Enhancements in localized classification for uterine cervical cancer digital histology image assessment

Guo, Peng; Almubarak, Haidar; Banerjee, Koyel; Stanley, R. Joe; Long, L. Rodney; Thoma, George R.; Frazier, Shelliane R.; Moss, Randy Hays; Stoecker, William V.

doi:10.4103/2153-3539.197193

Cited by 12 publications

(3 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…After the epithelial patches were extracted, we carried out the squamous epithelial tissue analysis. In the diagnosis of the squamous epithelium, the results of immunohistochemistry have a Then, in order to keep the growth orientation of the epithelium in image patches for tissue-level analysis, skeleton-based epithelium partition [15,23] was used to obtain vertically divided patches. The skeleton medial axis of each epithelium was extracted based on the distance transform, and the shorter axis was cut off.…”

Section: Squamous Epithelial Tissue Analysismentioning

confidence: 99%

Deep Learning-Based Recognition of Cervical Squamous Interepithelial Lesions

Ding

et al. 2023

Diagnostics

View full text Add to dashboard Cite

Cervical squamous intraepithelial lesions (SILs) are precursor lesions of cervical cancer, and their accurate diagnosis enables patients to be treated before malignancy manifests. However, the identification of SILs is usually laborious and has low diagnostic consistency due to the high similarity of pathological SIL images. Although artificial intelligence (AI), especially deep learning algorithms, has drawn a lot of attention for its good performance in cervical cytology tasks, the use of AI for cervical histology is still in its early stages. The feature extraction, representation capabilities, and use of p16 immunohistochemistry (IHC) among existing models are inadequate. Therefore, in this study, we first designed a squamous epithelium segmentation algorithm and assigned the corresponding labels. Second, p16-positive area of IHC slides were extracted with Whole Image Net (WI-Net), followed by mapping the p16-positive area back to the H&E slides and generating a p16-positive mask for training. Finally, the p16-positive areas were inputted into Swin-B and ResNet-50 to classify the SILs. The dataset comprised 6171 patches from 111 patients; patches from 80% of the 90 patients were used for the training set. The accuracy of the Swin-B method for high-grade squamous intraepithelial lesion (HSIL) that we propose was 0.914 [0.889–0.928]. The ResNet-50 model for HSIL achieved an area under the receiver operating characteristic curve (AUC) of 0.935 [0.921–0.946] at the patch level, and the accuracy, sensitivity, and specificity were 0.845, 0.922, and 0.829, respectively. Therefore, our model can accurately identify HSIL, assisting the pathologist in solving actual diagnostic issues and even directing the follow-up treatment of patients.

show abstract

Section: Squamous Epithelial Tissue Analysismentioning

confidence: 99%

Deep Learning-Based Recognition of Cervical Squamous Interepithelial Lesions

Ding

et al. 2023

Diagnostics

View full text Add to dashboard Cite

show abstract

“…The thresholding approaches [9], [10], [11] usually involves nonlinear contrast enhancement, the application of a threshold either globally or locally, and morphological operations such as erosion and dilation. Edge-based approaches [12], [13], [14] also utilize morphological operations and other image processing techniques, such as contrast enhancement or smoothing. Unlike thresholding, edge-based approaches use edge detection techniques to enhance the image rather than thresholding.…”

Section: Related Workmentioning

confidence: 99%

Analysis of food processing crystal images

Liang¹,

Shakouri²,

Allebach³

2023

View full text Add to dashboard Cite

In this paper, we study a food ingredient processing that has a total of ten fabrication stages, where the first stage is known as crystal seed count. In this stage, manufacturers study under a microscope the food sample and count crystal seeds per unit area. Sample preparation has imperfections (e.g. air bubbles) that need to be disregarded. The number of seed crystals per unit area is a key indicator for the quality of subsequent processing steps. This paper proposes a method for automating the crystal counting and air bubble removal processes in the seed count stage, in order to save manufacturers considerable time and money. An automated image processing pipeline for crystal counting and air bubble removal is employed in the proposed method. In addition to the proposed pipeline for crystal counting and air bubble removal automation, we also introduce an interactive GUI which is operated based on this pipeline. This GUI can not only be used by manufacturers of the particular crystals that we examine in this paper, but can also benefit other manufacturers in the food industry that face similar issues. By utilizing this automatic crystal counting pipeline and GUI, food crystal fabrication can be made more efficient. This paper contributes to the field of food processing by presenting a practical and innovative solution to a common challenge in the industry.

show abstract

“…However, most recent studies have applied techniques for the detection and classi cation of invasive cancers [8][9][10][11][12][13] rather than intraepithelial or premalignant lesions. With regard to cervical lesions, some studies have been devoted to the creation of computer-assisted reading systems for assessing cervical cytology specimens [14], and only a limited number of studies have focused on examining CINs [15][16][17][18][19][20][21]. In this study, we aimed to develop and assess an optimal convolutional neural network (CNN) model for classi cation of CINs.…”

Section: Introductionmentioning

confidence: 99%

Automated Diagnosis of Cervical Intraepithelial Neoplasia in Histology Images via Deep Learning

Cho

Kim

Park

et al. 2021

Preprint

View full text Add to dashboard Cite

Artificial intelligence has enabled the automated diagnosis of several cancer types. We aimed to develop and validate deep learning models that automatically classify cervical intraepithelial neoplasia (CIN) based on histological images. Microscopic images of CIN3, CIN2, CIN1, and non-neoplasm were obtained. The performances of two pre-trained convolutional neural network (CNN) models adopting DenseNet-161 and EfficientNet-B7 architectures were evaluated and compared with those of pathologists. The dataset comprised 1,106 images from 588 patients; images of 10% of patients were included in the test dataset. The mean accuracies for the 4-class classification were 88.5% (95% confidence interval [CI], 86.3–90.6%) by DenseNet-161 and 89.5% (95% CI, 83.3–95.7%) by EfficientNet-B7, which were similar to human performance (93.2% and 89.7%). The mean per-class area under the receiver operating characteristic curve values by EfficientNet-B7 were 0.996, 0.990, and 0.971 in the non-neoplasm, CIN3, CIN1 groups, respectively. The class activation map detected the diagnostic area for CIN lesions. In the 3-class classification of CIN2 and CIN3 as one group, the mean accuracies of DenseNet-161 and EfficientNet-B7 increased to 91.4% (95% CI, 88.8–94.0%) and 92.6% (95% CI, 90.4–94.9%), respectively. CNN-based deep learning is a promising tool for diagnosing CIN lesions on digital histological images.

show abstract

Enhancements in localized classification for uterine cervical cancer digital histology image assessment

Cited by 12 publications

References 18 publications

Deep Learning-Based Recognition of Cervical Squamous Interepithelial Lesions

Deep Learning-Based Recognition of Cervical Squamous Interepithelial Lesions

Analysis of food processing crystal images

Automated Diagnosis of Cervical Intraepithelial Neoplasia in Histology Images via Deep Learning

Contact Info

Product

Resources

About