Databases for mammography Maria Evelina Fantacci Dipar8mento di - - PowerPoint PPT Presentation

▶
databases for mammography
SMART_READER_LITE
LIVE PREVIEW

Databases for mammography Maria Evelina Fantacci Dipar8mento di - - PowerPoint PPT Presentation

Jan 11, 2024 370 likes •1.31k views

Databases for mammography Maria Evelina Fantacci Dipar8mento di Fisica and INFN Pisa, IT 3 Training School on Applica8on of computer models for advancement of X-ray breast imaging techniques Napoli, IT, 17-19 September 2018 OUTLINE

slide-1
SLIDE 1

Databases for mammography

Maria Evelina Fantacci Dipar8mento di Fisica and INFN Pisa, IT

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-2
SLIDE 2

OUTLINE

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Databases for mammography:

  • Introduc8on (“personal”)
  • Historical overview
  • Applica8ons
  • Requirements
  • Examples of public research databases
  • Conclusions (“personal”)
  • Perspec8ves
  • Conclusions
slide-3
SLIDE 3

Introduc<on (“personal”)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Thanks for this invita8on-opportunity to collect and evaluate the different aspects related to the topic “Databases for mammography”.

  • Very current, evergreen topic …

… for at least the last 25 years … with ups and downs … with different meanings, contents, applica8ons … strictly related to fundamental progress in imaging instrumenta8on storage and computa8onal power so[ware development

slide-4
SLIDE 4

Introduc<on (“personal”)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Other principal works but also Involved in:

  • Collec8on of an italian mul8centric dataset of digi8zed

mammograms (end of last century-millennium).

  • Collec8on of a pan-european dataset of digi8zed and digital

mammograms (beginning of this century-millennium). Dra[ of a book chapter about mammographic databases (few years ago). Meanwhile: physiological changes and news.

This year: need for a dataset for breast density studies (RADIOMA project). No available public annotated database of digital images OK for density studies with deep learning techniques.

slide-5
SLIDE 5

Historical overview

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • First working groups
  • Digi8zed mammograms
  • Digital mammograms
slide-6
SLIDE 6

Historical overview

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

1993 US:

Panel discussion, moderator F. Shtern, at IS&T/SPIE Symposium on Electronic Imaging: Science and Technology, 1993, San Jose, CA, US. SPIE Conference on Biomedical Image Processing and Biomedical Visualiza8on, 1993 Proceedings volume 1005.

slide-7
SLIDE 7

Historical overview

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

1994 UK:

slide-8
SLIDE 8

Historical overview

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

1995 US:

slide-9
SLIDE 9

Historical overview

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-10
SLIDE 10

Historical overview

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-11
SLIDE 11

Historical overview

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

2005 EU: The first Full Field Digital Mammography images in a research database (28 exams from Valdese Hospital, Turin. IT)

slide-12
SLIDE 12

Applica<ons

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Teaching and training
  • Archive of par8cular cases
  • Sta8s8cal and epidemiological studies
  • Development and valida8on of so[ware analysis tools
slide-13
SLIDE 13

Teaching and training

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-14
SLIDE 14

Teaching and training

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-15
SLIDE 15

Teaching and training

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-16
SLIDE 16

Teaching and training

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-17
SLIDE 17

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Archive of par<cular cases

slide-18
SLIDE 18

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Archive of par<cular cases

slide-19
SLIDE 19

Sta<s<cal and epidemiological studies

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

2016 US:

slide-20
SLIDE 20

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Sta<s<cal and epidemiological studies

slide-21
SLIDE 21

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Sta<s<cal and epidemiological studies

slide-22
SLIDE 22

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Sta<s<cal and epidemiological studies

slide-23
SLIDE 23

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Sta<s<cal and epidemiological studies

slide-24
SLIDE 24

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Development and valida<on of soQware analysis tools

  • CAD (Computer Aided Detec8on/Diagnosis) systems

(more than 1000 pubblica8ons)

  • Microcalcifica8on clusters
  • Tumour masses
  • Density assessment
  • BIRADS classes
  • Segmenta8on
slide-25
SLIDE 25

