A Strictly Unsupervised Deep Learning Method for HEp-2 Cell Image Classification

Abstract

Classifying the images that portray the Human Epithelial cells of type 2 (HEp-2) represents one of the most important steps in the diagnosis procedure of autoimmune diseases. Performing this classification manually represents an extremely complicated task due to the heterogeneity of these cellular images. Hence, an automated classification scheme appears to be necessary. However, the majority of the available methods prefer to utilize the supervised learning approach for this problem. The need for thousands of images labelled manually can represent a difficulty with this approach. The first contribution of this work is to demonstrate that classifying HEp-2 cell images can also be done using the unsupervised learning paradigm. Unlike the majority of the existing methods, we propose here a deep learning scheme that performs both the feature extraction and the cells’ discrimination through an end-to-end unsupervised paradigm. We propose the use of a deep convolutional autoencoder (DCAE) that performs feature extraction via an encoding–decoding scheme. At the same time, we embed in the network a clustering layer whose purpose is to automatically discriminate, during the feature learning process, the latent representations produced by the DCAE. Furthermore, we investigate how the quality of the network’s reconstruction can affect the quality of the produced representations. We have investigated the effectiveness of our method on some benchmark datasets and we demonstrate here that the unsupervised learning, when done properly, performs at the same level as the actual supervised learning-based state-of-the-art methods in terms of accuracy.

CCRL: Contrastive Cell Representation Learning

“…We also compared the results of our model with two state-of-the-art unsupervised cell clustering methods. The DCAE [31] model adopted a deep convolution auto-encoder model alongside a clustering layer to learn cell embeddings by preforming an image reconstruction task. Also, the authors of [20] developed a generative adversarial model for cell clustering by increasing the mutual information between the cell representation and a categorical noise vector.…”

“…A few recent studies have focused on unsupervised cell classification to address this problem. For example, Hue et al [20] take advantage of the InfoGAN [9] design to provide a categorical embedding for images, based on which they can differentiate between cell types and Vununu et al [31] propose using Deep Convolution Auto-encoder (DCAE) which learns feature embeddings by performing image reconstruction and clustering at the same time. However, all of these works are focused on only one tissue type.…”

Artificial Intelligence and Deep Learning for Rheumatologists

“…Broadly, deep learning has been applied in 2 ways to this task by 1) automatically extracting features and classifying cells or 2) automatically extracting features, which are then passed to an alternative model for classification. State-ofthe-art models using either technique have achieved accuracies exceeding 97% (26,27), which favorably compare to human accuracy (73.3%) (28). However, this comparison has been criticized, as the task of classifying a single HEp-2 cell, isolated from the broader context of the specimen, is not representative of how IIF tests for ANAs are performed in real clinical practice (29).…”

Hybrid deep learning model for the risk prediction of cognitive impairment in stroke patients

“…(1) Medical image analysis: The accuracy of DL in medical image analysis is equal to or even better than that of clinical experts [28] and saves time and economic cost. For example, DL is applied to the automatic analysis of inspection images, such as ANA system [29][30][31][32].…”