Collaborative Learning of Semi-Supervised Segmentation and Classification for Medical Images

TLDR

Medical image analysis typically separates disease grading, which uses image‑level labels, from fine‑grained lesion segmentation that requires costly pixel‑level annotations by experts. We propose a collaborative learning framework that jointly improves disease grading and lesion segmentation through semi‑supervised learning with an attention mechanism. Starting from a small set of pixel‑annotated images, a multi‑lesion segmentation model generates initial maps that are used to train a lesion‑attentive grading network, while the attention module refines the segmentation maps in a semi‑supervised loop and an adversarial architecture further enhances training. Experiments on diabetic retinopathy datasets demonstrate that our method consistently outperforms state‑of‑the‑art approaches across three public benchmarks.

Abstract

Medical image analysis has two important research areas: disease grading and fine-grained lesion segmentation. Although the former problem often relies on the latter, the two are usually studied separately. Disease severity grading can be treated as a classification problem, which only requires image-level annotations, while the lesion segmentation requires stronger pixel-level annotations. However, pixel-wise data annotation for medical images is highly time-consuming and requires domain experts. In this paper, we propose a collaborative learning method to jointly improve the performance of disease grading and lesion segmentation by semi-supervised learning with an attention mechanism. Given a small set of pixel-level annotated data, a multi-lesion mask generation model first performs the traditional semantic segmentation task. Then, based on initially predicted lesion maps for large quantities of image-level annotated data, a lesion attentive disease grading model is designed to improve the severity classification accuracy. Meanwhile, the lesion attention model can refine the lesion maps using class-specific information to fine-tune the segmentation model in a semi-supervised manner. An adversarial architecture is also integrated for training. With extensive experiments on a representative medical problem called diabetic retinopathy (DR), we validate the effectiveness of our method and achieve consistent improvements over state-of-the-art methods on three public datasets.