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

CAD systems: first aTempts

slide-26
SLIDE 26

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Microcalcifica<on clusters

CAD systems: first aTempts

slide-27
SLIDE 27

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Microcalcifica<on clusters

CAD systems: first aTempts

slide-28
SLIDE 28

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Tumour masses

CAD systems: first aTempts

slide-29
SLIDE 29

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

with MIAS database

CAD systems: first aTempts

slide-30
SLIDE 30

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

with mini-MIAS database

CAD systems today: recent work

slide-31
SLIDE 31

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

deep learning based

CAD systems today:

slide-32
SLIDE 32

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

CAD systems today: DREAM challenge

First classified: Yaroslav Nikulin THERAPIXEL (French company) and and Yuanfang Guan (Dept. of Computa8onal Medicine and Bioinforma8cs, University of Michigan).

slide-33
SLIDE 33

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

CAD systems today: DREAM challenge

slide-34
SLIDE 34

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

synte<c images

CAD systems today:

slide-35
SLIDE 35

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Density assessment

A D C B ACR (American College of Radiology) breast imaging repor8ng and data system (BIRADS)

slide-36
SLIDE 36

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Density assessment

slide-37
SLIDE 37

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Segmenta<on: density

slide-38
SLIDE 38

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Contents and requirements

  • Case Selec8on
  • Ground Truth
  • Associated Informa8ons
  • Requirements of the Digi8zers

(digital images characteris8cs)

  • Organiza8on
  • Distribu8on
slide-39
SLIDE 39

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Contents and requirements

slide-40
SLIDE 40

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Case Selec<on

The database should include various cases with images with none and all types of findings, and also all types of breast density. Normal images with structures that may be misleading (e.g. superimposed 8ssue that looks like a mass) are important in order to make the classifiers more robust. The cases should be collected by a specialist experienced in mammography, and each case should contain the four standard views, unless it is a case from a pa8ent with one breast only. It is considered that for each 100 cases, approximately 200 images should contain a lesion.

slide-41
SLIDE 41

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Case Selec<on

The database should include various cases with images with none and all types of findings, and also all types of breast density. Normal images with structures that may be misleading (e.g. superimposed 8ssue that looks like a mass) are important in order to make the classifiers more robust. The cases should be collected by a specialist experienced in mammography, and each case should contain the four standard views, unless it is a case from a pa8ent with one breast only. It is considered that for each 100 cases, approximately 200 images should contain a lesion.

slide-42
SLIDE 42

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Ground Truth

Biopsy proof for all cases should be available, and for cases in which a biopsy is not recommended, the mammography should have the same BI-RADS for at least three years. Annota8ons should include the “ground truth” concerning the degree of malignancy, the loca8on and the boundary of the lesion and this outline should be performed by a specialist.

slide-43
SLIDE 43

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Ground Truth

Biopsy proof for all cases should be available, and for cases in which a biopsy is not recommended, the mammography should have the same BI-RADS for at least three years. Annota8ons should include the “ground truth” concerning the degree of malignancy, the loca8on and the boundary of the lesion and this outline should be performed by a specialist.

slide-44
SLIDE 44

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Associated Informa<on

Clinical history, age, family history, previous biopsies.

Requirements of the digi<zer

This is s8ll a point of controversy, but one common approach is to digitalize at a very small pixel size, for example, at 25 microns.

Organiza<on of Database

A specific file format for digital mammograms does not exist. Medical images are usually saved in the DICOM (Digital Imaging and Communica8ons in Medicine) format that gathers not only the image but also some related metadata.

Distribu<on of Database

The database should be available, preferen8ally over the World Wide

  • Web. Con8nuous user support is also indispensable.
slide-45
SLIDE 45

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Associated Informa<on

Clinical history, age, family history, previous biopsies.

Requirements of the digi<zer

This is s8ll a point of controversy, but one common approach is to digitalize at a very small pixel size, for example, at 25 microns.

Organiza<on of Database

A specific file format for digital mammograms does not exist. Medical images are usually saved in the DICOM (Digital Imaging and Communica8ons in Medicine) format that gathers not only the image but also some related metadata.

