SlideShare a Scribd company logo
1 of 26
Download to read offline
#SIIM17
The Impact of Segmentation on the
Accuracy and Sensitivity of a
Melanoma Classifier
Based on Skin Lesion Images
Oge Marques, PhD
Professor
College of Engineering and Computer Science
Florida Atlantic University
#SIIM17
Our Team
Adrià Romero López Xavier Giró-i.Nieto
Image Processing Group
Signal Theory and
Communications Department
MIDDLE Research Group
Oge Marques Borko Furht Jack Burdick Janet Weinthal
NSF Award No. 1464537, I/UCRC Phase II under NSF 13-542
#SIIM17
Outline
• Motivation
• Context
• Scope and goals
• Challenges
• State of the art
• Our work
• Hypothesis
• Methodology
• Experimental results
• Ongoing and future work
• Concluding remarks
#SIIM17
Motivation
[Source: https://seer.cancer.gov/statfacts/html/melan.html]
Melanoma of the skin
#SIIM17
Context
• This is not a typical SIIM presentation
• No specialized imaging equipment
• No PACS
• No DICOM
• No metadata
• No workflows
• Instead...
• Regular photographs
• Unstructured (and purely visual) data
• Minimal ground truth
#SIIM17
Scope
• Skin Disease: An
Illustrated Taxonomy
• Our focus: skin
lesion analysis for
(early) melanoma
detection
[Source:	Esteva	et	al.,	Nature	(2017)]
#SIIM17
Scope and Goals
• Scope:
• Help physicians to detect melanoma (a 2-class
classifier)
• Goals:
• Design an intelligent medical imaging skin lesion
diagnosis system using deep learning techniques
• Achieve (or improve upon) state-of-the-art results for
skin lesion classification
#SIIM17
A Challenging Problem
8
#SIIM17
Success Rate - Sensitivity
Physicians, as low as:
• 43 % - naked eye
• 79 % - dermoscopy
(Vestergaard et al, 2008)
Deep learning based methods, as high as:
• 94.83 %
(Jafari et al, 2016; Premaladha and Ravichandran 2016)
#SIIM17
Classical Techniques
[Source: Center For Excellence In Dermatology - Kennewick, WA]
CASH (Henning et al., 2007)
Color
Architecture
Symmetry
Homogeneity
#SIIM17
Typical workflow (pre-DL)
Source: [Glaister2013]
11
#SIIM17
Challenges
• Segmentation
• Artifacts, such as: freckles, hair, shading, and noise.
• Irregular, fuzzy lesion border
• Low contrast between lesion and surrounding skin
• Multi-fragment lesions
• Classification
• Manual feature extraction is not effective
• False positives and false negatives have different impact
#SIIM17
State of the art
• Esteva et al. Nature 2017
#SIIM17
Our Approach
• Pretrained VGG16
(Simonyan & Zisserman, 2014)
• Transfer Learning
[Source:Blier 2016]
#SIIM17
Transfer Learning
Source: [http://www.slideshare.net/hunkim/transfer-defectlearningnew-completed]
#SIIM17
Transfer Learning
1. Train on
Imagenet
3. Medium dataset:
finetuning
more data = retrain more of
the network (or all of it)
2. Small dataset:
feature extractor
Freeze these
Train this
Freeze these
Train this
Slide credit: Bay Area Deep Learning School Presentation by A. Karpathy
Medical Imaging case
#SIIM17
Prediction Examples:
Getting It Right
True Positives True Negatives
#SIIM17
Prediction Examples:
Getting It Wrong
False Positives False Negatives
#SIIM17
Our Hypothesis
• Image segmentation improves the performance of skin lesion
classifiers using convolutional neural networks.
[Source:International Skin Imaging Collaboration Archive]
Not segmented
Perfectly segmented
Partially segmented
#SIIM17
Methods
• ISBI 2016 Challenge dataset
• Skin Lesion Analysis towards melanoma detection
• 1279 RGB images
• Labeled as either benign or malignant
Class
Benign Malignant Total Images
Training subset 727 173 900
Testing subset 304 75 379
#SIIM17
Methods
• Dataset balancing through downsampling.
• Dataset split: 70-30% training/testing
• Input images:
• Unsegmented images: straight from the dataset.
• Perfectly segmented images: bitwise AND operation of the unaltered
images and its corresponding binary mask provided by the ISIC
dataset.
• Partially segmented images: original binary masks morphologically
dilated with a disk-shaped structuring element (50 pixel radius).
• Additional preprocessing methods (resizing and normalization) were
also performed to match the input size expected by the VGG16
architecture.
#SIIM17
Results
#SIIM17
Further Investigation
• What if we vary the degree of border expansion?
Sensitivity Accuracy AUC
Perfect Segmentation 45.3% 58.7% 62.2%
+25 53.3% 61.3% 64.2%
+50 56.0% 60.7% 62.6%
+75 57.3% 59.3% 60.8%
+100 34.7% 55.3% 57.9%
Unsegmented 24.0% 51.3% 53.2%
#SIIM17
Ongoing and Future Work
•Additional / larger / more challenging datasets
•Other CNN architectures
•Better image preprocessing
•Partnerships and collaborations
•Mobile app
#SIIM17
Concluding remarks
• Challenges
• Difficulty in acquiring datasets and reproducing / benchmarking results
• The “black box” aspect of DL-based solutions
• Hard to tell positives from negatives
• Learning curve: TensorFlow, Keras, HPC, DL concepts and best practices, etc.
• Opportunities
• Many variations of the basic classification problem
• Mobile app market
• Tech-minded dermatology practices
#SIIM17

