Unable to run the existing deepImageJ trained bundled models for my dataset

My data mainly contains phase contrast microscopy images of ovarian cancer spheroids (I have attached one of the images below) and ImageJ itself failed to detect the individual spheroids using the analyse particles feature. I then tried to use some of the preexisting DeepImageJ models (U-Net Glyoblastoma segmentation and U-Net HeLa segmentation models) to analyse my datasets, but they kept throwing an error and I got a null output.
I realised that the input size of my images don’t match the predetermined input size. And the brightness and intensity of my images also vary from image to image. Does this prevent the model from giving the required outcome? What should I be modifying to use these models? Or how should I train the preexisting models for analysing my image? OVCAR3_24well_Day1_19_03_2021_17.bmp (14.4 MB)