Distribu<on of Database

The database should be available, preferen8ally over the World Wide

  • Web. Con8nuous user support is also indispensable.
slide-46
SLIDE 46

Examples of public research databases

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • MIAS (Mammographic Image Analysis Society)

mini-MIAS

  • DDSM (Digital Database for Screening Mammography)

CBIS-DDSM (Curated Breast Imaging Subset of DDSM)

  • MAGIC-5
  • BancoWeb LAPIMO
  • BCDR (Breast Cancer Digital Repository)
  • INbreast
  • LLNL/UCSF
  • MammoGrid Project
  • AMDI (Indexed Atlas of Digital Mammograms)
  • IRMA (Image Retrieval in Medical Applica8ons)
  • DBMS (DataBase Management System)
slide-47
SLIDE 47

MIAS database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Mammographic Image Analysis Society.
  • One center (UK), Na8onal Breast Screening Programme.
  • 322 screening images (161 pairs), only MLO view.
  • Digi8zed films (50 µm, 8 bit, PGM).
  • Info about density and abnormai8es (no BIRADS).
  • Various releases (different format, pixel size).

from 1998 to 2015.

slide-48
SLIDE 48

MIAS database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

hjps://www.repository.cam.ac.uk/handle/1810/250394

slide-49
SLIDE 49

mini-MIAS database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

hjp://peipa.essex.ac.uk/info/mias.html

slide-50
SLIDE 50

DDSM database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Digital Database for Screening Mammography.
  • Mul8centric (US): Massachusejs General Hospital,

University of South Florida, Sandia Na8onal Laboratories (1988-1999).

  • 2620 screening exams, 4 views/exam.
  • Digi8zed films (42,0, 43,5, 50,0 µm, 8/16 bit, PGM, TIFF,

non standard lossless JPEG).

  • Info about density and normal/benign/cancer (ACR-

BIRADS).

  • Various releases (different format, SW tools) from 1999.
slide-51
SLIDE 51

DDSM database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

hjp://marathon.csee.usf.edu/Mammography/Database.html

slide-52
SLIDE 52

DDSM database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-53
SLIDE 53

DDSM database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-54
SLIDE 54

CBIS-DDSM (Curated Breast Imaging Subset of DDSM)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-55
SLIDE 55

CBIS-DDSM (Curated Breast Imaging Subset of DDSM)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-56
SLIDE 56

MAGIC-5 database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Medical Applica8ons in a Grid Infrastructure Connec8on.
  • Mul8centric (IT): hospitals of Bari, Udine, Palermo,

Firenze, Torino, Napoli, Sassari (1999-2003).

  • 967 screening and clinical exams, 3369 images.
  • Digi8zed films (85 µm, 12 bit, DICOM).
  • Info about density (3 classes) and normal/benign/cancer

(ACR).

slide-57
SLIDE 57

BancoWeb LAPIMO database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Two centers (BR): San Paulo University (2011).
  • 320 most screening exams, 1400 images.
  • Digi8zed films (85, 150 µm, 12 bit, TIFF).
  • Info about density and pathology (BIRADS).
slide-58
SLIDE 58

BCDR database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Breast Cancer Digital Repository
  • Mul8center (PT and SP): Centro Hospitalar São João,

University of Porto, CETA-CIEMAT, Aviero University (2009-2013).

  • Digi8zed films (FM, Film Mammography) and DM

(Digital Mammography).

  • FM: 1010 pa8ents, 1125 exams, 3703 images.

Matrices: 720x1168, 8 bits, TIFF.

  • DM: 724 pa8ents, 1042 exams, 3612 images.

Matrices: 3328x4084 or 2560x3328 pixels, 14 bits, TIFF.

  • Info about pathology (BIRADS) and related clinical data.
slide-59
SLIDE 59

BCDR database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

hjps://bcdr.ceta-ciemat.es

slide-60
SLIDE 60

INbreast database

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

hjp://medicalresearch.inescporto.pt/breastresearch/ index.php/Get_INbreast_Database Released in 2012, similar to BCDM-DR

slide-61
SLIDE 61

LLNL/UCSF

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Mul8center (US): Lawrence Livermore Na8onal

Laboratories, University of California at San Francisco (UCSF) Radiology Deartment (2016).