More Related Content

What's hot

IISc Internship Report
IISc Internship ReportIISc Internship Report
IISc Internship ReportHarshilJain26
 
Codetecon #KRK 3 - Object detection with Deep Learning
Codetecon #KRK 3 - Object detection with Deep LearningCodetecon #KRK 3 - Object detection with Deep Learning
Codetecon #KRK 3 - Object detection with Deep LearningMatthew Opala
 
TRECVID 2016 POSTER CERTH-ITI
TRECVID 2016 POSTER CERTH-ITITRECVID 2016 POSTER CERTH-ITI
TRECVID 2016 POSTER CERTH-ITIMOVING Project
 
unrban-building-damage-detection-by-PJLi.ppt
unrban-building-damage-detection-by-PJLi.pptunrban-building-damage-detection-by-PJLi.ppt
unrban-building-damage-detection-by-PJLi.pptgrssieee
 
Transfer Learning Model for Image Segmentation by Integrating U-NetPlusPlus a...
Transfer Learning Model for Image Segmentation by Integrating U-NetPlusPlus a...Transfer Learning Model for Image Segmentation by Integrating U-NetPlusPlus a...
Transfer Learning Model for Image Segmentation by Integrating U-NetPlusPlus a...YutaSuzuki27
 
Deep Learning based Segmentation Pipeline for Label-Free Phase-Contrast Micro...
Deep Learning based Segmentation Pipeline for Label-Free Phase-Contrast Micro...Deep Learning based Segmentation Pipeline for Label-Free Phase-Contrast Micro...
Deep Learning based Segmentation Pipeline for Label-Free Phase-Contrast Micro...Fellowship at Vodafone FutureLab
 
Automatic Building detection for satellite Images using IGV and DSM
Automatic Building detection for satellite Images using IGV and DSMAutomatic Building detection for satellite Images using IGV and DSM
Automatic Building detection for satellite Images using IGV and DSMAmit Raikar
 
SPIE 10059-36(Reheman Baikejiang)
SPIE 10059-36(Reheman Baikejiang)SPIE 10059-36(Reheman Baikejiang)
SPIE 10059-36(Reheman Baikejiang)Reheman Baikejiang
 
Александр Заричковый "Faster than real-time face detection"
Александр Заричковый "Faster than real-time face detection"Александр Заричковый "Faster than real-time face detection"
Александр Заричковый "Faster than real-time face detection"Fwdays
 
Ieee 2016 nss mic poster N30-21
Ieee 2016 nss mic poster N30-21Ieee 2016 nss mic poster N30-21
Ieee 2016 nss mic poster N30-21Dae Woon Kim
 
ieee nss mic 2016 poster N30-21
ieee nss mic 2016 poster N30-21ieee nss mic 2016 poster N30-21
ieee nss mic 2016 poster N30-21Dae Woon Kim
 
Knowledge-based Fusion for Image Tampering Localization
Knowledge-based Fusion for Image Tampering LocalizationKnowledge-based Fusion for Image Tampering Localization
Knowledge-based Fusion for Image Tampering LocalizationSymeon Papadopoulos
 
Avihu Efrat's Viola and Jones face detection slides
Avihu Efrat's Viola and Jones face detection slidesAvihu Efrat's Viola and Jones face detection slides
Avihu Efrat's Viola and Jones face detection slideswolf
 
KaoNet: Face Recognition and Generation App using Deep Learning
KaoNet: Face Recognition and Generation App using Deep LearningKaoNet: Face Recognition and Generation App using Deep Learning
KaoNet: Face Recognition and Generation App using Deep LearningVan Huy
 
