Submissions to SPIE Medical Imaging 2014

Two abstracts have been submitted to SPIE Medical Imaging 2014. The Conferences will take place in San Diego, USA the 15-20 February 2014.

The abstracts are related to Multi-organ Probability Maps in whole body MR images and Pulmonary Embolism detection in 4D Dual Energy CT.

Registration open for Visceral Benchmark 1

Registration to participate in VISCERAL Benchmark 1 has opened. Links to an Specfication document and Registration website are available at Benchmark 1 page

35 volumes from 5 modalities are in the VISCERAL training set. In these volumes, 478 manual segmentations of individual organs are available. All 35 volumes also have a number of landmarks manually placed. The following organs are annotated: Kidneys, spleen, liver, lungs, urinary bladder, rectus abdominis muscle, lumbar vertebra #1, thyroid gland, pancreas, psoas major muscles, gall bladder, sternum, aorta, trachea, and adrenal glands.

 

Papers accepted at MICCAI 2013

We have one paper accepted in the main MICCAI 2013 conference, and one paper accepted in MICCAI-MCV 2013 workshop (Medical Computer Vision: Large Data in Medical Imaging):

Epileptogenic Lesion Quantication in MRI Using Contralateral 3D Texture Comparisons, Oscar Alfonso Jiménez del Toro, Antonio Foncubierta-Rodriguez, María Isabel Vargas Gómez, Henning Müller, Adrien Depeursinge, MICCAI 2013, Nagoya, Japan.

 

2D-Based 3D Volume Retrieval Using Singular Value Decomposition of Detected Regions, Alba G. Seco de Herrera, Antonio Foncubierta-Rodríguez, Emanuele Schiavi, Henning Müller, MICCAI-MCV 2013, Nagoya, Japan.

Visceral teaser dataset available

The VISCERAL benchmark 1 teaser dataset has been released. It includes 4 volumes, annotations, landmarks, and Matlab tutorial functions that allow participants to easily explore the data, and understand the coordinate and label conventions.

This data can be obtained by e-mail request from georg.langs[at]meduniwien.ac.at. This data can only be made available on condition that is not used in any publications and not distributed further. Please agree to these conditions in the request e-mail.

The full dataset will be available from the 1st of August, when VISCERAL Benchmark 1 starts. The full dataset will be subject to a detailed usage agreement, but will not be subject to the restriction on publication. 

Khresmoi project will present their radiology prototypes at the 2013 RSNA conference

The paper ‘A System for Visual Information Retrieval in Images, Radiology Reports and the Medical Literature’ by Dimitrios Markonis et. al. has been accepted for the 2013 RSNA conference in Chicago, IL, USA. With about 60’000 visitors is the biggest conference in the world in radiology.

Khresmoi will have a booth during the entire conference presenting the prototypes and discussing these prototypes with interested radiologists and the companies present at the end. 

 

Khresmoi professional beta prototype released

A beta version of the Khresmoi Professional prototype is now available. This prototype is aimed at medical professionals. It provides tools such as faceted search, multilingual search, a Personal Library for storing articles and query history. The link to access the prototype is: http://khresmoi.is.inf.uni-due.de:8182

tutorial outlining the capabilities of the system is available.

More technical information on this prototype can be found in Deliverable 8.3.

This system is still under development, and this prototype is a snapshot of its current state. This prototype is intended for evaluation use only and should not be used to guide treatment. Comments are welcome.

WIDTH first workshop: Warehousing Images in the Digital Hospital scheduled for July 15-16

The Newsletter from the WIDTH project May 2013 is now available online . It contains information about the upcoming first workshop: Warehousing Images in the Digital Hospital. It will take place next 15-16 July in Athens, Greece. 

The goal of this workshop is to bring together researchers in medical imaging, medical image retrieval, data mining, machine learning and Artificial Intelligence communities to discuss new techniques in multimodal mining and retrieval and their potential employment in supporting clinical decision makings.

Crowdsourcing approach with medical imaging highlighted in Crowdflower blog

The experiments for the ‘Ground truth generation in medical imaging: A crowd sourcing iterative based approach’ paper were highlighted in the CrowdFlower blog. The research was performed by Antonio Foncubierta and Henning Müller.
 
In this research crowdsourcing is used as a way to obtain quickly basic ground truth for radiological medical images. The paper can be found at our website publications link

Keynote presentation at Eurographics 3D Object Retrieval Workshop 2013

Prof. Müller was keynote speaker at the 3D Object Retrieval Workshop Series. The topic of the keynote talk was ‘Medical 3D data retrival’. 

Abstract:

Medical tomographic data is produced in enormous quantities and currently very little of the knowledge stored in past cases has been exploited as data is only used in the context of single patients. In difference to object retrieval, tomographic data represents solid  texture and the texture variation inside objects is usually more important than shapes of structures. The presentation will describe methods for retrieval of solid medical 3D texture in several applications and ways to better exploit existing medical data.

The VISCERAL challenge in medical image segmentation and retrieval will be presented that has the goal to bring medical 3D data analysis towards big data. Several terabytes of data will be made available to the scientific community to first identify important body parts and then retrieve similar cases for challenges that will be run inside the cloud.

Paper on texture analysis in dual energy CT accepted for EMBC 2013

The paper ‘Benefits of texture analysis of dual energy CT for computer-aided pulmonary embolism detection’ by Antonio Foncubierta-Rodríguez, Oscar Alfonso Jimenez del Toro, Alexandra Platon, Pierre-Alexandre Poletti, Henning Müller and Adrien Depeursinge has been accepted for publication. 

It will be presented in the 35th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC’13) to be held in Osaka International Convention Center, in Osaka, Japan on July 3-7, 2013.

 

'Medical 3D data retrieval' keynote talk at Eurographics2013

Prof. Henning Müller has been invited as a speaker at the Eurographics Workshop on 3D Object Retrieval. This event will take place in Girona (Spain) May 11-2013. 

The Keynote talk is called “Medical 3D data retrieval”:

Medical tomographic data is produced in enormous quantities and currently very little of the knowledge stored in past cases has been exploited as data is only used in the context of single patients. In difference to object retrieval, tomographic data represents solid texture and the texture variation inside objects is usually more important than shapes of structures. The presentation will describe methods for retrieval of solid medical 3D texture in several applications and ways to better exploit existing medical data. 

The VISCERAL challenge in medical images segmentation and retrieval will be presented that has the goal to bring medical 3Ddata analysis towards big data. Several terabytes of data will be made available to the scientific community to first identify important body parts and then retrieve similar cases for challenges that will be run inside the cloud.