  • Digi8zed films, 35 µm, 12 bits, .
  • 50 cases, 198 images.
  • 5 normal cases, 5 normal but difficult, 20 of obviously

benign microcalcifica8ons, 12 of suspicious, benign microcalcifica8ons, and 8 with malignant cluster of microcalcifica8ons.

hjp://marathon.csee.usf.edu/Mammography/OtherResources.html

slide-62
SLIDE 62

MammoGrid Project

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

As grid compu8ng promised to resolve many of the difficul8es in facilita8ng medical image analysis to allow clinicians to collaborate without having to co-locate, the MammoGrid project has inves8gated the feasibility of developing a Grid- enabled European database of mammograms so that a set of important healthcare applica8ons using a mammographic database have been enabled and the poten8al of the Grids has been harnessed to support co-working between healthcare professionals across the EU.

slide-63
SLIDE 63

MammoGrid Project

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Database: part of MAGIC-5 (Udine and Torino hospitals) and Cambridge University. AIMS:

  • to evaluate Grid technologies in implemen8ng a mammographic database using novel

Grid-compliant;

  • to provide improved access to distributed data and allow rapid deployment of so[ware

packages;

  • to operate on locally stored informa8on;
  • to deploy a standardiza8on system (SMF, Standard Mammogram Form) that enables

comparison of mammograms in terms of intrinsic 8ssue proper8es independently of scanner seqngs;

  • to develop so[ware tools to automa8cally extract image informa8on that can be used

to perform quality controls on the acquisi8on);

  • to develop so[ware tools to automa8cally extract 8ssue informa8on that can be used

to perform clinical studies.

slide-64
SLIDE 64

AMDI system

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Indexed Atlas of Digital Mammograms
  • Hospital de clinicas de Uberlandia (BR) and Ins8tuto Victorio

Valeri de diagnos8cos medicos (Brazil)

  • Based on an eXtension Rela8onal Database Management System

(XRDBMS)

  • Includes SISPRIM (Sistema de Pesquisa para Recuperação de

Imagens Mamográficas), a Research System for Retrieval of Mammographic Images for sta8s8cal studies.

  • Includes INDIAM (INterpreta8on and DIAgnosis of Mammograms)

for e-learning, comprising a module to simulate the analysis and diagnosis of breast cancer and a module for training the student in the interpreta8on of mammograms .

slide-65
SLIDE 65

AMDI system

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-66
SLIDE 66

IRMA project

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Departments of Diagnos8c Radiology and of Medical

Informa8cs, Aachen University of Technology.

  • Aim: development and implementa8on of high-level methods

for content-based image retrieval with prototypical applica8on to medico-diagnos8c tasks on a radiologic image archive.

  • Includes: 10509 images (MIAS, DDSM, LLNL databases and

images from the Rheinisch-WestÃlische Technische Hochschule, Aachen) and standardized coding of 8ssue type, tumor staging, and lesion descrip8on according to the ACR 8ssue codes and BIRADS.

  • It can be extended easily with further cases imported from a

picture archiving and communica8on system (PACS).

slide-67
SLIDE 67

DBMS (DataBase Management System)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • University of South Florida (DDSM), University of

Mississippi Medical Center (UMMC).

  • ACR standard BIRADS and Facility Oncology Registry

Data Standards by the Commission on Cancer (FORDS) .

  • Hierarchical architecture, able to combine cancer

registry data with the clinical radiology image and repor8ng categories.

  • Capability for the database system to store and retrieve

data for cancer registra8on, annota8on, sta8s8c analysis, research, teaching, etc.

slide-68
SLIDE 68

DBMS (DataBase Management System)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Top level: general informa8on collected from the cancer registry (informa8on on pa8ent, cancer iden8fica8on, co morbidi8es, staging and treatment, etc.). Middle level: the informa8on on breast cancer, repor8ng, pathology and scanning. At this level the informa8on becomes more specific and narrowed down to each case of breast cancer, Lowest level: en88es related to image and pathology of three lesion types, architecture distor8on, calcifica8on and mass. Each case has four scanned images. The informa8on held at this level is specific to each image, such as laterality, view, shape, distribu8on, boundary of annota8on, number of abnormali8es, etc.

slide-69
SLIDE 69

DBMS (DataBase Management System)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-70
SLIDE 70

DBMS (DataBase Management System)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-71
SLIDE 71

DBMS (DataBase Management System)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-72
SLIDE 72

DBMS (DataBase Management System)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-73
SLIDE 73

DBMS (DataBase Management System)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

slide-74
SLIDE 74

Conclusions (“personal”)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • 2018: need of a dataset of FFDM (Full Field Digital

Mammography) normal images to develop a deep-learning based system for automa8c assessment of breast density … not found!

