Detail of the student project

List
Topic:Automatická katalogizace MRI dat
Department:Katedra kybernetiky
Supervisor:MUDr. Martin Černý
Announce as:Diplomová práce, Bakalářská práce, Semestrální projekt
Description:Clinical practice produces high amounts of imaging data, however it's utility for big data research projects is limited by it's poorly labeled nature, unconsistent naming conventions and data often comming from various institutions with different imaging protocols. To unlock the richness of the data collected over the years, we need a reliable automatic catalogization method based on imaging metadata. The aim of this work is to automatically identify main clinical imaging sequences based on DICOM tags. The student will be provided with a large database of anonymized metadata from over 11 thousand MR brain scans and asked to classify them as one of the main clinical sequences. For example, the three most often used sequences T1, T2 and FLAIR can be told apart by looking at their TR and TE parameters (0018|0080 and 0018|0081 DICOM tags). A set of classification rules will be determined together with a clinical consultant to account for the main clinically relevant imaging parameters (administration of intravenous contrast agent, diffusion-weighted image, what kind of diffusion weighted image, whether a reconstruction algorithm was used or not, whether a sequence is rather a set of conseqent 2D slices or a true 3D volume). Any programming language can be used, the final output will be a command-line program accepting a DICOM image, extracting it's metadata and printing a classification result.

For the purpose of a diploma or bachelor thesis, the topic can by extended for example by performing an image based classification.

This project will by supervised by dr. Martin Černý (cerny.martin@uvn.cz), please contact him for more details.
Bibliography:Gauriau et al: Using DICOM Metadata for Radiological Image Series Categorization: a Feasibility Study on Large Clinical Brain MRI Datasets. Journal of Digital Imaging. 2020

List of DICOM tags: https://www.dicomlibrary.com/dicom/dicom-tags/
Responsible person: Petr Pošík