Advancing Cell Detection in Anterior Segment Optical Coherence Tomography Images

TL;DR

This work tackles the challenge of detecting inflammatory cells in AS-OCT images for anterior uveitis, where manual analysis is slow and thresholding-based methods show incomplete detection. It introduces ACCDor, a two-stage framework with a zero-shot chamber segmentation module (CSM) that uses the Segment Anything Model (SAM) via heuristic prompts, and a cell detection module (CDM) that localizes cells within the AC using an adjusted thresholding strategy and a cell classifier. The approach yields strong AC segmentation performance ($IoU=96.41\%$, $Dice=98.21\%$) and superior cell detection results (image-level $Precision=87.82\%$, $Recall=90.29\%$, $F1=89.02\%$; bounding-box level $Precision=86.88\%$, $Recall=87.23\%$, $F1=87.02\%$), outperforming thresholding and state-of-the-art detectors, and revealing that prior thresholds may undercount cells. By enabling more reliable, location-based cell detection, ACCDor has clear clinical implications for monitoring anterior uveitis and guiding treatment, with code publicly available for broader adoption.

Abstract

Anterior uveitis, a common form of eye inflammation, can lead to permanent vision loss if not promptly diagnosed. Monitoring this condition involves quantifying inflammatory cells in the anterior chamber (AC) of the eye, which can be captured using Anterior Segment Optical Coherence Tomography (AS-OCT). However, manually identifying cells in AS-OCT images is time-consuming and subjective. Moreover, existing automated approaches may have limitations in both the effectiveness of detecting cells and the reliability of their detection results. To address these challenges, we propose an automated framework to detect cells in the AS-OCT images. This framework consists of a zero-shot chamber segmentation module and a cell detection module. The first module segments the AC area in the image without requiring human-annotated training data. Subsequently, the second module identifies individual cells within the segmented AC region. Through experiments, our framework demonstrates superior performance compared to current state-of-the-art methods for both AC segmentation and cell detection tasks. Notably, we find that previous cell detection approaches could suffer from low recall, potentially overlooking a significant number of cells. In contrast, our framework offers an improved solution, which could benefit the diagnosis and study of anterior uveitis. Our code for cell detection is publicly available at: https://github.com/joeybyc/cell_detection.

Figures (5)

• Figure 1: The architecture of CSM.
• Figure 2: Illustration of the CSM for AC segmentation.
• Figure 3: The architecture of CDM.
• Figure 4: Visualization of the cell identification results.
• Figure 5: Trends of evaluation metrics with the change of $\alpha$ in validation set.