Agenda

Market Trends

Automated Cancer Diagnosis

Future Plans

Our Team

Computer-Aided Diagonosis in Medical Imaging

Market Trends

100 million eVisits in 2014
Global health care cloud computing market expected to grow from $1.82 billion in 2011 to $6.79 billion by 2018
Computer - aided detection market expected to hit $1.47 billion by 2020
Advanced medical imaging within reach of everyone
Patient-to-patient social media collaboration
Whole slide digital pathology imaging

Market Trends

Out of 600 million annual GP visits about half could also be solved by eVisit

100 million eVisits in 2014
Global health care cloud computing market expected to grow from $1.82 billion in 2011 to $6.79 billion by 2018
Computer - aided detection market expected to hit $1.47 billion by 2020
Advanced medical imaging within reach of everyone
Patient-to-patient social media collaboration
Whole slide digital pathology imaging

Market Trends

The University of Pittsburgh Medical Center decreased study management costs by 43% or $14 million

100 million eVisits in 2014
Global health care cloud computing market expected to grow from $1.82 billion in 2011 to $6.79 billion by 2018
Computer - aided detection market expected to hit $1.47 billion by 2020
Advanced medical imaging within reach of everyone
Patient-to-patient social media collaboration
Whole slide digital pathology imaging

Market Trends

Digital mammography is already better than film mammography in detecting early stage cancer

100 million eVisits in 2014
Global health care cloud computing market expected to grow from $1.82 billion in 2011 to $6.79 billion by 2018
Computer - aided detection market expected to hit $1.47 billion by 2020
Advanced medical imaging within reach of everyone
Patient-to-patient social media collaboration
Whole slide digital pathology imaging

Market Trends

100 million eVisits in 2014
Global health care cloud computing market expected to grow from $1.82 billion in 2011 to $6.79 billion by 2018
Computer - aided detection market expected to hit $1.47 billion by 2020
Advanced medical imaging within reach of everyone
Patient-to-patient social media collaboration
Whole slide digital pathology imaging

Since the launch of Pocket Pathologist in 2013 high resolution digital pathology images are available with just a smartphone

Market Trends

100 million eVisits in 2014
Global health care cloud computing market expected to grow from $1.82 billion in 2011 to $6.79 billion by 2018
Computer - aided detection market expected to hit $1.47 billion by 2020
Advanced medical imaging within reach of everyone
Patient-to-patient social media collaboration
Whole slide digital pathology imaging

Collaboration between patients and experts will be a leading trend in the future*

* Pathology Informatics Summit 2015: Shaping the Future of Clinical Informatics Michael Riben, MD

Market Trends

Whole slide image sharing and analysis is one of the crucial challenges of today

100 million eVisits in 2014
Global health care cloud computing market expected to grow from $1.82 billion in 2011 to $6.79 billion by 2018
Computer - aided detection market expected to hit $1.47 billion by 2020
Advanced medical imaging within reach of everyone
Patient-to-patient social media collaboration
Whole slide digital pathology imaging

Automated Cancer Diagnosis

Miccai Digital Pathology Challenge 2015

Classification Algorithm

Oligodendroglioma 87%

Astrocytoma 13%

Diagnosis

Automated Cancer Diagnosis

Classification Algorithm

Divide in Frames

Frame Segmentation

Classify each frame

Classify the sample

Extract frames of interest

Classify nuclei in a frame

Whole slide image is divided in small frames

Automated Cancer Diagnosis

Segmentation algorithm is used to extract nuclei

Divide in Frames

Frame Segmentation

Classify each frame

Classify the sample

Extract frames of interest

Classify nuclei in a frame

Classification Algorithm

Automated Cancer Diagnosis

Frames are ranked based on the number of nuclei and the most dense frames are found

Divide in Frames

Frame Segmentation

Classify each frame

Classify the sample

Extract frames of interest

Classify nuclei in a frame

Classification Algorithm

Automated Cancer Diagnosis

Each nucleus is classified with certain probability

Oligodendrocyte

Astrocyte

Divide in Frames

Frame Segmentation

Classify each frame

Classify the sample

Extract frames of interest

Classify nuclei in a frame

Classification Algorithm

Automated Cancer Diagnosis

Based on the percentage of nuclei of certain type each frame is classified

Oligodendrocyte

Astrocyte

34%

66%

Astrocytoma

Divide in Frames

Frame Segmentation

Classify each frame

Classify the sample

Extract frames of interest

Classify nuclei in a frame

Classification Algorithm

Automated Cancer Diagnosis

Divide in Frames

Frame Segmentation

Classify each frame

Classify the sample

Extract frames of interest

Classify nuclei in a frame

Astrocytoma Frames > T

Astrocytoma

If the amount of frames classified as certain cancer type is over specified threshold the whole sample is classified as that type

Classification Algorithm

Classification on MRI images is a viable alternative for some/many tumors
Preliminary results of our MRI classification algorithm are very promising: 88% though more data needed
Incorporation of MRI can improve results