  • Datasets available for analysis SW development and valida8on:

s8ll consis8ng mainly of digi8zed films.

  • Datasets used for sta8s8cal and epidemiological studies: non

publicly available.

  • 2D sysnthe8c images from DBT (Digital Breast Tomosynthesis)

to be considered.

slide-75
SLIDE 75

Conclusions (“personal”)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Realiza8on of a new database of clinical 2D FFDM images

acquired at AOUP (Azienda Ospedaliera Universitaria Pisana). Work in progress, 1962 exams (4 views/each case).

  • Realiza8on of a new database of clinical DBT images acquired

at AOUP (Azienda Ospedaliera Universitaria Pisana). Work in progress, 50 cases.

  • Project of a new mul8centric database of screening FFDM

(with clinical and family history informa8ons) in collabora8on with ATNO (Azienda USL Toscana Nord Ovest).

slide-76
SLIDE 76

Conclusions (“personal”)

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

AOUP FFDM exams:

slide-77
SLIDE 77

Perspec<ves

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Applica8on of Standard Mammographic Form to FFDM.
  • Realiza8on of synthe8c images from clinical data (breast CT,

breast MRI) or by hybrid approaches and 3D anatomic models.

  • Histopathological images and images acquired with other

diagnos8c modali8es (DBT, breast CT, PET/CT, breast MRI.

slide-78
SLIDE 78

Perspec<ves

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

The SMF tool is a fully automated, objec8ve measurement tool to es8mate the volume of glandular 8ssue in the breast from a mammogram which also compensates for varia8ons in X-ray imaging techniques and is able to standardize image scanned

  • n different film digi8zers. It

explicitly considers breast compression, exposure and tube voltage and incorporates a full physics model.

slide-79
SLIDE 79

Perspec<ves

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Produc8on of synthe8c mammograms

slide-80
SLIDE 80

Perspec<ves

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

With quan8ta8ve analysis of clinical images (CT) With quan8ta8ve analysis of clinical images (MR)

slide-81
SLIDE 81

Perspec<ves

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

3D anatomic breast model for genera8on of synthe8c mammograms and tomograms

slide-82
SLIDE 82

Perspec<ves

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

CT anatomic models including pathological strutures of regular shapes Inclusion of spiculated tumour masses

slide-83
SLIDE 83

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Perspec<ves

Other breast cancer databases

slide-84
SLIDE 84

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Perspec<ves

hjp://www.cancerimagingarchive.net hjps://wiki.nci.nih.gov/#all-updates Other breast cancer databases

slide-85
SLIDE 85

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Perspec<ves

hjps://wiki.nci.nih.gov/display/cip/rider

slide-86
SLIDE 86

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Perspec<ves

hjps://wiki.nci.nih.gov/display/cip/rider

slide-87
SLIDE 87

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Perspec<ves

hjps:// wiki.cancerimagingarchive.net/ display/Public/ISPY1

slide-88
SLIDE 88

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Perspec<ves

hjps://wiki.cancerimagingarchive.net/display/ Public/ACRIN-FLT-Breast

slide-89
SLIDE 89

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Perspec<ves

hjps://wiki.cancerimagingarchive.net/display/ Public/QIN-Breast

slide-90
SLIDE 90

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Perspec<ves

hjps://wiki.cancerimagingarchive.net/display/ Public/QIN-Breast

slide-91
SLIDE 91

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

Perspec<ves

hjps://wiki.cancerimagingarchive.net/display/ Public/TCGA-BRCA

slide-92
SLIDE 92

Conclusions

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018

  • Real esigence of reliable databases of (not only) mammographic

images.

  • Technological progress in acquisi8on instruments.
  • Increasing availability of storage and compu8ng power.
  • Characteris8cs and quan88es requested by (non only) the new

analysis tecnniques.

  • A lot of efforts in various research fields related to this topic.
slide-93
SLIDE 93

Thank you for your ajen8on

maria.evelina.fantacci@unipi.it

3° Training School on “Applica8on of computer models for advancement of X-ray breast imaging techniques” Napoli, IT, 17-19 September 2018