NuLite -- Lightweight and Fast Model for Nuclei Instance Segmentation and Classification

TL;DR

NuLite is a lightweight, fast, and accurate CNN-based model for nuclei segmentation and classification in pathology slides, significantly reducing computational costs while maintaining state-of-the-art performance.

Contribution

We introduce NuLite, a novel lightweight U-Net-like architecture based on Fast-ViT, achieving comparable accuracy to SOTA models with much fewer parameters and faster inference.

Findings

NuLite-S is 40x smaller in parameters than CellViT.

NuLite models are up to 8x faster than CellViT.

NuLite maintains SOTA quality on multiple datasets.

Abstract

In pathology, accurate and efficient analysis of Hematoxylin and Eosin (H\&E) slides is crucial for timely and effective cancer diagnosis. Although many deep learning solutions for nuclei instance segmentation and classification 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 a novel convolutional neural network, NuLite, a U-Net-like architecture designed explicitly on Fast-ViT, a state-of-the-art (SOTA) lightweight CNN. 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 equal CellViT (SOTA) in terms of panoptic quality and detection. However, our lightest model, NuLite-S, is 40 times smaller in terms of parameters and about 8 times smaller in terms of GFlops, while our heaviest model is 17 times smaller in terms of parameters and about 7 times smaller in terms of GFlops. Moreover, our model is up to about 8 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