# UltraDuke v1 dataset

The UltraDuke v1 dataset is an ultrasound dataset collected at Duke 
University with a Verasonics c52v probe. It contains delay-and-sum 
(DAS) beamformed data as well as data post-processed with Siemens 
Dynamic TCE for speckle reduction, contrast enhancement and 
improvement in conspicuity of anatomical structures. Images are
of phantoms as well as invivo images from Dr. Mark Palmeri

The `master_public.csv` file contains a list of .mat filenames which 
have a DAS image along with the post-processed image. These
filenames are unique across `verasonics_mark.tar` and 
`verasonics_phantom.tar`.

Other `.csv` files are training, testing splits used in the paper
[MimickNet, Matching Clinical Post-Processing Under Realistic 
Black-Box Constraints](https://arxiv.org/abs/1908.05782).

All `.csv` files have scan parameter metadata associated.
The `.csv` contains individual ultrasound frames of targets.
The `timestamp_id` column denotes one acquisition, so grouping by
`timestamp_id` would give you all the frames in an acquisition.

The column `harm` indicates whether or not the images was 
using harmonic imaging. The `initial_radius`, `final_radius`,
`initial_angle`, `finial_angle`, are parameters used to
scan convert the axial and lateral samples. `span_deg` can be
derived from the `initial_angle` and `final_angle` by converting
from radians to degrees.

`v` is the voltage used, `focus_cm` is the ultrasound transducer
transmit focus in centimeters. `f0` is the transducer center 
frequency in Hz. `f0_mhz` is the rounded center frequency in MHz.

The `scanner` and `probe` in v1 is only the [verasonics vantage
C5-2v probe](https://verasonics.com/verasonics-transducers/).

You can open the tar files in [Linux/Unix](https://www.pendrivelinux.com/how-to-open-a-tar-file-in-unix-or-linux/) or on in [Windows](https://wiki.haskell.org/How_to_unpack_a_tar_file_in_Windows). 

Please note that the timestamp_id cells will need to be changed 
to number format (no decimals) if you open in Excel otherwise they will not display
properly.