Automatic Assessment of Full LV Coverage in Cardiac Cine MRI Using Learned Intensity Attributes with Discriminative 3D CNNs
datasetposted on 03.11.2017 by Le Zhang, Ali Gooya, Bo Dong, Marco Pereañez, Stefan K. Piechnik, Stefan Neubauer, Steffen E. Petersen, Alejandro F Frangi
Datasets usually provide raw data for analysis. This raw data often comes in spreadsheet form, but can be any collection of data, on which analysis can be performed.
This dataset has been created to serve as complimentary material to a journal publication currently under review by IEEE Transactions on Medical Imaging.
The dataset contains two files, a comma-separated file (.csv) and a video (.avi). The first file contains the results of applying our Discriminative 3D CNN (D3D CNN) Image Quality Assessment (IQA) algorithm to a set of images from the UK Biobank dataset. The Image Quality Assessment file named "IQA spreadsheet.csv" contains 3 columns. The first column labeled "f.eid" lists 5000+ patient IDs, the next column labeled "MBS" (Missing Basal Slice) contains binary values indicating the presence or absence of the basal slice. Similarly the last column labeled "MAS" (Missing Apical Slice) indicates the presence or absence of the apical slice in the image volume.
The second file is a demo of the algorithm's functionality. The video shows a unique patient's cardiac image volume, and the result of the detection/non-detection of the basal and apical slices indicated with green and red labels on the images and on a side panel.