Speckle image stacks acquired on human skin with a high-speed camera

These datasets all consist of stacks of images of various body parts of of the same 50-year-old Caucasian male subject.

For each acquisition, the skin was first illuminated with a near-infrared LASER (wavelength 785 nm). Then, the skin was filmed with a Phantom VEO 710L camera, with an acquisition rate of 2000 to 30,000 Hz.

Datasets WRIST_1, WRIST_2 and WRIST_3 were acquired by imaging the anterior surface of the left wrist, with respective acquisition frequencies of 3000, 2000 and 10,000 Hz.

Dataset FINGERS consists of images of the end of two fingers of the left hand. It was acquired at 30,000 Hz.

Dataset EAR consists of images of the right ear. It was acquired at 2000 Hz.

Dataset PALM consists of images of the palm of the left hand, which was gently scraped beforehand to trigger a superficial inflammation and draw a smiley face. It was then filmed with an acquisition frequency of 10,000 Hz.

Each dataset is stored in a HDF5 file. The data array is stored as data at the root of the tree structure. Attributes T_exp and f_acq provide respectively the exposure time (in microseconds) and the acquisition rate (in Hertz) of the dataset. The script test.py shows how data can be accessed through Python and the library h5py. It can be used as follows:

python test.py filename.h5

All these datasets are used and referenced in our companion article "Dynamic speckle imaging of human skin vasculature with a high-speed camera" (to be published).