IRJET - Real Time Object Detection using YOLOv3
IRJET - Real Time Object Detection using YOLOv3IRJET - Real Time Object Detection using YOLOv3
IRJET - Real Time Object Detection using YOLOv3IRJET Journal
 
De-convolution on Digital Images
De-convolution on Digital ImagesDe-convolution on Digital Images
De-convolution on Digital ImagesMd. Shohel Rana
 
THE EFFECT OF PHYSICAL BASED FEATURES FOR RECOGNITION OF RECAPTURED IMAGES
THE EFFECT OF PHYSICAL BASED FEATURES FOR RECOGNITION OF RECAPTURED IMAGESTHE EFFECT OF PHYSICAL BASED FEATURES FOR RECOGNITION OF RECAPTURED IMAGES
THE EFFECT OF PHYSICAL BASED FEATURES FOR RECOGNITION OF RECAPTURED IMAGESijcsit
 

What's hot (20)

IISc Internship Report
IISc Internship ReportIISc Internship Report
IISc Internship Report
 
Codetecon #KRK 3 - Object detection with Deep Learning
Codetecon #KRK 3 - Object detection with Deep LearningCodetecon #KRK 3 - Object detection with Deep Learning
Codetecon #KRK 3 - Object detection with Deep Learning
 
TRECVID 2016 POSTER CERTH-ITI
TRECVID 2016 POSTER CERTH-ITITRECVID 2016 POSTER CERTH-ITI
TRECVID 2016 POSTER CERTH-ITI
 
unrban-building-damage-detection-by-PJLi.ppt
unrban-building-damage-detection-by-PJLi.pptunrban-building-damage-detection-by-PJLi.ppt
unrban-building-damage-detection-by-PJLi.ppt
 
Transfer Learning Model for Image Segmentation by Integrating U-NetPlusPlus a...
Transfer Learning Model for Image Segmentation by Integrating U-NetPlusPlus a...Transfer Learning Model for Image Segmentation by Integrating U-NetPlusPlus a...
Transfer Learning Model for Image Segmentation by Integrating U-NetPlusPlus a...
 
Deep Learning based Segmentation Pipeline for Label-Free Phase-Contrast Micro...
Deep Learning based Segmentation Pipeline for Label-Free Phase-Contrast Micro...Deep Learning based Segmentation Pipeline for Label-Free Phase-Contrast Micro...
Deep Learning based Segmentation Pipeline for Label-Free Phase-Contrast Micro...
 
Yolo
YoloYolo
Yolo
 
Automatic Building detection for satellite Images using IGV and DSM
Automatic Building detection for satellite Images using IGV and DSMAutomatic Building detection for satellite Images using IGV and DSM
Automatic Building detection for satellite Images using IGV and DSM
 
Fn2611681170
Fn2611681170Fn2611681170
Fn2611681170
 
Face Recognition - Elisa Sayrol - UPC Barcelona 2018
Face Recognition - Elisa Sayrol - UPC Barcelona 2018Face Recognition - Elisa Sayrol - UPC Barcelona 2018
Face Recognition - Elisa Sayrol - UPC Barcelona 2018
 
SPIE 10059-36(Reheman Baikejiang)
SPIE 10059-36(Reheman Baikejiang)SPIE 10059-36(Reheman Baikejiang)
SPIE 10059-36(Reheman Baikejiang)
 
Александр Заричковый "Faster than real-time face detection"
Александр Заричковый "Faster than real-time face detection"Александр Заричковый "Faster than real-time face detection"
Александр Заричковый "Faster than real-time face detection"
 
Ieee 2016 nss mic poster N30-21
Ieee 2016 nss mic poster N30-21Ieee 2016 nss mic poster N30-21
Ieee 2016 nss mic poster N30-21
 
ieee nss mic 2016 poster N30-21
ieee nss mic 2016 poster N30-21ieee nss mic 2016 poster N30-21
ieee nss mic 2016 poster N30-21
 
Knowledge-based Fusion for Image Tampering Localization
Knowledge-based Fusion for Image Tampering LocalizationKnowledge-based Fusion for Image Tampering Localization
Knowledge-based Fusion for Image Tampering Localization
 
Avihu Efrat's Viola and Jones face detection slides
Avihu Efrat's Viola and Jones face detection slidesAvihu Efrat's Viola and Jones face detection slides
Avihu Efrat's Viola and Jones face detection slides
 
KaoNet: Face Recognition and Generation App using Deep Learning
KaoNet: Face Recognition and Generation App using Deep LearningKaoNet: Face Recognition and Generation App using Deep Learning
KaoNet: Face Recognition and Generation App using Deep Learning
 
