughnessy, PhD R2 Technology, Inc.
BUSINESS SENSITIVE 1 Copyright © 2009 by James E. Dvorsky, Battelle Medical Device Solutions Compliance Engineering Jim Dvorsky Battelle Medical Device.
Financial Disclosure Robert Nishikawa: –paid consultant for Carestream and Siemens –shareholder in and receives royalties & research funding from Hologic,
ACRIN Breast Committee Fall Meeting : Comparison of Full-Field Digital Mammography with Digital Breast Tomosynthesis Image Acquisition in Relation.
Applicability of the AGREE II Instrument in Evaluating the Development Process and Quality of Current National Academy of Clinical Biochemistry Guidelines.
EDRN Approaches to Biomarker Validation DMCC Statisticians Fred Hutchinson Cancer Research Center Margaret Pepe Ziding Feng, Mark Thornquist, Yingye Zheng,
Biomedical Research Objective 2 Biomedical Research Methods.
Managing Your Grant Award August 23, 2012 Janet Stoeckert Director, Research Administration Sr. Administrator, Basic Sciences Keck School of Medicine 1.
New Draft Guidance for Multiplex Tests Elizabeth Mansfield and Michele Schoonmaker Office of In Vitro Diagnostic Device Evaluation and Safety (OIVD) CDRH/FDA.
Landmark Trials: Recommendations for Interpretation and Presentation Julianna Burzynski, PharmD, BCOP, BCPS Heme/Onc Clinical Pharmacy Specialist 11/29/07.
Component 6 - Health Management Information Systems Unit 1-2 What is Health Informatics?
EVALUATION OF HRD PROGRAMS Jayendra Rimal. The Purpose of HRD Evaluation HRD Evaluation – the systematic collection of descriptive and judgmental information.
Measuring and Improving Radiologists’ Interpretative Performance on Screening Mammography Karla Kerlikowske, MD Diana Buist, PhD Patricia Carney, PhD Berta.
Evaluating Impacts of MSP Grants Hilary Rhodes, PhD Ellen Bobronnikov February 22, 2010 Common Issues and Recommendations.
Regulatory Affairs and Adaptive Designs Greg Enas, PhD, RAC Director, Endocrinology/Metabolism US Regulatory Affairs Eli Lilly and Company.
COMPARABILITY PROTOCOLUPDATE ADVISORY COMMITTEE FOR PHARMACEUTICAL SCIENCE Manufacturing Subcommittee July 20-21, 2004 Stephen Moore, Ph.D. Chemistry Team.
Investigational Devices and Humanitarian Use Devices June 2007.
Screening Mammography Benchmarks – Modified Angoff: Screening Performance and Guidelines for Practice  Robert D Rosenberg and Patricia Carney for Breast.
Diagnostic reference levels in Medical Imaging. Concept and practice
Evaluation Requirements for MSP and Characteristics of Designs to Estimate Impacts with Confidence Ellen Bobronnikov February 16, 2011.
1 MITA Observations On Draft CADe Guidances Released by FDA October 21, 2009.
Jaro Vostal, MD, PhD Division of Hematology, OBRR, CBER, FDA
United States Department of Agriculture Food Safety and Inspection Service Overview of Trim Sampling Compliance Guidelines and Discussion Daniel Engeljohn,
Radiological Devices Advisory Panel Meeting Radiological Devices Advisory Panel Meeting Computer-Assisted Detection Devices Panel Questions Radiological.
Author : Elliot B. Sloane, Ph.D. American College of Clinical Engineering, President Villanova University Department of Decision.
T EST T OOLS U NIT VI This unit contains the overview of the test tools. Also prerequisites for applying these tools, tools selection and implementation.
12/10/02Sacks - Clinical Assessment1 Clinical Assessment – Part II William Sacks, PhD, MD Clinical Assessment – Part II William Sacks, PhD, MD COMPUTERIZED.
ACRIN Breast Committee Fall Meeting CADe study Etta Pisano, MD Martin Yaffe PhD Elodia Cole, MS Zheng Zhang, PhD ACRIN Breast Committee.
12/4/031 Drug Safety and Risk Management Advisory Committee Advancing the Science of Proprietary Drug Name Review Paul J. Seligman, MD.
Radiology Advisory Panel Meeting Radiology Advisory Panel Meeting Computer-Assisted Detection (CADe) Devices Joyce M. Whang Deputy Division Director Radiological.
Thursday Case of the Day History: The hypothetical results from a clinical trial of computer-aided detection (CADe) used in mammographic screening of 5,000,000.
United States Department of Agriculture Food Safety and Inspection Service Draft Labeling Policy Guidance for N- 60 Testing Claims for Boneless Beef Manufacturing.
Controlled Assessment Unit - CAU Investigative Skills Assessment - ISA 7 ISA section 2 exam – Additional / Separate Science Section 2 Exam: Up to 50 minutes.
Evaluation Requirements for MSP and Characteristics of Designs to Estimate Impacts with Confidence Ellen Bobronnikov March 23, 2011.
Contract Research organizations
U.S. FDA Center for Devices and Radiological Health Update
Adherence to the Labeling
On Draft CADe Guidances Released by FDA October 21, 2009
Premarket Notification 510(k) process
FDA’s IDE Decisions and Communications
NIR in Medicine Training and Legislation frame
Statistical Approaches to Support Device Innovation- FDA View
First-in-Man, First In The USA: What’s The Difference?
Deputy Director, Division of Biostatistics No Conflict of Interest
Quality Risk Management
Radiation Dose Management (RDM) Market to Cross $1.7 billion by 2023.
Crucial Statistical Caveats for Percutaneous Valve Trials
This teaching material has been made freely available by the KEMRI-Wellcome Trust (Kilifi, Kenya). You can freely download,
The Learning Agreement, Intellectual Property Rights and Project Approval Professor Dianne Ford Director of PhD Studies, Faculty of Medical Sciences.
The efficacy of using CAD for detection of
Linda M. Chatwin, Esq. RAC Business Manager, UL LLC
Thursday Case of the Day
Objective 2 Biomedical Research Methods
Regulatory Perspective of the Use of EHRs in RCTs
Presentation transcript:

