CAMEL: A Weakly Supervised Learning Framework for Histopathology Image Segmentation

Abstract

Histopathology image analysis plays a critical role in cancer diagnosis and treatment. To automatically segment the cancerous regions, fully supervised segmentation algorithms require labor-intensive and time-consuming labeling at the pixel level. In this research, we propose CAMEL, a weakly supervised learning framework for histopathology image segmentation using only image-level labels. Using multiple instance learning (MIL)-based label enrichment, CAMEL splits the image into latticed instances and automatically generates instance-level labels. After label enrichment, the instance-level labels are further assigned to the corresponding pixels, producing the approximate pixellevel labels and making fully supervised training of segmentation models possible. CAMEL achieves comparable performance with the fully supervised approaches in both instance-level classification and pixel-level segmentation on CAMELYON16 and a colorectal adenoma dataset. Moreover, the generality of the automatic labeling methodology may benefit future weakly supervised learning studies for histopathology image analysis.

Characterization of Pulmonary Nodules in Computed Tomography Images Based on Pseudo-Labeling Using Radiology Reports

“…Due to the difficulty of data annotation, semi-supervised learning is a popular approach in medical domain [2], [23], [24], [25], [26]. In this context, pseudo-labeling has been discussed [24], [25].…”

TransMIL: Transformer based Correlated Multiple Instance Learning for Whole Slide Image Classification

“…The application of MIL in WSIs can be divided into two categories. The first one is instance-level algorithms [7,23,24,25,26], where a CNN is first trained by assigning each instance a pseudo-label based on the bag-level label, and then the top-k instances are selected for aggregation. However, this method requires a large number of WSIs, since only a small number of instances within each slide can actually participate in the training.…”

Effective Immunohistochemistry Pathology Microscopy Image Generation Using CycleGAN

“…But the MIL algorithm is used to train a binary classifier to classify two kinds of patches, so the background patches should be removed. In this paper, an OTSU algorithm (Xu G. et al, 2019) is used to extract foreground from HE pathology microscopy images and Ki-67 pathology microscopy images. The second step is extracting large patches with weak annotations.…”

“…With incomplete or lacking annotations of pathology microscopy images, semi-supervised learning-based methods, unsupervised learning-based methods and self-supervised learning-based methods have been introduced to work on datasets with partial annotations or without any annotation and these methods have proven to be useful (Campanella et al, 2019;Xu G. et al, 2019). Among these methods, multiple instances learning (MIL) algorithms have been applied successfully with unannotated pathology microscopy images, so they have been adopted in this paper (Xu G. et al, 2019). Actually, the adversarial training process aims to align some feature vectors extracting from real images or fake images (Nguyen et al, 2017).…”