We created a small dataset to validate the lifetime denoising using the trained Noise2Noise-NBN model - the Fluorescence Microscopy Lifetime Denoising (FMLD) dataset - that is dedicated to perform the lifetime denoising from the noisy (raw) lifetime images. The dataset consists of 50 real fluorescence microscopy images obtained with custom-built two-photon microscope and representative fixed BPAE cell (F36924) samples. We use all 50 images for testing/inference. These images are captured with single different excitation wavelength (800 nm) and at a single power level (4 mW). Here, the dataset contains images of Intensity, G and S components of the lifetime processing as well as lifetime image stack (contains 50 images) with in the same filed of view (FOV).