Label-free metabolic fingerprinting of motile mammalian spermatozoa with subcellular resolution

Acquisition

The fluorescence lifetime images of mouse sperm cells ex vivo were acquired using a Carl Zeiss LSM880 NLO microscope (Carl Zeiss, Germany) coupled to a (\(690-1040\,\textrm{nm} \)) Ti:Saphire laser system (Chameleon Ultra II, Coherent) pulsing at 80 MHz and a TCSPC card (HydraHarp 400, PicoQuant). Sperm samples were kept in an Okolab chamber at \(37^{\circ}\) and \(5\%\,\textrm{CO}_2\). The samples were excited with a laser of \(740\,\textrm{nm}\) wavelength using a \(40\, \times \, 1.1 \) NA water immersion objective. A dichroic mirror (\(690\,\textrm{nm} \)) was used to separate the fluorescence signal from the excitation laser. NAD(P)H fluorescence was collected through a \(460 (\pm20)\,\textrm{nm}\) filter by a hybrid detector (HPM-100-40, Becker & Hickl). An additional filter was used to block near-infrared light.  The FLIM data were recorded using the SymPhoTime 64 software (PicoQuant) using 16 ps TCSPC sampling (length of the time slot). 16 time frames with dimension \(512\times512\) were acquired, with the scanning speed \(5\, \mu \textrm{s/pixel}\) (giving full acquisition time of 1 minute and 23 seconds).

Export

Custom software, called TTTR Data Analysis, was used to open the files and adjust them for export. Each file is a composition of 16 time frames of the same scene. The sperm were not immobilised, each scene was acquired multiple times sequentially, thus, each time frame may include the sperm in a different position. Therefore, some of the sperm may be blurry. For this dataset we provide only one time frame of the field from the selected data, trying to pick the time frame with the least movement of the sperm. 

Manual Annotation

Masks for the training, validation and test dataset were manually prepared in the binary editor of NIS-Elements by Fitore Kusari. Two binary masks are provided for each image, for the heads and the mitochondrial midpiece of the sperm.

Final dataset size is summarized in:

Images
Midpiece Objects
Head Objects

Train
235
17952
8620

Validation
16
1448
678

Test
30
2507
1167

Sum
281
21907
10465