Semi-Supervised Segmentation Framework for Gastrointestinal Lesion Diagnosis in Endoscopic Images

Background: Accurate gastrointestinal (GI) lesion segmentation is crucial in diagnosing digestive tract diseases. An automatic lesion segmentation in endoscopic images is vital to relieving physicians' burden and improving the survival rate of patients. However, pixel-wise annotations are highly intensive, especially in clinical settings, while numerous unlabeled image datasets could be available, although the significant results of deep learning approaches in several tasks heavily depend on large labeled datasets. Limited labeled data also hinder trained models' generalizability under fully supervised learning for computer-aided diagnosis (CAD) systems. Methods: This work proposes a generative adversarial learning-based semi-supervised segmentation framework for GI lesion diagnosis in endoscopic images to tackle the challenge of limited annotations. The proposed approach leverages limited annotated and large unlabeled datasets in the training networks. We extensively tested the proposed method on 4880 endoscopic images. Results: Compared with current related works, the proposed method validates better results (Dice similarity coefficient = 89.42 +/- 3.92, Intersection over union = 80.04 +/- 5.75, Precision = 91.72 +/- 4.05, Recall = 90.11 +/- 5.64, and Hausdorff distance = 23.28 +/- 14.36) on the challenging multi-sited datasets, confirming the effectiveness of the proposed framework. Conclusion: We explore a semi-supervised lesion segmentation method to employ the full use of multiple unlabeled endoscopic images to improve lesion segmentation accuracy. Experimental results confirmed the potential of our method and outperformed promising results compared with the current related works. The proposed CAD system can minimize diagnostic errors.

Automated summary

This paper proposes a generative adversarial learning-based semi-supervised segmentation framework for GI lesion diagnosis in endoscopic images to tackle the challenge of limited annotations. The proposed method is extensively tested on 4880 endoscopic images and achieves better results compared to current related works, confirming its effectiveness.