NuLite - lightweight and fast model for nuclei instance segmentation and classification

In pathology, accurate and efficient analysis of Hematoxylin and Eosin (H&E) slides is crucial for timely and effective cancer diagnosis. For these reasons, nuclei instance segmentation and classification tools are helpful, allowing pathologists to detect and identify regions of interest and perform quantitative analysis. Although many deep-learning solutions for this task exist in the literature, they often entail high computational costs and resource requirements, thus limiting their practical usage in medical applications. To address this issue, we introduce NuLite, an architecture designed explicitly to be lightweight and fast. We obtained three versions of our model, NuLite-S, NuLite-M, and NuLite-H, trained on the PanNuke dataset. The experimental results prove that our models are equivalent to CellViT (SOTA) in terms of panoptic quality and F-score. However, our lightest model, NuLite-T, is about 58 times smaller in terms of parameters and about 10 times smaller in terms of GFlops. In comparison, our heaviest model is about 15 times smaller in terms of parameters and about 7 times smaller in terms of GFlops. Moreover, considering the GPU latency, our model is up to about 13 times faster than CellViT. Lastly, to prove the effectiveness of our solution, we provide a robust comparison of external datasets, namely CoNseP, MoNuSeg, and GlySAC. Our model is publicly available at https://github.com/CosmoIknosLab/NuLite.