Deep Learning-Based Classification of Uterine Cervical and Endometrial Cancer Subtypes from Whole-Slide Histopathology Images

Abstract

Uterine cervical and endometrial cancers have different subtypes with different clinical outcomes. Therefore, cancer subtyping is essential for proper treatment decisions. Furthermore, an endometrial and endocervical origin for an adenocarcinoma should also be distinguished. Although the discrimination can be helped with various immunohistochemical markers, there is no definitive marker. Therefore, we tested the feasibility of deep learning (DL)-based classification for the subtypes of cervical and endometrial cancers and the site of origin of adenocarcinomas from whole slide images (WSIs) of tissue slides. WSIs were split into 360 × 360-pixel image patches at 20× magnification for classification. Then, the average of patch classification results was used for the final classification. The area under the receiver operating characteristic curves (AUROCs) for the cervical and endometrial cancer classifiers were 0.977 and 0.944, respectively. The classifier for the origin of an adenocarcinoma yielded an AUROC of 0.939. These results clearly demonstrated the feasibility of DL-based classifiers for the discrimination of cancers from the cervix and uterus. We expect that the performance of the classifiers will be much enhanced with an accumulation of WSI data. Then, the information from the classifiers can be integrated with other data for more precise discrimination of cervical and endometrial cancers.

Detection of malignancy in whole slide images of endometrial cancer biopsies using artificial intelligence

“…The three studies that used CPTAC aimed to predict the same information as genetic sequencing [11] or illustrate features in H&E slides that could identify different cancer variants [12,13] and hence allow more personalised treatment. A paper by Song et al [14] uses a dataset from the cancer genome atlas along with the CPTAC dataset to distinguish between subtypes of endonmetrial cancer. A paper by Zhang et al [15] aims to split slides into endometrial cancer or not, achieving sensitivity and specificity of 0.924 and 0.801.…”

Deep Learning for the Pathologic Diagnosis of Hepatocellular Carcinoma, Cholangiocarcinoma, and Metastatic Colorectal Cancer

“…The discrimination of cancer tissue subtypes is the most basic task in cancer diagnosis. Recently, DL has been widely used for tissue subtyping in various cancers [31,[46][47][48][49]. In the present study, we successfully trained classifiers to discriminate between HCC, CC, and mCRC.…”

“…First, proper tissue images in WSIs should be discriminated from multiple artifacts, including air bubbles, compression artifacts, out-of-focus blurring, pen markings, tissue folding, and white backgrounds. We reused a tissue/non-tissue classifier from our previous study to eliminate these artifacts [31]. A normal/tumor classifier was then trained to discriminate cancerous tissues from normal tissues.…”

Role of artificial intelligence in digital pathology for gynecological cancers

“…Additionally, this model offers insights into molecular subtypes and mutation status [40] . On the other hand, the Inception-v3 model attained an AUC value of 0.944 for classifying the EC subtype [41] .…”

“… Author [ref], year Cancer type WSI type Objectives Sample size Participants BenTaieb et al [36] , 2017 Ovarian cancer H&E Subtype classification 133 Ovarian carcinoma patients with different subtypes (HGSC, EN, MC, LGSC, CC) Jiang et al [37] , 2021 Ovarian cancer H&E Subtype classification 30 SBOT and HGSOC patients were retrieved from the institutional pathology system database Wu et al [38] , 2018 Ovarian cancer H&E Subtype classification 85 Ovarian cancer patients with different subtypes (serous carcinoma, MC, endometrioid, and CC) were obtained from the First Affiliated Hospital of Xinjiang Medical University Farahani et al [39] , 2022 Ovarian cancer H&E IHC Subtype classification 485 Patients from the OVCARE archives and the University of Calgary Hong et al [40] , 2021 Endometrial cancer H&E Subtype classification 456 Train, validate, and test data from the TCGA and the Clinical Proteomic Tumor Analysis Consortium (CPTAC). Independent dataset from New York University (NYU) hospitals Song et al [41] , 2022 Endometrial cancer and Cervical cancer H&E Subtype classification 230 (70 for CC, 160 for EC) Data from The Cancer Genome Atlas (TCGA) program and The National Cancer Institute’s Clinical Proteomic Tumor Analysis Consortium (CPTAC) endometrial cancer dataset Li et al [42] , 2023 Cervical cancer H&E Subtype classification 229 Cervical specimens from January 2018 and December 2020 were acquired from the Department of Pathology, Xinhua hospital Chongming branch affiliated with Shanghai Jiaotong University Habtemariam et al [43] , 2022 Cervical cancer H&E Subtype classification 915 WSIs ...…”