IRJET - Real Time Object Detection using YOLOv3
IRJET - Real Time Object Detection using YOLOv3IRJET - Real Time Object Detection using YOLOv3
IRJET - Real Time Object Detection using YOLOv3
 
De-convolution on Digital Images
De-convolution on Digital ImagesDe-convolution on Digital Images
De-convolution on Digital Images
 
THE EFFECT OF PHYSICAL BASED FEATURES FOR RECOGNITION OF RECAPTURED IMAGES
THE EFFECT OF PHYSICAL BASED FEATURES FOR RECOGNITION OF RECAPTURED IMAGESTHE EFFECT OF PHYSICAL BASED FEATURES FOR RECOGNITION OF RECAPTURED IMAGES
THE EFFECT OF PHYSICAL BASED FEATURES FOR RECOGNITION OF RECAPTURED IMAGES
 

Similar to Segmentation Improves Melanoma Classification Accuracy

Detection and recognition of face using neural network
Detection and recognition of face using neural networkDetection and recognition of face using neural network
Detection and recognition of face using neural networkSmriti Tikoo
 
Face recognition: A Comparison of Appearance Based Approaches
Face recognition: A Comparison of Appearance Based ApproachesFace recognition: A Comparison of Appearance Based Approaches
Face recognition: A Comparison of Appearance Based Approachessadique_ghitm
 
IP%20FACE%20RECOGNITION%20ATTENDANCE%20SYSTEM.pptx
IP%20FACE%20RECOGNITION%20ATTENDANCE%20SYSTEM.pptxIP%20FACE%20RECOGNITION%20ATTENDANCE%20SYSTEM.pptx
IP%20FACE%20RECOGNITION%20ATTENDANCE%20SYSTEM.pptxJyotiradityaPandey1
 
Face and Eye Detection Varying Scenarios With Haar Classifier_2015
Face and Eye Detection Varying Scenarios With Haar Classifier_2015Face and Eye Detection Varying Scenarios With Haar Classifier_2015
Face and Eye Detection Varying Scenarios With Haar Classifier_2015Showrav Mazumder
 
braintumordetectionusingimagesegmentationppt-210830184640 (1).pdf
braintumordetectionusingimagesegmentationppt-210830184640 (1).pdfbraintumordetectionusingimagesegmentationppt-210830184640 (1).pdf
braintumordetectionusingimagesegmentationppt-210830184640 (1).pdfSunnyYadav735981
 
braintumordetectionusingimagesegmentationppt-210830184640.pdf
braintumordetectionusingimagesegmentationppt-210830184640.pdfbraintumordetectionusingimagesegmentationppt-210830184640.pdf
braintumordetectionusingimagesegmentationppt-210830184640.pdfAlHussieniAbdulAziz
 
braintumordetectionusingimagesegmentationppt-210830184640.pdf
braintumordetectionusingimagesegmentationppt-210830184640.pdfbraintumordetectionusingimagesegmentationppt-210830184640.pdf
braintumordetectionusingimagesegmentationppt-210830184640.pdfAlHussieniAbdulAziz
 
Brain tumor detection using image segmentation ppt
Brain tumor detection using image segmentation pptBrain tumor detection using image segmentation ppt
Brain tumor detection using image segmentation pptRoshini Vijayakumar
 
Cross Pose Facial Recognition Method for Tracking any Person's Location an Ap...
Cross Pose Facial Recognition Method for Tracking any Person's Location an Ap...Cross Pose Facial Recognition Method for Tracking any Person's Location an Ap...
Cross Pose Facial Recognition Method for Tracking any Person's Location an Ap...ijtsrd
 
Report Face Detection
Report Face DetectionReport Face Detection
Report Face DetectionJugal Patel
 
FACE DETECTION USING PRINCIPAL COMPONENT ANALYSIS
FACE DETECTION USING PRINCIPAL COMPONENT ANALYSISFACE DETECTION USING PRINCIPAL COMPONENT ANALYSIS
FACE DETECTION USING PRINCIPAL COMPONENT ANALYSISIAEME Publication
 
Face Recognition Technology
Face Recognition TechnologyFace Recognition Technology
Face Recognition TechnologyShravan Halankar
 
Facial recognition
Facial recognitionFacial recognition
Facial recognitionSonam1891
 
IRJET- Survey on Face Detection Methods
IRJET- Survey on Face Detection MethodsIRJET- Survey on Face Detection Methods
IRJET- Survey on Face Detection MethodsIRJET Journal
 
