Abstract:Improving the segmentation accuracy of cells from microscopy images directly benefits medical practitioners for differential cell counting, disease diagnosis, and treatment. However, separating each cell continues to be challenging, especially when the cells appear in clumps. Recent advancements in deep learning have demonstrated success in automating cell segmentation and counting, but they necessitate large volumes of annotated data for training. Creating precise ground truth for cell segmentation is laborious and demands expertise due to the intricate and varied characteristics of microscopy images. Consequently, this complexity poses a challenge to the practicality of implementing deep neural network-based solutions for microscopy cell segmentation. This paper addresses a significant research gap by exploring the underutilized traditional image features to build a lightweight deep-learning framework for cell boundary segmentation that can be effectively trained with a limited amount of labeled data. We have developed Saliency Guided Attention Module Network (SGAM-net), which uses perceptual features inherent along the cell boundaries and poses cell segmentation as a thick boundary prediction problem with the aim to separate the cells. SGAM-net uses a specialized attention mechanism to combine the perceptual feature maps with the learned features. When tested on the three public bench-mark microscopy datasets, with respect to the base Unet model, there is an increase of 1.36%, 1.29%, and 6.35% in Dice score and 2.39%, 2.31% and 6.99% in Intersection Over Union (IOU) score. The model requires only 23.07% less parameters than Unet++ model. The potential of SGAM-net in segmenting boundaries semantically has been demonstrated, and it is further processed to compute cell counts. The model outperformed the state-of-the-art instance separation method 'Stardist' in cell counting by giving a reduction ratio of 2.34, 1.89, and 2.03 on the metric 'mean absolute error'.

An Enhancement Neighborhood connected Segmentation for 2D-Cellular Image

An Automatic Seeded Region Growing for 2D Biomedical Image Segmentation

Interactive Color Image Segmentation by Region Growing Combined with Image Enhancement Based on Bezier Model

Edge Segmentation: Empowering Mobile Telemedicine with Compressed Cellular Neural Networks

Robust Segmentation of Cell Nuclei in 3-D Microscopy Images

Robust cell segmentation based on gradient detection, Gabor filtering and morphological erosion

Learn to segment single cells with deep distance estimator and deep cell detector

Attention-Guided Residual U-Net with SE Connection and ASPP for Watershed-Based Cell Segmentation in Microscopy Images

Saliency and boundary guided segmentation framework for cell counting in microscopy images

An Efficient Data Analytics and Optimized Algorithm for Enhancing the Performance of Image Segmentation Using Deep Learning Model

Accurate Cell Segmentation in Digital Pathology Images via Attention Enforced Networks

Adaptive strategy for superpixel-based region-growing image segmentation

Automatic enhancement preprocessing for segmentation of low quality cell images

A Novel Approach to Optimizing Convolutional Neural Networks for Improved Digital Image Segmentation

A Novel Automatic Seed Placement Approach for Region Growing segmentation in Mammograms

A hybrid cell image segmentation method based on the multilevel improvement of data

Segmentation with Residual Attention U-Net and an Edge-Enhancement Approach Preserves Cell Shape Features

A New Image Segmentation Algorithm: A Community Detection Approach

Semantic Segmentation Refinement by Monte Carlo Region Growing of High Confidence Detections

CycleSegNet: Object Co-segmentation with Cycle Refinement and Region Correspondence

Advanced Multi-Microscopic Views Cell Semi-supervised Segmentation