Giger, FDA 2009 Accepting CAD for Clinical Practice Maryellen L. Giger, Ph.D., FAAPM Professor & Vice-Chair for Basic Science Research Department of Radiology & Committee on Medical Physics Carl J. Vyborny Translational Laboratory for Breast Imaging Research University of Chicago

Giger, FDA 2009 I am representing myself. My research is supported with grants from NIH, the U.S. Army DOD, the DOE, and the University of Chicago. Conflict of Interest: I am a stockholder in R2/Hologic, and I receiving royalties from Hologic, GE Medical Systems, MEDIAN Technologies, Riverain Medical, Mitsubishi, and Toshiba. Conflict of Interest: I am the current President of the American Association of Physicists in Medicine, however, I am NOT representing the AAPM here today.

Giger, FDA 2009 Concern As a CAD researcher, I am concerned about the timeliness and consistency of the translation of CAD developments to clinical use.

Giger, FDA 2009 Future of Computerized Image Analysis in Medical Decision Making Computers are increasingly being incorporated into our lives It is inevitable that they will be incorporated into medical imaging to aid in patient management What is important: –How to further the progress of CAD research –How to evaluate new CAD devices –How to expedite the process so that they are incorporated into clinical practice in a timely manner –Realization by users that radiologist training with the computer output is crucial when introducing CAD into the clinical arena

Giger, FDA 2009 How to evaluate a new CADe system??? Least burdensome approach to demonstrate substantial equivalence Standardization of testing including scoring method and ground truth Maintenance of the integrity of “test set”, and avoid reuse of data for testing Case mix: allow for enrichment of a test set with different lesion types, cancer prevalence, etc. to match the clinical context Reproducibility Reader mix: Realization that in ultimate clinical practice, the user will change with different CADe systems - so potentially could take user out of the evaluation??