Face Recognition Based Attendance System with Auto Alert to Guardian using Ca...
Face Recognition Based Attendance System with Auto Alert to Guardian using Ca...Face Recognition Based Attendance System with Auto Alert to Guardian using Ca...
Face Recognition Based Attendance System with Auto Alert to Guardian using Ca...ijtsrd
 

Similar to Segmentation Improves Melanoma Classification Accuracy (20)

Face recognition
Face recognition Face recognition
Face recognition
 
Detection and recognition of face using neural network
Detection and recognition of face using neural networkDetection and recognition of face using neural network
Detection and recognition of face using neural network
 
ACET.pptx
ACET.pptxACET.pptx
ACET.pptx
 
Face recognition: A Comparison of Appearance Based Approaches
Face recognition: A Comparison of Appearance Based ApproachesFace recognition: A Comparison of Appearance Based Approaches
Face recognition: A Comparison of Appearance Based Approaches
 
IP%20FACE%20RECOGNITION%20ATTENDANCE%20SYSTEM.pptx
IP%20FACE%20RECOGNITION%20ATTENDANCE%20SYSTEM.pptxIP%20FACE%20RECOGNITION%20ATTENDANCE%20SYSTEM.pptx
IP%20FACE%20RECOGNITION%20ATTENDANCE%20SYSTEM.pptx
 
face recognition
face recognitionface recognition
face recognition
 
dt.ppt
dt.pptdt.ppt
dt.ppt
 
Face and Eye Detection Varying Scenarios With Haar Classifier_2015
Face and Eye Detection Varying Scenarios With Haar Classifier_2015Face and Eye Detection Varying Scenarios With Haar Classifier_2015
Face and Eye Detection Varying Scenarios With Haar Classifier_2015
 
braintumordetectionusingimagesegmentationppt-210830184640 (1).pdf
braintumordetectionusingimagesegmentationppt-210830184640 (1).pdfbraintumordetectionusingimagesegmentationppt-210830184640 (1).pdf
braintumordetectionusingimagesegmentationppt-210830184640 (1).pdf
 
braintumordetectionusingimagesegmentationppt-210830184640.pdf
braintumordetectionusingimagesegmentationppt-210830184640.pdfbraintumordetectionusingimagesegmentationppt-210830184640.pdf
braintumordetectionusingimagesegmentationppt-210830184640.pdf
 
braintumordetectionusingimagesegmentationppt-210830184640.pdf
braintumordetectionusingimagesegmentationppt-210830184640.pdfbraintumordetectionusingimagesegmentationppt-210830184640.pdf
braintumordetectionusingimagesegmentationppt-210830184640.pdf
 
Brain tumor detection using image segmentation ppt
Brain tumor detection using image segmentation pptBrain tumor detection using image segmentation ppt
Brain tumor detection using image segmentation ppt
 
Cross Pose Facial Recognition Method for Tracking any Person's Location an Ap...
Cross Pose Facial Recognition Method for Tracking any Person's Location an Ap...Cross Pose Facial Recognition Method for Tracking any Person's Location an Ap...
Cross Pose Facial Recognition Method for Tracking any Person's Location an Ap...
 
Report Face Detection
Report Face DetectionReport Face Detection
Report Face Detection
 
FACE DETECTION USING PRINCIPAL COMPONENT ANALYSIS
FACE DETECTION USING PRINCIPAL COMPONENT ANALYSISFACE DETECTION USING PRINCIPAL COMPONENT ANALYSIS
FACE DETECTION USING PRINCIPAL COMPONENT ANALYSIS
 
Face Recognition Technology
Face Recognition TechnologyFace Recognition Technology
Face Recognition Technology
 
Facial recognition
Facial recognitionFacial recognition
Facial recognition
 
IRJET- Survey on Face Detection Methods
IRJET- Survey on Face Detection MethodsIRJET- Survey on Face Detection Methods
IRJET- Survey on Face Detection Methods
 
Face Recognition Based Attendance System with Auto Alert to Guardian using Ca...
Face Recognition Based Attendance System with Auto Alert to Guardian using Ca...Face Recognition Based Attendance System with Auto Alert to Guardian using Ca...
Face Recognition Based Attendance System with Auto Alert to Guardian using Ca...
 
