Medical & Biological Engineering & Computing

A lightweight xAI approach to cervical cancer classification

Abstract Cervical cancer is caused in the vast majority of cases by the human papilloma virus (HPV) through sexual contact and requires a specific molecular-based analysis to be detected. As an HPV vaccine is available, the incidence of cervical cancer is up to ten times higher in areas without adequate healthcare resources. In recent years, liquid cytology has been used to overcome these shortcomings and perform mass screening. In addition, classifiers based on convolutional neural networks can be developed to help pathologists diagnose the disease. However, these systems always require the final verification of a pathologist to make a final diagnosis. For this reason, explainable AI techniques are required to highlight the most significant data to the healthcare professional, as it can be used to determine the confidence in the results and the areas of the image used for classification (allowing the professional to point out the areas he/she thinks are most important and cross-check them against those detected by the system in order to create incremental learning systems). In this work, a 4-phase optimization process is used to obtain a custom deep-learning classifier for distinguishing between 4 severity classes of cervical cancer with liquid-cytology images. The final classifier obtains an accuracy over 97% for 4 classes and 100% for 2 classes with execution times under 1 s (including the final report generation). Compared to previous works, the proposed classifier obtains better accuracy results with a lower computational cost. Graphical abstract

A whole-slide image grading benchmark and tissue classification for cervical cancer precursor lesions with inter-observer variability

The cervical cancer developing from the precancerous lesions caused by the human papillomavirus (HPV) has been one of the preventable cancers with the help of periodic screening. Cervical intraepithelial neoplasia (CIN) and squamous intraepithelial lesion (SIL) are two types of grading conventions widely accepted by pathologists. On the other hand, inter-observer variability is an important issue for final diagnosis. In this paper, a whole-slide image grading benchmark for cervical cancer precursor lesions is created and the "Uterine Cervical Cancer Database" introduced in this article is the first publicly available cervical tissue microscopy image dataset. In addition, a morphological feature representing the angle between the basal membrane (BM) and the major axis of each nucleus in the tissue is proposed. The presence of papillae of the cervical epithelium and overlapping cell problems are also discussed. Besides that, the inter-observer variability is also evaluated by thorough comparisons among decisions of pathologists, as well as the final diagnosis.

Papanicolaou stain unmixing for RGB image using weighted nucleus sparsity and total variation regularization

Abstract The Papanicolaou stain, consisting of five dyes, provides extensive color information essential for cervical cancer cytological screening. The visual observation of these colors is subjective and difficult to characterize. Direct RGB quantification is unreliable because RGB intensities vary with staining and imaging conditions. Stain unmixing offers a promising alternative by quantifying dye amounts. In previous work, multispectral imaging was utilized to estimate the dye amounts of Papanicolaou stain. However, its application to RGB images presents a challenge since the number of dyes exceeds the three RGB channels. This paper proposes a novel training-free Papanicolaou stain unmixing method for RGB images. This model enforces (i) nonnegativity, (ii) weighted nucleus sparsity for hematoxylin, and (iii) total variation smoothness, resulting in a convex optimization problem. Our method achieved excellent performance in stain quantification when validated against the results of multispectral imaging. We further used it to distinguish cells in lobular endocervical glandular hyperplasia (LEGH), a precancerous gastric-type adenocarcinoma lesion, from normal endocervical cells. Stain abundance features clearly separated the two groups, and a classifier based on stain abundance achieved 98.0% accuracy. By converting subjective color impressions into numerical markers, this technique highlights the strong promise of RGB-based stain unmixing for quantitative diagnosis. Graphical abstract

Unsupervised cervical cell instance segmentation method integrating cellular characteristics

Cell instance segmentation is a key technology for cervical cancer auxiliary diagnosis systems. However, pixel-level annotation is time-consuming and labor-intensive, making it difficult to obtain a large amount of annotated data. This results in the model not being fully trained. In response to these problems, this paper proposes an unsupervised cervical cell instance segmentation method that integrates cell characteristics. Cervical cells have a clear corresponding structure between the nucleus and cytoplasm. This method fully takes this feature into account by building a dual-flow framework to locate the nucleus and cytoplasm and generate high-quality pseudo-labels. In the nucleus segmentation stage, the position and range of the nucleus are determined using the standard cell-restricted nucleus segmentation method. In the cytoplasm segmentation stage, a multi-angle collaborative segmentation method is used to achieve the positioning of the cytoplasm. First, taking advantage of the self-similarity characteristics of pixel blocks in cells, a cytoplasmic segmentation method based on self-similarity map iteration is proposed. The pixel blocks are mapped from the perspective of local details, and the iterative segmentation is repeated. Secondly, using low-level features such as cell color and shape, a self-supervised heatmap-aware cytoplasm segmentation method is proposed to obtain the activation map of the cytoplasm from the perspective of global attention. The two methods are fused to determine cytoplasmic regions, and combined with nuclear locations, high-quality pseudo-labels are generated. These pseudo-labels are used to train the model cyclically, and the loss strategy is used to encourage the model to discover new object masks, thereby obtaining a segmentation model with better performance. Experimental results show that this method achieves good results in cytoplasm segmentation. On the three datasets of ISBI, MS_CellSeg, and Cx22, 54.32%, 44.64%, and 66.52% AJI were obtained, respectively, which is better than other typical unsupervised methods selected in this article.