Giger, FDA 2009 Suggest a Two-Stage Method of Evaluation BACKGROUND: “Independent Double Reading” Double reading improves detection sensitivity with some increase in recall rate –Occurs if intended goal of double reading is to improve sensitivity Double reading is accepted (although no FDA requirement) The increase in sensitivity and recall rate are similar for double read and single read with CAD

Giger, FDA 2009 Suggest a Two-Stage Method of Evaluation STAGE ONE: Determine a performance level standard for CAD based on published data and a cooperative study –With academia, government agencies, and commercial CAD manufacturers, demonstrate equivalency between “single read with CAD” and “double read”, [or demonstrate improvement with CADe as compared to no CADe]. –Using radiologists trained in CAD usage and CADe system(s) –Perform the cooperative study as an “ACRIN-like” study Output from this ONE LARGE READER STUDY will yield the needed “minimum bar” level of a computer alone performance in terms of detection sensitivity and false- positives marks per image.

Giger, FDA 2009 Suggest a Two-Stage Method of Evaluation After we know that CADe is useful via the “STAGE ONE” Study STAGE TWO (a): For evaluation of new or improved prior CADe systems Since the necessary performance level of CADe will be known from STAGE ONE, now it will only be necessary to demonstrate that the system’s STAND ALONE performance meets or exceeds the earlier- specified stand alone performance standard of sensitivity and FPs/image. Similar to the testing of image acquisition devices in terms of physical image quality metrics (e..g, MTF, SNR, etc.) Test result could include, for various “populations”: sensitivity and FPs/image, and perhaps FROC - Note location is included

Giger, FDA 2009 Suggest a Two-Stage Method of Evaluation STAGE TWO (b): An independent “technology assessment institute” (TAI) could be tasked with the assessment of the stand alone performance of all new or improved CAD devices –Institute would have a protected and sufficiently large database with appropriate distributions of cancer/abnormality sub-types and prevalence to allow for random sampling of cases for a CAD system assessment Can also vary the “enrichment” of different case types –Integrity of the “tests set” would be maintained since each test set would be randomly drawn from a LARGE set (within the TAI) according to a specific case distribution and cancer prevalence

Giger, FDA 2009 Suggest a Two-Stage Method of Evaluation STAGE TWO (c): An independent “technology assessment institute” (TAI) could be tasked with the assessment of the stand alone performance of all new or improved CAD devices –Only the test result (sensitivity and FP/image) would be given back to the manufacturer, so the manufacturer would not be able to “train to the test set” –Evaluation method eliminates the variation of the radiologist mindset, skill, level of training, etc. from the evaluation –Retesting of the CADe system by the TAI would yield measures of reproducibility (since a different testing set would be randomly drawn from the main TAI database) –Blackbox of the manufacturer is maintained –Output from such a TAI evaluation would be “acceptable” for FDA submissions

Giger, FDA 2009 Technology Assessment Institute (TAI) Independent Supported by government grants/contracts and industry (fee for testing service) Could have oversite from a scientific association such as AAPM (American Association of Physicists in Medicine) Expect would be used for FDA 510(k) for all types of CADe including mammography [SF & FFDM] in order to test the “label” and include the test score in the “label”

Giger, FDA 2009 Summary of Proposed Evaluation Method - 1 How to evaluate a new CADe system??? TAI approach allows for: Least burdensome approach to demonstrate substantial equivalence Standardization of testing including scoring method and ground truth Maintenance of the integrity of “test set”, and avoid reuse of data for testing -- Eliminates teaching to the test concern Case mix: Allow for enrichment of a test set with different lesion types, cancer prevalence, etc. to match the clinical context –Includes random samples of population Reproducibility metrics

Giger, FDA 2009 Summary of Proposed Evaluation Method - 2 How to evaluate a new CADe system??? TAI approach allows for: Reader Mix: Realization that in ultimate clinical practice, the user will change with different CADe systems - so potentially could take user out of the evaluation –TAI approach eliminates reader variability, bias, training level, performance level, problem with small number of readers that may not generalize, etc. –Clinical Utility would have been shown in Stage ONE

Giger, FDA 2009 Summary of Proposed Evaluation Method - 3 How to evaluate a new CADe system??? - TAI However, need to consider role of user/computer interface of any particular CADe system –Display device considerations –Necessary for only large differences in prompting methods Need to consider situation in which different types of FPs occur –Could be realized during post market evaluations –Note that the TAI test score would be included in “label” –[Recall that for CADe in mammography the output helps decide if additional imaging is needed, it does not inform on any other patient management decisions.]

Giger, FDA 2009 Consider the role of a TAI in pre- and post- market evaluations for CADe. Will also be useful for CADx. Thank you