Anits dip
Anits dipAnits dip
Anits dip
 

More from Oge Marques

Visual Information Retrieval: Advances, Challenges and Opportunities
Visual Information Retrieval: Advances, Challenges and OpportunitiesVisual Information Retrieval: Advances, Challenges and Opportunities
Visual Information Retrieval: Advances, Challenges and OpportunitiesOge Marques
 
Image Processing and Computer Vision in iOS
Image Processing and Computer Vision in iOSImage Processing and Computer Vision in iOS
Image Processing and Computer Vision in iOSOge Marques
 
Using games to improve computer vision solutions
Using games to improve computer vision solutionsUsing games to improve computer vision solutions
Using games to improve computer vision solutionsOge Marques
 
Image retrieval: challenges and opportunities
Image retrieval: challenges and opportunitiesImage retrieval: challenges and opportunities
Image retrieval: challenges and opportunitiesOge Marques
 
Advances and Challenges in Visual Information Search and Retrieval (WVC 2012 ...
Advances and Challenges in Visual Information Search and Retrieval (WVC 2012 ...Advances and Challenges in Visual Information Search and Retrieval (WVC 2012 ...
Advances and Challenges in Visual Information Search and Retrieval (WVC 2012 ...Oge Marques
 
Mobile Visual Search
Mobile Visual SearchMobile Visual Search
Mobile Visual SearchOge Marques
 
Advances in Image Search and Retrieval
Advances in Image Search and RetrievalAdvances in Image Search and Retrieval
Advances in Image Search and RetrievalOge Marques
 
Image Processing and Computer Vision in iPhone and iPad
Image Processing and Computer Vision in iPhone and iPadImage Processing and Computer Vision in iPhone and iPad
Image Processing and Computer Vision in iPhone and iPadOge Marques
 
Recent advances in visual information retrieval marques klu june 2010
Recent advances in visual information retrieval marques klu june 2010Recent advances in visual information retrieval marques klu june 2010
Recent advances in visual information retrieval marques klu june 2010Oge Marques
 
Oge Marques (FAU) - invited talk at WISMA 2010 (Barcelona, May 2010)
Oge Marques (FAU) - invited talk at WISMA 2010 (Barcelona, May 2010)Oge Marques (FAU) - invited talk at WISMA 2010 (Barcelona, May 2010)
Oge Marques (FAU) - invited talk at WISMA 2010 (Barcelona, May 2010)Oge Marques
 

More from Oge Marques (10)

Visual Information Retrieval: Advances, Challenges and Opportunities
Visual Information Retrieval: Advances, Challenges and OpportunitiesVisual Information Retrieval: Advances, Challenges and Opportunities
Visual Information Retrieval: Advances, Challenges and Opportunities
 
Image Processing and Computer Vision in iOS
Image Processing and Computer Vision in iOSImage Processing and Computer Vision in iOS
Image Processing and Computer Vision in iOS
 
Using games to improve computer vision solutions
Using games to improve computer vision solutionsUsing games to improve computer vision solutions
Using games to improve computer vision solutions
 
Image retrieval: challenges and opportunities
Image retrieval: challenges and opportunitiesImage retrieval: challenges and opportunities
Image retrieval: challenges and opportunities
 
Advances and Challenges in Visual Information Search and Retrieval (WVC 2012 ...
Advances and Challenges in Visual Information Search and Retrieval (WVC 2012 ...Advances and Challenges in Visual Information Search and Retrieval (WVC 2012 ...
Advances and Challenges in Visual Information Search and Retrieval (WVC 2012 ...
 
Mobile Visual Search
Mobile Visual SearchMobile Visual Search
Mobile Visual Search
 
Advances in Image Search and Retrieval
Advances in Image Search and RetrievalAdvances in Image Search and Retrieval
Advances in Image Search and Retrieval
 
Image Processing and Computer Vision in iPhone and iPad
Image Processing and Computer Vision in iPhone and iPadImage Processing and Computer Vision in iPhone and iPad
Image Processing and Computer Vision in iPhone and iPad
 
Recent advances in visual information retrieval marques klu june 2010
Recent advances in visual information retrieval marques klu june 2010Recent advances in visual information retrieval marques klu june 2010
Recent advances in visual information retrieval marques klu june 2010
 
Oge Marques (FAU) - invited talk at WISMA 2010 (Barcelona, May 2010)
Oge Marques (FAU) - invited talk at WISMA 2010 (Barcelona, May 2010)Oge Marques (FAU) - invited talk at WISMA 2010 (Barcelona, May 2010)
Oge Marques (FAU) - invited talk at WISMA 2010 (Barcelona, May 2010)
 

Recently uploaded

GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]📊 Markus Baersch
 
Semantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptxSemantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptxMike Bennett
 
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...Amil Baba Dawood bangali
 
2006_GasProcessing_HB (1).pdf HYDROCARBON PROCESSING
2006_GasProcessing_HB (1).pdf HYDROCARBON PROCESSING2006_GasProcessing_HB (1).pdf HYDROCARBON PROCESSING
2006_GasProcessing_HB (1).pdf HYDROCARBON PROCESSINGmarianagonzalez07
 
Biometric Authentication: The Evolution, Applications, Benefits and Challenge...
Biometric Authentication: The Evolution, Applications, Benefits and Challenge...Biometric Authentication: The Evolution, Applications, Benefits and Challenge...
Biometric Authentication: The Evolution, Applications, Benefits and Challenge...GQ Research
 
Predicting Salary Using Data Science: A Comprehensive Analysis.pdf
Predicting Salary Using Data Science: A Comprehensive Analysis.pdfPredicting Salary Using Data Science: A Comprehensive Analysis.pdf
Predicting Salary Using Data Science: A Comprehensive Analysis.pdfBoston Institute of Analytics
 
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改yuu sss
 
How we prevented account sharing with MFA
How we prevented account sharing with MFAHow we prevented account sharing with MFA
How we prevented account sharing with MFAAndrei Kaleshka
 
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...dajasot375
 
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...Jack DiGiovanna
 
Student Profile Sample report on improving academic performance by uniting gr...
Student Profile Sample report on improving academic performance by uniting gr...Student Profile Sample report on improving academic performance by uniting gr...
Student Profile Sample report on improving academic performance by uniting gr...Seán Kennedy
 
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degreeyuu sss
 
Student profile product demonstration on grades, ability, well-being and mind...
Student profile product demonstration on grades, ability, well-being and mind...Student profile product demonstration on grades, ability, well-being and mind...
Student profile product demonstration on grades, ability, well-being and mind...Seán Kennedy
 
While-For-loop in python used in college
While-For-loop in python used in collegeWhile-For-loop in python used in college
While-For-loop in python used in collegessuser7a7cd61
 
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...ssuserf63bd7
 
From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...Florian Roscheck
 
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一fhwihughh
 
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024thyngster
 
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...Boston Institute of Analytics
 
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理e4aez8ss
 

Recently uploaded (20)

GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]
 
Semantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptxSemantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptx
 
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...
 
2006_GasProcessing_HB (1).pdf HYDROCARBON PROCESSING
2006_GasProcessing_HB (1).pdf HYDROCARBON PROCESSING2006_GasProcessing_HB (1).pdf HYDROCARBON PROCESSING
2006_GasProcessing_HB (1).pdf HYDROCARBON PROCESSING
 
Biometric Authentication: The Evolution, Applications, Benefits and Challenge...
Biometric Authentication: The Evolution, Applications, Benefits and Challenge...Biometric Authentication: The Evolution, Applications, Benefits and Challenge...
Biometric Authentication: The Evolution, Applications, Benefits and Challenge...
 
Predicting Salary Using Data Science: A Comprehensive Analysis.pdf
Predicting Salary Using Data Science: A Comprehensive Analysis.pdfPredicting Salary Using Data Science: A Comprehensive Analysis.pdf
Predicting Salary Using Data Science: A Comprehensive Analysis.pdf
 
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
专业一比一美国俄亥俄大学毕业证成绩单pdf电子版制作修改
 
How we prevented account sharing with MFA
How we prevented account sharing with MFAHow we prevented account sharing with MFA
How we prevented account sharing with MFA
 
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
 
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
 
Student Profile Sample report on improving academic performance by uniting gr...
Student Profile Sample report on improving academic performance by uniting gr...Student Profile Sample report on improving academic performance by uniting gr...
Student Profile Sample report on improving academic performance by uniting gr...
 
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
 
Student profile product demonstration on grades, ability, well-being and mind...
Student profile product demonstration on grades, ability, well-being and mind...Student profile product demonstration on grades, ability, well-being and mind...
Student profile product demonstration on grades, ability, well-being and mind...
 
While-For-loop in python used in college
While-For-loop in python used in collegeWhile-For-loop in python used in college
While-For-loop in python used in college
 
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
 
From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...
 
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
 
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024
 
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...
NLP Data Science Project Presentation:Predicting Heart Disease with NLP Data ...
 
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
 

Segmentation Improves Melanoma Classification Accuracy

  • 1. #SIIM17 The Impact of Segmentation on the Accuracy and Sensitivity of a Melanoma Classifier Based on Skin Lesion Images Oge Marques, PhD Professor College of Engineering and Computer Science Florida Atlantic University
  • 2. #SIIM17 Our Team Adrià Romero López Xavier Giró-i.Nieto Image Processing Group Signal Theory and Communications Department MIDDLE Research Group Oge Marques Borko Furht Jack Burdick Janet Weinthal NSF Award No. 1464537, I/UCRC Phase II under NSF 13-542
  • 3. #SIIM17 Outline • Motivation • Context • Scope and goals • Challenges • State of the art • Our work • Hypothesis • Methodology • Experimental results • Ongoing and future work • Concluding remarks
  • 5. #SIIM17 Context • This is not a typical SIIM presentation • No specialized imaging equipment • No PACS • No DICOM • No metadata • No workflows • Instead... • Regular photographs • Unstructured (and purely visual) data • Minimal ground truth
  • 6. #SIIM17 Scope • Skin Disease: An Illustrated Taxonomy • Our focus: skin lesion analysis for (early) melanoma detection [Source: Esteva et al., Nature (2017)]
  • 7. #SIIM17 Scope and Goals • Scope: • Help physicians to detect melanoma (a 2-class classifier) • Goals: • Design an intelligent medical imaging skin lesion diagnosis system using deep learning techniques • Achieve (or improve upon) state-of-the-art results for skin lesion classification
  • 9. #SIIM17 Success Rate - Sensitivity Physicians, as low as: • 43 % - naked eye • 79 % - dermoscopy (Vestergaard et al, 2008) Deep learning based methods, as high as: • 94.83 % (Jafari et al, 2016; Premaladha and Ravichandran 2016)
  • 10. #SIIM17 Classical Techniques [Source: Center For Excellence In Dermatology - Kennewick, WA] CASH (Henning et al., 2007) Color Architecture Symmetry Homogeneity
  • 12. #SIIM17 Challenges • Segmentation • Artifacts, such as: freckles, hair, shading, and noise. • Irregular, fuzzy lesion border • Low contrast between lesion and surrounding skin • Multi-fragment lesions • Classification • Manual feature extraction is not effective • False positives and false negatives have different impact
  • 13. #SIIM17 State of the art • Esteva et al. Nature 2017
  • 14. #SIIM17 Our Approach • Pretrained VGG16 (Simonyan & Zisserman, 2014) • Transfer Learning [Source:Blier 2016]
  • 16. #SIIM17 Transfer Learning 1. Train on Imagenet 3. Medium dataset: finetuning more data = retrain more of the network (or all of it) 2. Small dataset: feature extractor Freeze these Train this Freeze these Train this Slide credit: Bay Area Deep Learning School Presentation by A. Karpathy Medical Imaging case
  • 17. #SIIM17 Prediction Examples: Getting It Right True Positives True Negatives
  • 18. #SIIM17 Prediction Examples: Getting It Wrong False Positives False Negatives
  • 19. #SIIM17 Our Hypothesis • Image segmentation improves the performance of skin lesion classifiers using convolutional neural networks. [Source:International Skin Imaging Collaboration Archive] Not segmented Perfectly segmented Partially segmented
  • 20. #SIIM17 Methods • ISBI 2016 Challenge dataset • Skin Lesion Analysis towards melanoma detection • 1279 RGB images • Labeled as either benign or malignant Class Benign Malignant Total Images Training subset 727 173 900 Testing subset 304 75 379
  • 21. #SIIM17 Methods • Dataset balancing through downsampling. • Dataset split: 70-30% training/testing • Input images: • Unsegmented images: straight from the dataset. • Perfectly segmented images: bitwise AND operation of the unaltered images and its corresponding binary mask provided by the ISIC dataset. • Partially segmented images: original binary masks morphologically dilated with a disk-shaped structuring element (50 pixel radius). • Additional preprocessing methods (resizing and normalization) were also performed to match the input size expected by the VGG16 architecture.
  • 23. #SIIM17 Further Investigation • What if we vary the degree of border expansion? Sensitivity Accuracy AUC Perfect Segmentation 45.3% 58.7% 62.2% +25 53.3% 61.3% 64.2% +50 56.0% 60.7% 62.6% +75 57.3% 59.3% 60.8% +100 34.7% 55.3% 57.9% Unsegmented 24.0% 51.3% 53.2%
  • 24. #SIIM17 Ongoing and Future Work •Additional / larger / more challenging datasets •Other CNN architectures •Better image preprocessing •Partnerships and collaborations •Mobile app
  • 25. #SIIM17 Concluding remarks • Challenges • Difficulty in acquiring datasets and reproducing / benchmarking results • The “black box” aspect of DL-based solutions • Hard to tell positives from negatives • Learning curve: TensorFlow, Keras, HPC, DL concepts and best practices, etc. • Opportunities • Many variations of the basic classification problem • Mobile app market • Tech-minded dermatology practices